From ff70166ff5210049e9108b69fb07b0f76ca9830c Mon Sep 17 00:00:00 2001
From: auphelia <jakobapk@web.de>
Date: Tue, 12 Nov 2019 14:56:15 +0000
Subject: [PATCH] [Docker] Added finn-hlslib and modified Dockerfile and
run-docker to use it inside the container
---
Dockerfile | 2 +
finn-hlslib/CONTRIBUTING.md | 2 +
finn-hlslib/Jenkinsfile | 91 +
finn-hlslib/LICENSE | 30 +
finn-hlslib/README.md | 7 +
finn-hlslib/activations.hpp | 205 ++
finn-hlslib/bnn-library.h | 63 +
finn-hlslib/convlayer.h | 202 ++
finn-hlslib/dma.h | 201 ++
finn-hlslib/docs/Makefile | 45 +
finn-hlslib/docs/README.md | 4 +
finn-hlslib/docs/doxygen/Doxyfile | 2494 +++++++++++++++++
finn-hlslib/docs/requirements.txt | 29 +
finn-hlslib/docs/source/conf.py | 93 +
finn-hlslib/docs/source/index.rst | 65 +
.../docs/source/library/activations.rst | 51 +
finn-hlslib/docs/source/library/convlayer.rst | 42 +
finn-hlslib/docs/source/library/dma.rst | 41 +
finn-hlslib/docs/source/library/fclayer.rst | 42 +
finn-hlslib/docs/source/library/mac.rst | 38 +
.../docs/source/library/matrixvector.rst | 44 +
finn-hlslib/docs/source/library/maxpool.rst | 38 +
finn-hlslib/docs/source/library/mmv.rst | 38 +
.../docs/source/library/streamtools.rst | 38 +
finn-hlslib/docs/source/library/swg.rst | 42 +
finn-hlslib/docs/source/library/weights.rst | 45 +
finn-hlslib/fclayer.h | 113 +
finn-hlslib/interpret.hpp | 286 ++
finn-hlslib/mac.hpp | 206 ++
finn-hlslib/maxpool.h | 387 +++
finn-hlslib/mmv.hpp | 60 +
finn-hlslib/mvau.hpp | 183 ++
finn-hlslib/requirements.txt | 59 +
finn-hlslib/slidingwindow.h | 331 +++
finn-hlslib/streamtools.h | 704 +++++
finn-hlslib/tb/.gitignore | 5 +
finn-hlslib/tb/README.md | 7 +
finn-hlslib/tb/add_config.h | 7 +
finn-hlslib/tb/add_tb.cpp | 65 +
finn-hlslib/tb/add_top.cpp | 15 +
finn-hlslib/tb/cnvlayer_batch_unit.cpp | 41 +
finn-hlslib/tb/conv.hpp | 91 +
finn-hlslib/tb/conv3_tb.cpp | 166 ++
finn-hlslib/tb/conv3mmv_tb.cpp | 164 ++
finn-hlslib/tb/conv_mmv.cpp | 59 +
finn-hlslib/tb/conv_top.cpp | 58 +
finn-hlslib/tb/dup_stream_config.h | 2 +
finn-hlslib/tb/dup_stream_tb.cpp | 50 +
finn-hlslib/tb/dup_stream_top.cpp | 11 +
finn-hlslib/tb/dwc_config.h | 34 +
finn-hlslib/tb/dwc_tb.cpp | 112 +
finn-hlslib/tb/dwc_top.cpp | 51 +
finn-hlslib/tb/gen_weigths.py | 100 +
finn-hlslib/tb/input_gen.cpp | 58 +
finn-hlslib/tb/input_gen.h | 41 +
finn-hlslib/tb/maxpool_tb.cpp | 124 +
finn-hlslib/tb/pool.hpp | 67 +
finn-hlslib/tb/pool_config.h | 38 +
finn-hlslib/tb/pool_top.cpp | 51 +
finn-hlslib/tb/swg_tb.cpp | 105 +
finn-hlslib/tb/test_add.tcl | 11 +
finn-hlslib/tb/test_conv3.tcl | 51 +
finn-hlslib/tb/test_convmmv.tcl | 51 +
finn-hlslib/tb/test_dup_stream.tcl | 11 +
finn-hlslib/tb/test_dwc.tcl | 51 +
finn-hlslib/tb/test_pool.tcl | 51 +
finn-hlslib/tb/test_swg.tcl | 51 +
finn-hlslib/utils.hpp | 104 +
finn-hlslib/weights.hpp | 152 +
run-docker.sh | 5 +
70 files changed, 8381 insertions(+)
create mode 100644 finn-hlslib/CONTRIBUTING.md
create mode 100755 finn-hlslib/Jenkinsfile
create mode 100755 finn-hlslib/LICENSE
create mode 100755 finn-hlslib/README.md
create mode 100755 finn-hlslib/activations.hpp
create mode 100755 finn-hlslib/bnn-library.h
create mode 100755 finn-hlslib/convlayer.h
create mode 100755 finn-hlslib/dma.h
create mode 100755 finn-hlslib/docs/Makefile
create mode 100755 finn-hlslib/docs/README.md
create mode 100755 finn-hlslib/docs/doxygen/Doxyfile
create mode 100644 finn-hlslib/docs/requirements.txt
create mode 100755 finn-hlslib/docs/source/conf.py
create mode 100755 finn-hlslib/docs/source/index.rst
create mode 100755 finn-hlslib/docs/source/library/activations.rst
create mode 100755 finn-hlslib/docs/source/library/convlayer.rst
create mode 100755 finn-hlslib/docs/source/library/dma.rst
create mode 100755 finn-hlslib/docs/source/library/fclayer.rst
create mode 100755 finn-hlslib/docs/source/library/mac.rst
create mode 100755 finn-hlslib/docs/source/library/matrixvector.rst
create mode 100755 finn-hlslib/docs/source/library/maxpool.rst
create mode 100755 finn-hlslib/docs/source/library/mmv.rst
create mode 100755 finn-hlslib/docs/source/library/streamtools.rst
create mode 100755 finn-hlslib/docs/source/library/swg.rst
create mode 100755 finn-hlslib/docs/source/library/weights.rst
create mode 100755 finn-hlslib/fclayer.h
create mode 100755 finn-hlslib/interpret.hpp
create mode 100755 finn-hlslib/mac.hpp
create mode 100755 finn-hlslib/maxpool.h
create mode 100755 finn-hlslib/mmv.hpp
create mode 100755 finn-hlslib/mvau.hpp
create mode 100755 finn-hlslib/requirements.txt
create mode 100755 finn-hlslib/slidingwindow.h
create mode 100755 finn-hlslib/streamtools.h
create mode 100755 finn-hlslib/tb/.gitignore
create mode 100755 finn-hlslib/tb/README.md
create mode 100755 finn-hlslib/tb/add_config.h
create mode 100755 finn-hlslib/tb/add_tb.cpp
create mode 100755 finn-hlslib/tb/add_top.cpp
create mode 100755 finn-hlslib/tb/cnvlayer_batch_unit.cpp
create mode 100755 finn-hlslib/tb/conv.hpp
create mode 100755 finn-hlslib/tb/conv3_tb.cpp
create mode 100755 finn-hlslib/tb/conv3mmv_tb.cpp
create mode 100755 finn-hlslib/tb/conv_mmv.cpp
create mode 100755 finn-hlslib/tb/conv_top.cpp
create mode 100755 finn-hlslib/tb/dup_stream_config.h
create mode 100755 finn-hlslib/tb/dup_stream_tb.cpp
create mode 100755 finn-hlslib/tb/dup_stream_top.cpp
create mode 100755 finn-hlslib/tb/dwc_config.h
create mode 100755 finn-hlslib/tb/dwc_tb.cpp
create mode 100755 finn-hlslib/tb/dwc_top.cpp
create mode 100755 finn-hlslib/tb/gen_weigths.py
create mode 100755 finn-hlslib/tb/input_gen.cpp
create mode 100755 finn-hlslib/tb/input_gen.h
create mode 100755 finn-hlslib/tb/maxpool_tb.cpp
create mode 100755 finn-hlslib/tb/pool.hpp
create mode 100755 finn-hlslib/tb/pool_config.h
create mode 100755 finn-hlslib/tb/pool_top.cpp
create mode 100755 finn-hlslib/tb/swg_tb.cpp
create mode 100755 finn-hlslib/tb/test_add.tcl
create mode 100755 finn-hlslib/tb/test_conv3.tcl
create mode 100755 finn-hlslib/tb/test_convmmv.tcl
create mode 100755 finn-hlslib/tb/test_dup_stream.tcl
create mode 100755 finn-hlslib/tb/test_dwc.tcl
create mode 100755 finn-hlslib/tb/test_pool.tcl
create mode 100755 finn-hlslib/tb/test_swg.tcl
create mode 100755 finn-hlslib/utils.hpp
create mode 100755 finn-hlslib/weights.hpp
diff --git a/Dockerfile b/Dockerfile
index 11d4d9aa9..06cb4e5d5 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -17,6 +17,8 @@ ENV PYTHONPATH "${PYTHONPATH}:/workspace/finn/src"
ENV PYTHONPATH "${PYTHONPATH}:/workspace/brevitas_cnv_lfc/training_scripts"
ENV PYTHONPATH "${PYTHONPATH}:/workspace/brevitas"
ENV PYTHONPATH "${PYTHONPATH}:/workspace/cnpy"
+ENV PYTHONPATH "${PYTHONPATH}:/workspace/finn-hlslib"
+
ARG GID
ARG GNAME
diff --git a/finn-hlslib/CONTRIBUTING.md b/finn-hlslib/CONTRIBUTING.md
new file mode 100644
index 000000000..86209240d
--- /dev/null
+++ b/finn-hlslib/CONTRIBUTING.md
@@ -0,0 +1,2 @@
+To get started, <a href="https://www.clahub.com/agreements/Xilinx/finn-hlslib">sign the Contributor License Agreement</a>.
+
diff --git a/finn-hlslib/Jenkinsfile b/finn-hlslib/Jenkinsfile
new file mode 100755
index 000000000..fd0d4c200
--- /dev/null
+++ b/finn-hlslib/Jenkinsfile
@@ -0,0 +1,91 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+
+node {
+ def app
+
+ stage('Clone repository') {
+ /* Let's make sure we have the repository cloned to our workspace */
+ checkout scm
+ }
+
+ parallel firstBranch: {
+ stage('Run tests SWG') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('source /proj/xbuilds/2019.1_released/installs/lin64/Vivado/2019.1/settings64.sh; cd tb; vivado_hls -f test_swg.tcl')
+ }
+ }, secondBranch: {
+ stage('Run tests POOL') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('source /proj/xbuilds/2019.1_released/installs/lin64/Vivado/2019.1/settings64.sh; cd tb; vivado_hls -f test_pool.tcl')
+ }
+ }, thirdBranch: {
+ stage('Run tests DWC') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('source /proj/xbuilds/2019.1_released/installs/lin64/Vivado/2019.1/settings64.sh; cd tb; vivado_hls -f test_dwc.tcl')
+ }
+ }, fourthBranch: {
+ stage('Run tests ADD') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('source /proj/xbuilds/2019.1_released/installs/lin64/Vivado/2019.1/settings64.sh; cd tb; vivado_hls -f test_add.tcl')
+ }
+ }, fifthBranch: {
+ stage('Run tests DUP_STREAM') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('source /proj/xbuilds/2019.1_released/installs/lin64/Vivado/2019.1/settings64.sh; cd tb; vivado_hls -f test_dup_stream.tcl')
+ }
+ }, sixthBranch: {
+ stage('Set-up virtual env') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('virtualenv venv; source venv/bin/activate;pip3.7 install -r requirements.txt')
+ }
+ stage('Generate weigths fro conv test') {
+ sh('source venv/bin/activate; cd tb; python3.7 gen_weigths.py;')
+ }
+ stage('Run tests CONV3') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('source /proj/xbuilds/2019.1_released/installs/lin64/Vivado/2019.1/settings64.sh; cd tb; vivado_hls -f test_conv3.tcl')
+ }
+ stage('Run tests CONVMMV') {
+ env.FINN_HLS_ROOT = "${env.WORKSPACE}"
+ echo "${env.FINN_HLS_ROOT}"
+ sh('source /proj/xbuilds/2019.1_released/installs/lin64/Vivado/2019.1/settings64.sh; cd tb; vivado_hls -f test_convmmv.tcl')
+ }
+ }
+}
diff --git a/finn-hlslib/LICENSE b/finn-hlslib/LICENSE
new file mode 100755
index 000000000..7023ac47c
--- /dev/null
+++ b/finn-hlslib/LICENSE
@@ -0,0 +1,30 @@
+Copyright (c) 2019, Xilinx, Inc.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice,
+ this list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its
+ contributors may be used to endorse or promote products derived from
+ this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+
diff --git a/finn-hlslib/README.md b/finn-hlslib/README.md
new file mode 100755
index 000000000..46916314c
--- /dev/null
+++ b/finn-hlslib/README.md
@@ -0,0 +1,7 @@
+# finn-hlslib [](https://finn-hlslib.readthedocs.io/en/latest/?badge=latest)
+
+
+This repo contains the HLS library for hardware acceleration of Quantized Neural Network (QNN) using FINN.
+
+For more information please refer to the documentation available <a href="https://finn-hlslib.readthedocs.io" target="_blank"> here </a>.
+
diff --git a/finn-hlslib/activations.hpp b/finn-hlslib/activations.hpp
new file mode 100755
index 000000000..f2461258b
--- /dev/null
+++ b/finn-hlslib/activations.hpp
@@ -0,0 +1,205 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *******************************************************************************/
+
+/*******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * @file activations.hpp
+ *
+ * Library of templated HLS classes for BNN deployment.
+ * This file lists a set of classes used to implement
+ * threshold memory in neural network.
+ *
+ * This project has received funding from the European Union's Framework
+ * Programme for Research and Innovation Horizon 2020 (2014-2020) under
+ * the Marie Skłodowska-Curie Grant Agreement No. 751339.
+ *
+ *******************************************************************************/
+
+#ifndef ACTIVATIONS_HPP
+#define ACTIVATIONS_HPP
+
+#include "interpret.hpp"
+
+/**
+ * General contract for activation functions.
+ *
+ * This class itself has no formal significance for the implementation
+ * of the MVAU. Implementations of activation functions are encouraged
+ * to implement it nonetheless to guarantee appropriate function
+ * signatures.
+ */
+template<typename TA, typename TO>
+class Activation {
+public:
+ TA init(unsigned const nf, unsigned const pe) const {
+#pragma HLS inline
+ return TA(0);
+ }
+
+ /**
+ * Compute the activation of the passed accumulator value accu in row idx.
+ */
+ TO activate(unsigned const nf, unsigned const pe, TA const &accu) const;
+};
+
+/**
+ * A no-op activation that simply outputs the computed accumulator
+ * output as the final result.
+ */
+template<typename T>
+class PassThroughActivation : public Activation<T, T> {
+public:
+ T activate(unsigned const nf, unsigned const pe, T const &accu) const {
+#pragma HLS inline
+ return accu;
+ }
+};
+
+/**
+ * Use a simple global threshold comparison as activation function.
+ *
+ * The constant threshold is initialized at construction.
+ * The default comparison returns true if the threshold value is
+ * smaller than the passed accumulator value.
+ */
+template<typename TA, typename Compare = std::less<TA>>
+class ThresholdActivation : public Activation<TA, bool> {
+ TA const m_threshold;
+public:
+ ThresholdActivation(TA const &threshold) : m_threshold(threshold) {
+#pragma HLS inline
+ }
+
+public:
+ bool activate(unsigned const nf, unsigned const pe, TA const &accu) const {
+#pragma HLS inline
+ return Compare()(m_threshold, accu);
+ }
+};
+
+/**
+ * Use a simple per-row threshold comparison as activation function.
+ *
+ * The thresholds are taken from an array indexed by output row.
+ * It is currently public to allow direct initialization and
+ * to make its name accessible for top-level HLS pragmas.
+ *
+ * The default comparison returns true if the threshold value defined for
+ * the indexed row is smaller than the passed accumulator value.
+ */
+template<unsigned NF, unsigned PE, unsigned NumTH,
+ typename TA, typename TR, int ActVal = 0, typename Compare = std::less<TA>>
+class ThresholdsActivation {
+public:
+ TA m_thresholds[PE][NF][NumTH];
+
+public:
+ TA init(unsigned const nf, unsigned const pe) const {
+#pragma HLS inline
+ return TA(0);
+ }
+
+public:
+ TR activate(unsigned const nf, unsigned const pe, TA const &accu) const {
+#pragma HLS inline
+ TR result=ActVal;
+ for(unsigned int i=0; i< NumTH; i++){
+#pragma HLS unroll
+ result+=Compare()(m_thresholds[pe][nf][i], accu);
+ }
+ return result;
+ }
+};
+
+/**
+ * \brief Thresholding function for multiple images
+ *
+ * The function performs thresholds comparison with input activation vector,
+ * and generating output based on the comparison results
+ *
+ * \tparam ImgDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam NumChannels Heigth of the input matrix
+ * \tparam PE Number of output rows computed in parallel
+ * \tparam TSrcI DataType of the input activation (as used in the MAC)
+ * \tparam TDstI DataType of the output activation (as generated by the activation)
+ * \tparam TI DataType of the input stream - safely deducible from the paramaters
+ * \tparam TO DataType of the output stream - safely deducible from the paramaters
+ * \tparam TA DataType of the activation class (e.g. thresholds) - safely deducible from the paramaters
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param activation Activation class
+ * \param reps Number of time the function has to be repeatedly executed (e.g. number of images)
+ */
+template <
+ unsigned ImgDim, unsigned NumChannels, unsigned PE,
+ typename TSrcI = Identity, typename TDstI = Identity,
+ typename TI, typename TO, typename TA>
+void Thresholding_Batch(hls::stream<TI> &in,
+ hls::stream<TO> &out,
+ TA const &activation,
+ int const reps)
+{
+
+ // how many different rows each neuron will compute
+ // alternatively: number of vertical matrix chunks
+ unsigned const NF = NumChannels / PE;
+
+ unsigned nf = 0;
+ unsigned tile = 0; // invariant: tile = nf*SF + sf
+
+ // everything merged into a common iteration space (one "big" loop instead
+ // of smaller nested loops) to get the pipelinening the way we want
+ for (unsigned i = 0; i < reps * ImgDim * ImgDim * NF; i++)
+ {
+ TI inElem;
+ inElem = in.read();
+ auto outElem = TDstI().template operator()<TO>();
+ for (unsigned pe = 0; pe < PE; pe++)
+ {
+#pragma HLS UNROLL
+ auto const act = TSrcI()(inElem);
+ outElem[pe] = activation.activate(nf, pe, act[pe]);
+ }
+ out.write(outElem);
+ if (++nf == NF)
+ {
+ nf = 0;
+ }
+ }
+}
+#endif
diff --git a/finn-hlslib/bnn-library.h b/finn-hlslib/bnn-library.h
new file mode 100755
index 000000000..1080655d8
--- /dev/null
+++ b/finn-hlslib/bnn-library.h
@@ -0,0 +1,63 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *****************************************************************************/
+
+/*****************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * @file bnn-library.h
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * Include this file in the network top funtion.
+ *
+ *****************************************************************************/
+
+#include <hls_stream.h>
+#include "ap_int.h"
+#include <iostream>
+#include <string>
+
+using namespace hls;
+using namespace std;
+
+#define CASSERT_DATAFLOW(x) {if (!(x)) {std::cout<< "CASSERT_DATAFLOW condition is not met " << endl; exit(-1); }}
+
+#include "mmv.hpp"
+#include "streamtools.h"
+#include "dma.h"
+#include "slidingwindow.h"
+#include "maxpool.h"
+#include "fclayer.h"
+#include "convlayer.h"
diff --git a/finn-hlslib/convlayer.h b/finn-hlslib/convlayer.h
new file mode 100755
index 000000000..e0517a469
--- /dev/null
+++ b/finn-hlslib/convlayer.h
@@ -0,0 +1,202 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *****************************************************************************/
+
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * \file convlayer.h
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * This file lists a set of convenience functions used to implement
+ * convolutional layers.
+ *
+ *****************************************************************************/
+
+#ifndef CONVLAYER_H
+#define CONVLAYER_H
+
+#include <ap_int.h>
+#include <hls_stream.h>
+
+#include "streamtools.h"
+#include "mvau.hpp"
+/**
+ * \brief Convolutional layer implementation
+ *
+ * The function implements a generic convolutional layer, and it's basically composed of the sliding window generator
+ * implemeting the im2col algorithm and the Matrix_Vector_Activate_Batch function to perform computation.
+ *
+ * \tparam ConvKernelDim Dimension of the convolutional kernel (assumed square)
+ * \tparam IFMChannels Number of Input Feature Maps
+ * \tparam IFMDim Width and Height of the Input Feature Map (assumed square)
+ * \tparam OFMChannels Number of Output Feature Maps
+ * \tparam OFMDim Width and Height of the Output Feature Map (assumed square)
+ * \tparam SIMD Number of input columns computed in parallel
+ * \tparam PE Number of output rows computed in parallel
+ * \tparam TSrcI DataType of the input activation (as used in the MAC)
+ * \tparam TDstI DataType of the output activation (as generated by the activation)
+ * \tparam TWeightI DataType of the weights (as used in the MAC)
+ * \tparam InStreamW Width of the input stream
+ * \tparam OutStreamW Width of the output stream
+ * \tparam TW DataType of the weights matrix - safely deducible from the paramaters
+ * \tparam TA DataType of the activation class (e.g. thresholds) - safely deducible from the paramaters
+ * \tparam R DataType for the resource used for FPGA implementation of the MAC - safely deducible from the paramaters
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param weights Weights matrix (currently supports BinaryWeights or FixedPointWeights)
+ * \param activation Activation class
+ * \param reps Number of time the function has to be repeatedly executed (e.g. number of images)
+ * \param r Resource type for the hardware implementation of the MAC block
+ */
+
+template<
+ unsigned int ConvKernelDim,
+ unsigned int IFMChannels,
+ unsigned int IFMDim,
+ unsigned int OFMChannels,
+ unsigned int OFMDim,
+
+ unsigned int SIMD, // number of SIMD lanes
+ unsigned int PE, // number of PEs
+
+ typename TSrcI = Identity, // redefine I/O interpretation as needed for input activations
+ typename TDstI = Identity, // redefine I/O interpretation as needed for output activations
+ typename TWeightI = Identity, // redefine I/O interpretation as needed for weigths
+
+ int InStreamW, int OutStreamW, // safely deducible (stream width must be int though!)
+ typename TW, typename TA, typename R
+>
+void ConvLayer_Batch(hls::stream<ap_uint<InStreamW>> &in,
+ hls::stream<ap_uint<OutStreamW>> &out,
+ TW const &weights,
+ TA const &activation,
+ unsigned const reps,
+ R const &r) {
+#pragma HLS INLINE
+ unsigned const MatrixW = ConvKernelDim * ConvKernelDim * IFMChannels;
+ unsigned const MatrixH = OFMChannels;
+ unsigned const InpPerImage = IFMDim*IFMDim*IFMChannels/InStreamW * TSrcI::width;
+ WidthAdjustedInputStream <InStreamW, SIMD*TSrcI::width, InpPerImage> wa_in (in, reps);
+ WidthAdjustedOutputStream <PE*TDstI::width, OutStreamW, OFMDim * OFMDim * (OFMChannels / PE)> mvOut (out, reps);
+ hls::stream<ap_uint<SIMD*TSrcI::width> > convInp("StreamingConvLayer_Batch.convInp");
+ ConvolutionInputGenerator<ConvKernelDim, IFMChannels, TSrcI::width, IFMDim,
+ OFMDim, SIMD,1>(wa_in, convInp, reps);
+ Matrix_Vector_Activate_Batch<MatrixW, MatrixH, SIMD, PE, 1, TSrcI, TDstI, TWeightI>
+ (static_cast<hls::stream<ap_uint<SIMD*TSrcI::width>>&>(convInp),
+ static_cast<hls::stream<ap_uint<PE*TDstI::width>>&> (mvOut),
+ weights, activation, reps* OFMDim * OFMDim, r);
+}
+
+
+/**
+ * \brief Convolutional layer implementation
+ *
+ * The function implements a generic convolutional layer, and it's basically composed of the sliding window generator
+ * implemeting the im2col algorithm and the Matrix_Vector_Activate_Batch function to perform computation.
+ *
+ * \tparam ConvKernelDim Dimension of the convolutional kernel (assumed square)
+ * \tparam IFMChannels Number of Input Feature Maps
+ * \tparam IFMDim Width and Height of the Input Feature Map (assumed square)
+ * \tparam OFMChannels Number of Output Feature Maps
+ * \tparam OFMDim Width and Height of the Output Feature Map (assumed square)
+ * \tparam STRIDE Stride of the convolutional kernel
+ *
+ * \tparam SIMD Number of input columns computed in parallel
+ * \tparam PE Number of output rows computed in parallel
+ * \tparam MMV Number of output pixels computed in parallel
+ *
+ * \tparam TSrcI DataType of the input activation (as used in the MAC)
+ * \tparam TDstI DataType of the output activation (as generated by the activation)
+ * \tparam TWeightI DataType of the weights (as used in the MAC)
+ * \tparam InStreamW Width of the input stream
+ * \tparam OutStreamW Width of the output stream
+ * \tparam TW DataType of the weights matrix - safely deducible from the paramaters
+ * \tparam TA DataType of the activation class (e.g. thresholds) - safely deducible from the paramaters
+ * \tparam R DataType for the resource used for FPGA implementation of the MAC - safely deducible from the paramaters
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param weights Weights matrix (currently supports BinaryWeights or FixedPointWeights)
+ * \param activation Activation class
+ * \param reps Number of time the function has to be repeatedly executed (e.g. number of images)
+ * \param r Resource type for the hardware implementation of the MAC block
+ */
+template<
+ unsigned int ConvKernelDim,
+ unsigned int IFMChannels,
+ unsigned int IFMDim,
+ unsigned int OFMChannels,
+ unsigned int OFMDim,
+ unsigned int STRIDE,
+
+ unsigned int SIMD, // number of SIMD lanes
+ unsigned int PE, // number of PEs
+ unsigned int MMV,
+
+ typename TSrcI = Identity, // redefine I/O interpretation as needed for input activations
+ typename TDstI = Identity, // redefine I/O interpretation as needed for output activations
+ typename TWeightI = Identity, // redefine I/O interpretation as needed for weigths
+
+ int InStreamW, int OutStreamW, // safely deducible (stream width must be int though!)
+ typename TW, typename TA, typename R
+>
+void ConvLayer_Batch_MMV(hls::stream<ap_uint<InStreamW>> &in,
+ hls::stream<ap_uint<OutStreamW>> &out,
+ TW const &weights,
+ TA const &activation,
+ unsigned const reps,
+ R const &r) {
+#pragma HLS INLINE
+ unsigned const MatrixW = ConvKernelDim * ConvKernelDim * IFMChannels;
+ unsigned const MatrixH = OFMChannels;
+ unsigned const InpPerImage = IFMDim*IFMDim*IFMChannels/InStreamW * TSrcI::width;
+ const unsigned int mmvReps = (reps * OFMDim * OFMDim) / MMV;
+ WidthAdjustedInputStream <InStreamW, SIMD*TSrcI::width, InpPerImage> wa_in (in, reps);
+ WidthAdjustedOutputStream <PE*TDstI::width*MMV, OutStreamW, OFMDim * OFMDim * (OFMChannels / PE)/MMV> mvOut (out, reps);
+ stream<MultiChanData<MMV, PE * TDstI::width> > mmv2dwc("mmv2dwc");
+ hls::stream<MultiChanData<MMV, SIMD *TSrcI::width> > convInp("StreamingConvLayer_Batch.convInp");
+ ConvolutionInputGenerator_MMV<ConvKernelDim, IFMChannels, TSrcI::width, IFMDim,
+ OFMDim, SIMD, STRIDE, MMV>(wa_in, convInp, reps);
+ Matrix_Vector_Activate_Batch<MatrixW, MatrixH, SIMD, PE, MMV, TSrcI, TDstI, TWeightI>
+ (static_cast<hls::stream<MultiChanData<MMV,SIMD*TSrcI::width>>&>(convInp),
+ static_cast<hls::stream<MultiChanData<MMV,PE*TDstI::width>>&>(mmv2dwc),
+ weights, activation, mmvReps, r);
+ FlattenMultiChanData<MMV, PE * TDstI::width>(mmv2dwc, mvOut, mmvReps); // I guess it should be flattened after the MMV DWC
+
+}
+
+#endif
diff --git a/finn-hlslib/dma.h b/finn-hlslib/dma.h
new file mode 100755
index 000000000..6d2ac77e8
--- /dev/null
+++ b/finn-hlslib/dma.h
@@ -0,0 +1,201 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * \file dma.h
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * This file lists a set of functions to access memory mapped values into
+ * streams.
+ *
+ *****************************************************************************/
+#ifndef DMA_HPP
+#define DMA_HPP
+
+
+#include <ap_int.h>
+#include <hls_stream.h>
+
+/*!
+ * \brief DMA block accessing AXI4 memory and output HLS streams
+ *
+ *
+ * \tparam DataWidth Width, in number of bits, of the AXI4 memory pointer and the output HLS stream
+ * \tparam numBytes Number of bytes to be read from the memory
+ *
+ * \param in Input memory pointer
+ * \param out Output HLS stream
+ */
+template<unsigned int DataWidth, unsigned int numBytes>
+void Mem2Stream(ap_uint<DataWidth> * in, hls::stream<ap_uint<DataWidth> > & out);
+
+/*!
+ * \brief DMA block accessing AXI4 memory and output HLS streams multiple times
+ *
+ * It basically calls Mem2Stream function multiple times, possibly with bigger sizes so to increase
+ * the burst size
+ *
+ * \tparam DataWidth Width, in number of bits, of the AXI4 memory pointer and the output HLS stream
+ * \tparam numBytes Number of bytes to be read from the memory
+ *
+ * \param in Input memory pointer
+ * \param out Output HLS stream
+ * \param numReps Number of times the Stream2Mem function has to be called
+ */
+template<unsigned int DataWidth, unsigned int numBytes>
+void Mem2Stream_Batch(ap_uint<DataWidth> * in, hls::stream<ap_uint<DataWidth> > & out, const unsigned int numReps);
+
+/*!
+ * \brief DMA block writing HLS streams content in AXI4 pointed memory
+ *
+ *
+ * \tparam DataWidth Width, in number of bits, of the AXI4 memory pointer and the output HLS stream
+ * \tparam numBytes Number of bytes to be read from the memory
+ *
+ * \param in Input HLS stream
+ * \param out Output memory pointer
+ */
+template<unsigned int DataWidth, unsigned int numBytes>
+void Stream2Mem(hls::stream<ap_uint<DataWidth> > & in, ap_uint<DataWidth> * out);
+
+/*!
+ * \brief DMA block that accesses the external memory and outputs HLS streams multiple times
+ *
+ * It basically calls Mem2Stream function multiple times, possibly with bigger sizes so to increase
+ * the burst size
+ *
+ * \tparam DataWidth Width, in number of bits, of the AXI4 memory pointer and the output HLS stream
+ * \tparam numBytes Number of bytes to be read from the memory
+ *
+ * \param in Input pointer to external memory
+ * \param out Output the generated HLS sream
+ * \param numReps Number of times the Mem2Stream function has to be called
+ */
+template<unsigned int DataWidth, unsigned int numBytes>
+void Mem2Stream_Batch_external_wmem(ap_uint<DataWidth> * in,
+ stream<ap_uint<DataWidth> > & out, const unsigned int numReps) {
+ unsigned int rep = 0;
+ while (rep != numReps) {
+ Mem2Stream<DataWidth, numBytes>(&in[0], out);
+ rep += 1;
+ }
+}
+
+/*!
+ * \brief DMA block writing HLS streams content in AXI4 pointed memory multiple times
+ *
+ * It basically calls Stream2Mem function multiple times, possibly with bigger sizes so to increase
+ * the burst size
+ *
+ * \tparam DataWidth Width, in number of bits, of the AXI4 memory pointer and the output HLS stream
+ * \tparam numBytes Number of bytes to be read from the memory
+ *
+ * \param in Input HLS stream
+ * \param out Output memory pointer
+ * \param numReps Number of times the Stream2Mem function has to be called
+ */
+template<unsigned int DataWidth, unsigned int numBytes>
+void Stream2Mem_Batch(hls::stream<ap_uint<DataWidth> > & in, ap_uint<DataWidth> * out, const unsigned int numReps);
+
+template<unsigned int DataWidth, unsigned int numBytes>
+void Mem2Stream(ap_uint<DataWidth> * in, hls::stream<ap_uint<DataWidth> > & out) {
+ CASSERT_DATAFLOW(DataWidth % 8 == 0);
+ const unsigned int numWords = numBytes / (DataWidth / 8);
+ CASSERT_DATAFLOW(numWords != 0);
+ for (unsigned int i = 0; i < numWords; i++) {
+#pragma HLS PIPELINE II=1
+ ap_uint<DataWidth> e = in[i];
+ out.write(e);
+ }
+}
+
+
+template<unsigned int DataWidth, unsigned int numBytes>
+void Stream2Mem(hls::stream<ap_uint<DataWidth> > & in, ap_uint<DataWidth> * out) {
+ CASSERT_DATAFLOW(DataWidth % 8 == 0);
+ const unsigned int numWords = numBytes / (DataWidth / 8);
+ CASSERT_DATAFLOW(numWords != 0);
+ for (unsigned int i = 0; i < numWords; i++) {
+#pragma HLS PIPELINE II=1
+ ap_uint<DataWidth> e = in.read();
+ out[i] = e;
+ }
+}
+
+template<unsigned int DataWidth, unsigned int numBytes>
+void Mem2Stream_Batch(ap_uint<DataWidth> * in, hls::stream<ap_uint<DataWidth> > & out, const unsigned int numReps) {
+ const unsigned int indsPerRep = numBytes / (DataWidth / 8);
+ unsigned int rep = 0;
+ // make sure Mem2Stream does not get inlined here
+ // we lose burst inference otherwise
+ while (rep != numReps) {
+ unsigned int repsLeft = numReps - rep;
+ if ((repsLeft & 0xF) == 0) {
+ // repsLeft divisable by 16, read 16 images
+ Mem2Stream<DataWidth, numBytes * 16>(&in[rep * indsPerRep], out);
+ rep += 16;
+ } else {
+ // fallback, read single image
+ Mem2Stream<DataWidth, numBytes>(&in[rep * indsPerRep], out);
+ rep += 1;
+ }
+ }
+}
+
+
+template<unsigned int DataWidth, unsigned int numBytes>
+void Stream2Mem_Batch(hls::stream<ap_uint<DataWidth> > & in, ap_uint<DataWidth> * out, const unsigned int numReps) {
+ const unsigned int indsPerRep = numBytes / (DataWidth / 8);
+ unsigned int rep = 0;
+ // make sure Stream2Mem does not get inlined here
+ // we lose burst inference otherwise
+ while (rep != numReps) {
+ unsigned int repsLeft = numReps - rep;
+ if ((repsLeft & 0xF) == 0) {
+ // repsLeft divisable by 16, write 16 images
+ Stream2Mem<DataWidth, numBytes * 16>(in, &out[rep * indsPerRep]);
+ rep += 16;
+ } else {
+ // fallback, write single image
+ Stream2Mem<DataWidth, numBytes>(in, &out[rep * indsPerRep]);
+ rep += 1;
+ }
+ }
+}
+
+#endif
diff --git a/finn-hlslib/docs/Makefile b/finn-hlslib/docs/Makefile
new file mode 100755
index 000000000..ed1272792
--- /dev/null
+++ b/finn-hlslib/docs/Makefile
@@ -0,0 +1,45 @@
+# Copyright (c) 2019, Xilinx, Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice,
+# this list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in the
+# documentation and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+# OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+# ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# You can set these variables from the command line.
+SPHINXOPTS =
+SPHINXBUILD = sphinx-build
+SOURCEDIR = source
+BUILDDIR = build
+
+# Put it first so that "make" without argument is like "make help".
+help:
+ @$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
+
+.PHONY: help Makefile
+
+# Catch-all target: route all unknown targets to Sphinx using the new
+# "make mode" option. $(O) is meant as a shortcut for $(SPHINXOPTS).
+%: Makefile
+ @$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
diff --git a/finn-hlslib/docs/README.md b/finn-hlslib/docs/README.md
new file mode 100755
index 000000000..e9322cd09
--- /dev/null
+++ b/finn-hlslib/docs/README.md
@@ -0,0 +1,4 @@
+# Documentation
+
+
+Documentation for FINN HLS library is generated using <a href="http://www.sphinx-doc.org/" target="_blank"> Sphinx </a> and <a href="https://breathe.readthedocs.io/" target="_blank"> Breathe, which works on top of <a href="http://www.doxygen.nl/index.html" target="_blank"> Doxygen </a> output.
diff --git a/finn-hlslib/docs/doxygen/Doxyfile b/finn-hlslib/docs/doxygen/Doxyfile
new file mode 100755
index 000000000..3f85a6e5d
--- /dev/null
+++ b/finn-hlslib/docs/doxygen/Doxyfile
@@ -0,0 +1,2494 @@
+# Doxyfile 1.8.13
+
+# This file describes the settings to be used by the documentation system
+# doxygen (www.doxygen.org) for a project.
+#
+# All text after a double hash (##) is considered a comment and is placed in
+# front of the TAG it is preceding.
+#
+# All text after a single hash (#) is considered a comment and will be ignored.
+# The format is:
+# TAG = value [value, ...]
+# For lists, items can also be appended using:
+# TAG += value [value, ...]
+# Values that contain spaces should be placed between quotes (\" \").
+
+#---------------------------------------------------------------------------
+# Project related configuration options
+#---------------------------------------------------------------------------
+
+# This tag specifies the encoding used for all characters in the config file
+# that follow. The default is UTF-8 which is also the encoding used for all text
+# before the first occurrence of this tag. Doxygen uses libiconv (or the iconv
+# built into libc) for the transcoding. See http://www.gnu.org/software/libiconv
+# for the list of possible encodings.
+# The default value is: UTF-8.
+
+DOXYFILE_ENCODING = UTF-8
+
+# The PROJECT_NAME tag is a single word (or a sequence of words surrounded by
+# double-quotes, unless you are using Doxywizard) that should identify the
+# project for which the documentation is generated. This name is used in the
+# title of most generated pages and in a few other places.
+# The default value is: My Project.
+
+PROJECT_NAME = "My Project"
+
+# The PROJECT_NUMBER tag can be used to enter a project or revision number. This
+# could be handy for archiving the generated documentation or if some version
+# control system is used.
+
+PROJECT_NUMBER =
+
+# Using the PROJECT_BRIEF tag one can provide an optional one line description
+# for a project that appears at the top of each page and should give viewer a
+# quick idea about the purpose of the project. Keep the description short.
+
+PROJECT_BRIEF =
+
+# With the PROJECT_LOGO tag one can specify a logo or an icon that is included
+# in the documentation. The maximum height of the logo should not exceed 55
+# pixels and the maximum width should not exceed 200 pixels. Doxygen will copy
+# the logo to the output directory.
+
+PROJECT_LOGO =
+
+# The OUTPUT_DIRECTORY tag is used to specify the (relative or absolute) path
+# into which the generated documentation will be written. If a relative path is
+# entered, it will be relative to the location where doxygen was started. If
+# left blank the current directory will be used.
+
+OUTPUT_DIRECTORY =
+
+# If the CREATE_SUBDIRS tag is set to YES then doxygen will create 4096 sub-
+# directories (in 2 levels) under the output directory of each output format and
+# will distribute the generated files over these directories. Enabling this
+# option can be useful when feeding doxygen a huge amount of source files, where
+# putting all generated files in the same directory would otherwise causes
+# performance problems for the file system.
+# The default value is: NO.
+
+CREATE_SUBDIRS = NO
+
+# If the ALLOW_UNICODE_NAMES tag is set to YES, doxygen will allow non-ASCII
+# characters to appear in the names of generated files. If set to NO, non-ASCII
+# characters will be escaped, for example _xE3_x81_x84 will be used for Unicode
+# U+3044.
+# The default value is: NO.
+
+ALLOW_UNICODE_NAMES = NO
+
+# The OUTPUT_LANGUAGE tag is used to specify the language in which all
+# documentation generated by doxygen is written. Doxygen will use this
+# information to generate all constant output in the proper language.
+# Possible values are: Afrikaans, Arabic, Armenian, Brazilian, Catalan, Chinese,
+# Chinese-Traditional, Croatian, Czech, Danish, Dutch, English (United States),
+# Esperanto, Farsi (Persian), Finnish, French, German, Greek, Hungarian,
+# Indonesian, Italian, Japanese, Japanese-en (Japanese with English messages),
+# Korean, Korean-en (Korean with English messages), Latvian, Lithuanian,
+# Macedonian, Norwegian, Persian (Farsi), Polish, Portuguese, Romanian, Russian,
+# Serbian, Serbian-Cyrillic, Slovak, Slovene, Spanish, Swedish, Turkish,
+# Ukrainian and Vietnamese.
+# The default value is: English.
+
+OUTPUT_LANGUAGE = English
+
+# If the BRIEF_MEMBER_DESC tag is set to YES, doxygen will include brief member
+# descriptions after the members that are listed in the file and class
+# documentation (similar to Javadoc). Set to NO to disable this.
+# The default value is: YES.
+
+BRIEF_MEMBER_DESC = YES
+
+# If the REPEAT_BRIEF tag is set to YES, doxygen will prepend the brief
+# description of a member or function before the detailed description
+#
+# Note: If both HIDE_UNDOC_MEMBERS and BRIEF_MEMBER_DESC are set to NO, the
+# brief descriptions will be completely suppressed.
+# The default value is: YES.
+
+REPEAT_BRIEF = YES
+
+# This tag implements a quasi-intelligent brief description abbreviator that is
+# used to form the text in various listings. Each string in this list, if found
+# as the leading text of the brief description, will be stripped from the text
+# and the result, after processing the whole list, is used as the annotated
+# text. Otherwise, the brief description is used as-is. If left blank, the
+# following values are used ($name is automatically replaced with the name of
+# the entity):The $name class, The $name widget, The $name file, is, provides,
+# specifies, contains, represents, a, an and the.
+
+ABBREVIATE_BRIEF = "The $name class" \
+ "The $name widget" \
+ "The $name file" \
+ is \
+ provides \
+ specifies \
+ contains \
+ represents \
+ a \
+ an \
+ the
+
+# If the ALWAYS_DETAILED_SEC and REPEAT_BRIEF tags are both set to YES then
+# doxygen will generate a detailed section even if there is only a brief
+# description.
+# The default value is: NO.
+
+ALWAYS_DETAILED_SEC = NO
+
+# If the INLINE_INHERITED_MEMB tag is set to YES, doxygen will show all
+# inherited members of a class in the documentation of that class as if those
+# members were ordinary class members. Constructors, destructors and assignment
+# operators of the base classes will not be shown.
+# The default value is: NO.
+
+INLINE_INHERITED_MEMB = NO
+
+# If the FULL_PATH_NAMES tag is set to YES, doxygen will prepend the full path
+# before files name in the file list and in the header files. If set to NO the
+# shortest path that makes the file name unique will be used
+# The default value is: YES.
+
+FULL_PATH_NAMES = YES
+
+# The STRIP_FROM_PATH tag can be used to strip a user-defined part of the path.
+# Stripping is only done if one of the specified strings matches the left-hand
+# part of the path. The tag can be used to show relative paths in the file list.
+# If left blank the directory from which doxygen is run is used as the path to
+# strip.
+#
+# Note that you can specify absolute paths here, but also relative paths, which
+# will be relative from the directory where doxygen is started.
+# This tag requires that the tag FULL_PATH_NAMES is set to YES.
+
+STRIP_FROM_PATH =
+
+# The STRIP_FROM_INC_PATH tag can be used to strip a user-defined part of the
+# path mentioned in the documentation of a class, which tells the reader which
+# header file to include in order to use a class. If left blank only the name of
+# the header file containing the class definition is used. Otherwise one should
+# specify the list of include paths that are normally passed to the compiler
+# using the -I flag.
+
+STRIP_FROM_INC_PATH =
+
+# If the SHORT_NAMES tag is set to YES, doxygen will generate much shorter (but
+# less readable) file names. This can be useful is your file systems doesn't
+# support long names like on DOS, Mac, or CD-ROM.
+# The default value is: NO.
+
+SHORT_NAMES = NO
+
+# If the JAVADOC_AUTOBRIEF tag is set to YES then doxygen will interpret the
+# first line (until the first dot) of a Javadoc-style comment as the brief
+# description. If set to NO, the Javadoc-style will behave just like regular Qt-
+# style comments (thus requiring an explicit @brief command for a brief
+# description.)
+# The default value is: NO.
+
+JAVADOC_AUTOBRIEF = NO
+
+# If the QT_AUTOBRIEF tag is set to YES then doxygen will interpret the first
+# line (until the first dot) of a Qt-style comment as the brief description. If
+# set to NO, the Qt-style will behave just like regular Qt-style comments (thus
+# requiring an explicit \brief command for a brief description.)
+# The default value is: NO.
+
+QT_AUTOBRIEF = NO
+
+# The MULTILINE_CPP_IS_BRIEF tag can be set to YES to make doxygen treat a
+# multi-line C++ special comment block (i.e. a block of //! or /// comments) as
+# a brief description. This used to be the default behavior. The new default is
+# to treat a multi-line C++ comment block as a detailed description. Set this
+# tag to YES if you prefer the old behavior instead.
+#
+# Note that setting this tag to YES also means that rational rose comments are
+# not recognized any more.
+# The default value is: NO.
+
+MULTILINE_CPP_IS_BRIEF = NO
+
+# If the INHERIT_DOCS tag is set to YES then an undocumented member inherits the
+# documentation from any documented member that it re-implements.
+# The default value is: YES.
+
+INHERIT_DOCS = YES
+
+# If the SEPARATE_MEMBER_PAGES tag is set to YES then doxygen will produce a new
+# page for each member. If set to NO, the documentation of a member will be part
+# of the file/class/namespace that contains it.
+# The default value is: NO.
+
+SEPARATE_MEMBER_PAGES = NO
+
+# The TAB_SIZE tag can be used to set the number of spaces in a tab. Doxygen
+# uses this value to replace tabs by spaces in code fragments.
+# Minimum value: 1, maximum value: 16, default value: 4.
+
+TAB_SIZE = 4
+
+# This tag can be used to specify a number of aliases that act as commands in
+# the documentation. An alias has the form:
+# name=value
+# For example adding
+# "sideeffect=@par Side Effects:\n"
+# will allow you to put the command \sideeffect (or @sideeffect) in the
+# documentation, which will result in a user-defined paragraph with heading
+# "Side Effects:". You can put \n's in the value part of an alias to insert
+# newlines.
+
+ALIASES =
+
+# This tag can be used to specify a number of word-keyword mappings (TCL only).
+# A mapping has the form "name=value". For example adding "class=itcl::class"
+# will allow you to use the command class in the itcl::class meaning.
+
+TCL_SUBST =
+
+# Set the OPTIMIZE_OUTPUT_FOR_C tag to YES if your project consists of C sources
+# only. Doxygen will then generate output that is more tailored for C. For
+# instance, some of the names that are used will be different. The list of all
+# members will be omitted, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_FOR_C = YES
+
+# Set the OPTIMIZE_OUTPUT_JAVA tag to YES if your project consists of Java or
+# Python sources only. Doxygen will then generate output that is more tailored
+# for that language. For instance, namespaces will be presented as packages,
+# qualified scopes will look different, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_JAVA = NO
+
+# Set the OPTIMIZE_FOR_FORTRAN tag to YES if your project consists of Fortran
+# sources. Doxygen will then generate output that is tailored for Fortran.
+# The default value is: NO.
+
+OPTIMIZE_FOR_FORTRAN = NO
+
+# Set the OPTIMIZE_OUTPUT_VHDL tag to YES if your project consists of VHDL
+# sources. Doxygen will then generate output that is tailored for VHDL.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_VHDL = NO
+
+# Doxygen selects the parser to use depending on the extension of the files it
+# parses. With this tag you can assign which parser to use for a given
+# extension. Doxygen has a built-in mapping, but you can override or extend it
+# using this tag. The format is ext=language, where ext is a file extension, and
+# language is one of the parsers supported by doxygen: IDL, Java, Javascript,
+# C#, C, C++, D, PHP, Objective-C, Python, Fortran (fixed format Fortran:
+# FortranFixed, free formatted Fortran: FortranFree, unknown formatted Fortran:
+# Fortran. In the later case the parser tries to guess whether the code is fixed
+# or free formatted code, this is the default for Fortran type files), VHDL. For
+# instance to make doxygen treat .inc files as Fortran files (default is PHP),
+# and .f files as C (default is Fortran), use: inc=Fortran f=C.
+#
+# Note: For files without extension you can use no_extension as a placeholder.
+#
+# Note that for custom extensions you also need to set FILE_PATTERNS otherwise
+# the files are not read by doxygen.
+
+EXTENSION_MAPPING =
+
+# If the MARKDOWN_SUPPORT tag is enabled then doxygen pre-processes all comments
+# according to the Markdown format, which allows for more readable
+# documentation. See http://daringfireball.net/projects/markdown/ for details.
+# The output of markdown processing is further processed by doxygen, so you can
+# mix doxygen, HTML, and XML commands with Markdown formatting. Disable only in
+# case of backward compatibilities issues.
+# The default value is: YES.
+
+MARKDOWN_SUPPORT = YES
+
+# When the TOC_INCLUDE_HEADINGS tag is set to a non-zero value, all headings up
+# to that level are automatically included in the table of contents, even if
+# they do not have an id attribute.
+# Note: This feature currently applies only to Markdown headings.
+# Minimum value: 0, maximum value: 99, default value: 0.
+# This tag requires that the tag MARKDOWN_SUPPORT is set to YES.
+
+TOC_INCLUDE_HEADINGS = 0
+
+# When enabled doxygen tries to link words that correspond to documented
+# classes, or namespaces to their corresponding documentation. Such a link can
+# be prevented in individual cases by putting a % sign in front of the word or
+# globally by setting AUTOLINK_SUPPORT to NO.
+# The default value is: YES.
+
+AUTOLINK_SUPPORT = YES
+
+# If you use STL classes (i.e. std::string, std::vector, etc.) but do not want
+# to include (a tag file for) the STL sources as input, then you should set this
+# tag to YES in order to let doxygen match functions declarations and
+# definitions whose arguments contain STL classes (e.g. func(std::string);
+# versus func(std::string) {}). This also make the inheritance and collaboration
+# diagrams that involve STL classes more complete and accurate.
+# The default value is: NO.
+
+BUILTIN_STL_SUPPORT = NO
+
+# If you use Microsoft's C++/CLI language, you should set this option to YES to
+# enable parsing support.
+# The default value is: NO.
+
+CPP_CLI_SUPPORT = NO
+
+# Set the SIP_SUPPORT tag to YES if your project consists of sip (see:
+# http://www.riverbankcomputing.co.uk/software/sip/intro) sources only. Doxygen
+# will parse them like normal C++ but will assume all classes use public instead
+# of private inheritance when no explicit protection keyword is present.
+# The default value is: NO.
+
+SIP_SUPPORT = NO
+
+# For Microsoft's IDL there are propget and propput attributes to indicate
+# getter and setter methods for a property. Setting this option to YES will make
+# doxygen to replace the get and set methods by a property in the documentation.
+# This will only work if the methods are indeed getting or setting a simple
+# type. If this is not the case, or you want to show the methods anyway, you
+# should set this option to NO.
+# The default value is: YES.
+
+IDL_PROPERTY_SUPPORT = YES
+
+# If member grouping is used in the documentation and the DISTRIBUTE_GROUP_DOC
+# tag is set to YES then doxygen will reuse the documentation of the first
+# member in the group (if any) for the other members of the group. By default
+# all members of a group must be documented explicitly.
+# The default value is: NO.
+
+DISTRIBUTE_GROUP_DOC = NO
+
+# If one adds a struct or class to a group and this option is enabled, then also
+# any nested class or struct is added to the same group. By default this option
+# is disabled and one has to add nested compounds explicitly via \ingroup.
+# The default value is: NO.
+
+GROUP_NESTED_COMPOUNDS = NO
+
+# Set the SUBGROUPING tag to YES to allow class member groups of the same type
+# (for instance a group of public functions) to be put as a subgroup of that
+# type (e.g. under the Public Functions section). Set it to NO to prevent
+# subgrouping. Alternatively, this can be done per class using the
+# \nosubgrouping command.
+# The default value is: YES.
+
+SUBGROUPING = YES
+
+# When the INLINE_GROUPED_CLASSES tag is set to YES, classes, structs and unions
+# are shown inside the group in which they are included (e.g. using \ingroup)
+# instead of on a separate page (for HTML and Man pages) or section (for LaTeX
+# and RTF).
+#
+# Note that this feature does not work in combination with
+# SEPARATE_MEMBER_PAGES.
+# The default value is: NO.
+
+INLINE_GROUPED_CLASSES = NO
+
+# When the INLINE_SIMPLE_STRUCTS tag is set to YES, structs, classes, and unions
+# with only public data fields or simple typedef fields will be shown inline in
+# the documentation of the scope in which they are defined (i.e. file,
+# namespace, or group documentation), provided this scope is documented. If set
+# to NO, structs, classes, and unions are shown on a separate page (for HTML and
+# Man pages) or section (for LaTeX and RTF).
+# The default value is: NO.
+
+INLINE_SIMPLE_STRUCTS = NO
+
+# When TYPEDEF_HIDES_STRUCT tag is enabled, a typedef of a struct, union, or
+# enum is documented as struct, union, or enum with the name of the typedef. So
+# typedef struct TypeS {} TypeT, will appear in the documentation as a struct
+# with name TypeT. When disabled the typedef will appear as a member of a file,
+# namespace, or class. And the struct will be named TypeS. This can typically be
+# useful for C code in case the coding convention dictates that all compound
+# types are typedef'ed and only the typedef is referenced, never the tag name.
+# The default value is: NO.
+
+TYPEDEF_HIDES_STRUCT = NO
+
+# The size of the symbol lookup cache can be set using LOOKUP_CACHE_SIZE. This
+# cache is used to resolve symbols given their name and scope. Since this can be
+# an expensive process and often the same symbol appears multiple times in the
+# code, doxygen keeps a cache of pre-resolved symbols. If the cache is too small
+# doxygen will become slower. If the cache is too large, memory is wasted. The
+# cache size is given by this formula: 2^(16+LOOKUP_CACHE_SIZE). The valid range
+# is 0..9, the default is 0, corresponding to a cache size of 2^16=65536
+# symbols. At the end of a run doxygen will report the cache usage and suggest
+# the optimal cache size from a speed point of view.
+# Minimum value: 0, maximum value: 9, default value: 0.
+
+LOOKUP_CACHE_SIZE = 0
+
+#---------------------------------------------------------------------------
+# Build related configuration options
+#---------------------------------------------------------------------------
+
+# If the EXTRACT_ALL tag is set to YES, doxygen will assume all entities in
+# documentation are documented, even if no documentation was available. Private
+# class members and static file members will be hidden unless the
+# EXTRACT_PRIVATE respectively EXTRACT_STATIC tags are set to YES.
+# Note: This will also disable the warnings about undocumented members that are
+# normally produced when WARNINGS is set to YES.
+# The default value is: NO.
+
+EXTRACT_ALL = YES
+
+# If the EXTRACT_PRIVATE tag is set to YES, all private members of a class will
+# be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PRIVATE = NO
+
+# If the EXTRACT_PACKAGE tag is set to YES, all members with package or internal
+# scope will be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PACKAGE = NO
+
+# If the EXTRACT_STATIC tag is set to YES, all static members of a file will be
+# included in the documentation.
+# The default value is: NO.
+
+EXTRACT_STATIC = NO
+
+# If the EXTRACT_LOCAL_CLASSES tag is set to YES, classes (and structs) defined
+# locally in source files will be included in the documentation. If set to NO,
+# only classes defined in header files are included. Does not have any effect
+# for Java sources.
+# The default value is: YES.
+
+EXTRACT_LOCAL_CLASSES = YES
+
+# This flag is only useful for Objective-C code. If set to YES, local methods,
+# which are defined in the implementation section but not in the interface are
+# included in the documentation. If set to NO, only methods in the interface are
+# included.
+# The default value is: NO.
+
+EXTRACT_LOCAL_METHODS = NO
+
+# If this flag is set to YES, the members of anonymous namespaces will be
+# extracted and appear in the documentation as a namespace called
+# 'anonymous_namespace{file}', where file will be replaced with the base name of
+# the file that contains the anonymous namespace. By default anonymous namespace
+# are hidden.
+# The default value is: NO.
+
+EXTRACT_ANON_NSPACES = NO
+
+# If the HIDE_UNDOC_MEMBERS tag is set to YES, doxygen will hide all
+# undocumented members inside documented classes or files. If set to NO these
+# members will be included in the various overviews, but no documentation
+# section is generated. This option has no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_MEMBERS = NO
+
+# If the HIDE_UNDOC_CLASSES tag is set to YES, doxygen will hide all
+# undocumented classes that are normally visible in the class hierarchy. If set
+# to NO, these classes will be included in the various overviews. This option
+# has no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_CLASSES = NO
+
+# If the HIDE_FRIEND_COMPOUNDS tag is set to YES, doxygen will hide all friend
+# (class|struct|union) declarations. If set to NO, these declarations will be
+# included in the documentation.
+# The default value is: NO.
+
+HIDE_FRIEND_COMPOUNDS = NO
+
+# If the HIDE_IN_BODY_DOCS tag is set to YES, doxygen will hide any
+# documentation blocks found inside the body of a function. If set to NO, these
+# blocks will be appended to the function's detailed documentation block.
+# The default value is: NO.
+
+HIDE_IN_BODY_DOCS = NO
+
+# The INTERNAL_DOCS tag determines if documentation that is typed after a
+# \internal command is included. If the tag is set to NO then the documentation
+# will be excluded. Set it to YES to include the internal documentation.
+# The default value is: NO.
+
+INTERNAL_DOCS = NO
+
+# If the CASE_SENSE_NAMES tag is set to NO then doxygen will only generate file
+# names in lower-case letters. If set to YES, upper-case letters are also
+# allowed. This is useful if you have classes or files whose names only differ
+# in case and if your file system supports case sensitive file names. Windows
+# and Mac users are advised to set this option to NO.
+# The default value is: system dependent.
+
+CASE_SENSE_NAMES = NO
+
+# If the HIDE_SCOPE_NAMES tag is set to NO then doxygen will show members with
+# their full class and namespace scopes in the documentation. If set to YES, the
+# scope will be hidden.
+# The default value is: NO.
+
+HIDE_SCOPE_NAMES = YES
+
+# If the HIDE_COMPOUND_REFERENCE tag is set to NO (default) then doxygen will
+# append additional text to a page's title, such as Class Reference. If set to
+# YES the compound reference will be hidden.
+# The default value is: NO.
+
+HIDE_COMPOUND_REFERENCE= NO
+
+# If the SHOW_INCLUDE_FILES tag is set to YES then doxygen will put a list of
+# the files that are included by a file in the documentation of that file.
+# The default value is: YES.
+
+SHOW_INCLUDE_FILES = YES
+
+# If the SHOW_GROUPED_MEMB_INC tag is set to YES then Doxygen will add for each
+# grouped member an include statement to the documentation, telling the reader
+# which file to include in order to use the member.
+# The default value is: NO.
+
+SHOW_GROUPED_MEMB_INC = NO
+
+# If the FORCE_LOCAL_INCLUDES tag is set to YES then doxygen will list include
+# files with double quotes in the documentation rather than with sharp brackets.
+# The default value is: NO.
+
+FORCE_LOCAL_INCLUDES = NO
+
+# If the INLINE_INFO tag is set to YES then a tag [inline] is inserted in the
+# documentation for inline members.
+# The default value is: YES.
+
+INLINE_INFO = YES
+
+# If the SORT_MEMBER_DOCS tag is set to YES then doxygen will sort the
+# (detailed) documentation of file and class members alphabetically by member
+# name. If set to NO, the members will appear in declaration order.
+# The default value is: YES.
+
+SORT_MEMBER_DOCS = YES
+
+# If the SORT_BRIEF_DOCS tag is set to YES then doxygen will sort the brief
+# descriptions of file, namespace and class members alphabetically by member
+# name. If set to NO, the members will appear in declaration order. Note that
+# this will also influence the order of the classes in the class list.
+# The default value is: NO.
+
+SORT_BRIEF_DOCS = NO
+
+# If the SORT_MEMBERS_CTORS_1ST tag is set to YES then doxygen will sort the
+# (brief and detailed) documentation of class members so that constructors and
+# destructors are listed first. If set to NO the constructors will appear in the
+# respective orders defined by SORT_BRIEF_DOCS and SORT_MEMBER_DOCS.
+# Note: If SORT_BRIEF_DOCS is set to NO this option is ignored for sorting brief
+# member documentation.
+# Note: If SORT_MEMBER_DOCS is set to NO this option is ignored for sorting
+# detailed member documentation.
+# The default value is: NO.
+
+SORT_MEMBERS_CTORS_1ST = NO
+
+# If the SORT_GROUP_NAMES tag is set to YES then doxygen will sort the hierarchy
+# of group names into alphabetical order. If set to NO the group names will
+# appear in their defined order.
+# The default value is: NO.
+
+SORT_GROUP_NAMES = NO
+
+# If the SORT_BY_SCOPE_NAME tag is set to YES, the class list will be sorted by
+# fully-qualified names, including namespaces. If set to NO, the class list will
+# be sorted only by class name, not including the namespace part.
+# Note: This option is not very useful if HIDE_SCOPE_NAMES is set to YES.
+# Note: This option applies only to the class list, not to the alphabetical
+# list.
+# The default value is: NO.
+
+SORT_BY_SCOPE_NAME = NO
+
+# If the STRICT_PROTO_MATCHING option is enabled and doxygen fails to do proper
+# type resolution of all parameters of a function it will reject a match between
+# the prototype and the implementation of a member function even if there is
+# only one candidate or it is obvious which candidate to choose by doing a
+# simple string match. By disabling STRICT_PROTO_MATCHING doxygen will still
+# accept a match between prototype and implementation in such cases.
+# The default value is: NO.
+
+STRICT_PROTO_MATCHING = NO
+
+# The GENERATE_TODOLIST tag can be used to enable (YES) or disable (NO) the todo
+# list. This list is created by putting \todo commands in the documentation.
+# The default value is: YES.
+
+GENERATE_TODOLIST = YES
+
+# The GENERATE_TESTLIST tag can be used to enable (YES) or disable (NO) the test
+# list. This list is created by putting \test commands in the documentation.
+# The default value is: YES.
+
+GENERATE_TESTLIST = YES
+
+# The GENERATE_BUGLIST tag can be used to enable (YES) or disable (NO) the bug
+# list. This list is created by putting \bug commands in the documentation.
+# The default value is: YES.
+
+GENERATE_BUGLIST = YES
+
+# The GENERATE_DEPRECATEDLIST tag can be used to enable (YES) or disable (NO)
+# the deprecated list. This list is created by putting \deprecated commands in
+# the documentation.
+# The default value is: YES.
+
+GENERATE_DEPRECATEDLIST= YES
+
+# The ENABLED_SECTIONS tag can be used to enable conditional documentation
+# sections, marked by \if <section_label> ... \endif and \cond <section_label>
+# ... \endcond blocks.
+
+ENABLED_SECTIONS =
+
+# The MAX_INITIALIZER_LINES tag determines the maximum number of lines that the
+# initial value of a variable or macro / define can have for it to appear in the
+# documentation. If the initializer consists of more lines than specified here
+# it will be hidden. Use a value of 0 to hide initializers completely. The
+# appearance of the value of individual variables and macros / defines can be
+# controlled using \showinitializer or \hideinitializer command in the
+# documentation regardless of this setting.
+# Minimum value: 0, maximum value: 10000, default value: 30.
+
+MAX_INITIALIZER_LINES = 30
+
+# Set the SHOW_USED_FILES tag to NO to disable the list of files generated at
+# the bottom of the documentation of classes and structs. If set to YES, the
+# list will mention the files that were used to generate the documentation.
+# The default value is: YES.
+
+SHOW_USED_FILES = YES
+
+# Set the SHOW_FILES tag to NO to disable the generation of the Files page. This
+# will remove the Files entry from the Quick Index and from the Folder Tree View
+# (if specified).
+# The default value is: YES.
+
+SHOW_FILES = YES
+
+# Set the SHOW_NAMESPACES tag to NO to disable the generation of the Namespaces
+# page. This will remove the Namespaces entry from the Quick Index and from the
+# Folder Tree View (if specified).
+# The default value is: YES.
+
+SHOW_NAMESPACES = YES
+
+# The FILE_VERSION_FILTER tag can be used to specify a program or script that
+# doxygen should invoke to get the current version for each file (typically from
+# the version control system). Doxygen will invoke the program by executing (via
+# popen()) the command command input-file, where command is the value of the
+# FILE_VERSION_FILTER tag, and input-file is the name of an input file provided
+# by doxygen. Whatever the program writes to standard output is used as the file
+# version. For an example see the documentation.
+
+FILE_VERSION_FILTER =
+
+# The LAYOUT_FILE tag can be used to specify a layout file which will be parsed
+# by doxygen. The layout file controls the global structure of the generated
+# output files in an output format independent way. To create the layout file
+# that represents doxygen's defaults, run doxygen with the -l option. You can
+# optionally specify a file name after the option, if omitted DoxygenLayout.xml
+# will be used as the name of the layout file.
+#
+# Note that if you run doxygen from a directory containing a file called
+# DoxygenLayout.xml, doxygen will parse it automatically even if the LAYOUT_FILE
+# tag is left empty.
+
+LAYOUT_FILE =
+
+# The CITE_BIB_FILES tag can be used to specify one or more bib files containing
+# the reference definitions. This must be a list of .bib files. The .bib
+# extension is automatically appended if omitted. This requires the bibtex tool
+# to be installed. See also http://en.wikipedia.org/wiki/BibTeX for more info.
+# For LaTeX the style of the bibliography can be controlled using
+# LATEX_BIB_STYLE. To use this feature you need bibtex and perl available in the
+# search path. See also \cite for info how to create references.
+
+CITE_BIB_FILES =
+
+#---------------------------------------------------------------------------
+# Configuration options related to warning and progress messages
+#---------------------------------------------------------------------------
+
+# The QUIET tag can be used to turn on/off the messages that are generated to
+# standard output by doxygen. If QUIET is set to YES this implies that the
+# messages are off.
+# The default value is: NO.
+
+QUIET = NO
+
+# The WARNINGS tag can be used to turn on/off the warning messages that are
+# generated to standard error (stderr) by doxygen. If WARNINGS is set to YES
+# this implies that the warnings are on.
+#
+# Tip: Turn warnings on while writing the documentation.
+# The default value is: YES.
+
+WARNINGS = YES
+
+# If the WARN_IF_UNDOCUMENTED tag is set to YES then doxygen will generate
+# warnings for undocumented members. If EXTRACT_ALL is set to YES then this flag
+# will automatically be disabled.
+# The default value is: YES.
+
+WARN_IF_UNDOCUMENTED = YES
+
+# If the WARN_IF_DOC_ERROR tag is set to YES, doxygen will generate warnings for
+# potential errors in the documentation, such as not documenting some parameters
+# in a documented function, or documenting parameters that don't exist or using
+# markup commands wrongly.
+# The default value is: YES.
+
+WARN_IF_DOC_ERROR = YES
+
+# This WARN_NO_PARAMDOC option can be enabled to get warnings for functions that
+# are documented, but have no documentation for their parameters or return
+# value. If set to NO, doxygen will only warn about wrong or incomplete
+# parameter documentation, but not about the absence of documentation.
+# The default value is: NO.
+
+WARN_NO_PARAMDOC = NO
+
+# If the WARN_AS_ERROR tag is set to YES then doxygen will immediately stop when
+# a warning is encountered.
+# The default value is: NO.
+
+WARN_AS_ERROR = NO
+
+# The WARN_FORMAT tag determines the format of the warning messages that doxygen
+# can produce. The string should contain the $file, $line, and $text tags, which
+# will be replaced by the file and line number from which the warning originated
+# and the warning text. Optionally the format may contain $version, which will
+# be replaced by the version of the file (if it could be obtained via
+# FILE_VERSION_FILTER)
+# The default value is: $file:$line: $text.
+
+WARN_FORMAT = "$file:$line: $text"
+
+# The WARN_LOGFILE tag can be used to specify a file to which warning and error
+# messages should be written. If left blank the output is written to standard
+# error (stderr).
+
+WARN_LOGFILE =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the input files
+#---------------------------------------------------------------------------
+
+# The INPUT tag is used to specify the files and/or directories that contain
+# documented source files. You may enter file names like myfile.cpp or
+# directories like /usr/src/myproject. Separate the files or directories with
+# spaces. See also FILE_PATTERNS and EXTENSION_MAPPING
+# Note: If this tag is empty the current directory is searched.
+
+INPUT = ../../
+
+# This tag can be used to specify the character encoding of the source files
+# that doxygen parses. Internally doxygen uses the UTF-8 encoding. Doxygen uses
+# libiconv (or the iconv built into libc) for the transcoding. See the libiconv
+# documentation (see: http://www.gnu.org/software/libiconv) for the list of
+# possible encodings.
+# The default value is: UTF-8.
+
+INPUT_ENCODING = UTF-8
+
+# If the value of the INPUT tag contains directories, you can use the
+# FILE_PATTERNS tag to specify one or more wildcard patterns (like *.cpp and
+# *.h) to filter out the source-files in the directories.
+#
+# Note that for custom extensions or not directly supported extensions you also
+# need to set EXTENSION_MAPPING for the extension otherwise the files are not
+# read by doxygen.
+#
+# If left blank the following patterns are tested:*.c, *.cc, *.cxx, *.cpp,
+# *.c++, *.java, *.ii, *.ixx, *.ipp, *.i++, *.inl, *.idl, *.ddl, *.odl, *.h,
+# *.hh, *.hxx, *.hpp, *.h++, *.cs, *.d, *.php, *.php4, *.php5, *.phtml, *.inc,
+# *.m, *.markdown, *.md, *.mm, *.dox, *.py, *.pyw, *.f90, *.f95, *.f03, *.f08,
+# *.f, *.for, *.tcl, *.vhd, *.vhdl, *.ucf and *.qsf.
+
+FILE_PATTERNS = *.c \
+ *.cc \
+ *.cxx \
+ *.cpp \
+ *.c++ \
+ *.java \
+ *.ii \
+ *.ixx \
+ *.ipp \
+ *.i++ \
+ *.inl \
+ *.idl \
+ *.ddl \
+ *.odl \
+ *.h \
+ *.hh \
+ *.hxx \
+ *.hpp \
+ *.h++ \
+ *.cs \
+ *.d \
+ *.php \
+ *.php4 \
+ *.php5 \
+ *.phtml \
+ *.inc \
+ *.m \
+ *.markdown \
+ *.md \
+ *.mm \
+ *.dox \
+ *.py \
+ *.pyw \
+ *.f90 \
+ *.f95 \
+ *.f03 \
+ *.f08 \
+ *.f \
+ *.for \
+ *.tcl \
+ *.vhd \
+ *.vhdl \
+ *.ucf \
+ *.qsf
+
+# The RECURSIVE tag can be used to specify whether or not subdirectories should
+# be searched for input files as well.
+# The default value is: NO.
+
+RECURSIVE = NO
+
+# The EXCLUDE tag can be used to specify files and/or directories that should be
+# excluded from the INPUT source files. This way you can easily exclude a
+# subdirectory from a directory tree whose root is specified with the INPUT tag.
+#
+# Note that relative paths are relative to the directory from which doxygen is
+# run.
+
+EXCLUDE =
+
+# The EXCLUDE_SYMLINKS tag can be used to select whether or not files or
+# directories that are symbolic links (a Unix file system feature) are excluded
+# from the input.
+# The default value is: NO.
+
+EXCLUDE_SYMLINKS = NO
+
+# If the value of the INPUT tag contains directories, you can use the
+# EXCLUDE_PATTERNS tag to specify one or more wildcard patterns to exclude
+# certain files from those directories.
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories for example use the pattern */test/*
+
+EXCLUDE_PATTERNS =
+
+# The EXCLUDE_SYMBOLS tag can be used to specify one or more symbol names
+# (namespaces, classes, functions, etc.) that should be excluded from the
+# output. The symbol name can be a fully qualified name, a word, or if the
+# wildcard * is used, a substring. Examples: ANamespace, AClass,
+# AClass::ANamespace, ANamespace::*Test
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories use the pattern */test/*
+
+EXCLUDE_SYMBOLS =
+
+# The EXAMPLE_PATH tag can be used to specify one or more files or directories
+# that contain example code fragments that are included (see the \include
+# command).
+
+EXAMPLE_PATH =
+
+# If the value of the EXAMPLE_PATH tag contains directories, you can use the
+# EXAMPLE_PATTERNS tag to specify one or more wildcard pattern (like *.cpp and
+# *.h) to filter out the source-files in the directories. If left blank all
+# files are included.
+
+EXAMPLE_PATTERNS = *
+
+# If the EXAMPLE_RECURSIVE tag is set to YES then subdirectories will be
+# searched for input files to be used with the \include or \dontinclude commands
+# irrespective of the value of the RECURSIVE tag.
+# The default value is: NO.
+
+EXAMPLE_RECURSIVE = NO
+
+# The IMAGE_PATH tag can be used to specify one or more files or directories
+# that contain images that are to be included in the documentation (see the
+# \image command).
+
+IMAGE_PATH =
+
+# The INPUT_FILTER tag can be used to specify a program that doxygen should
+# invoke to filter for each input file. Doxygen will invoke the filter program
+# by executing (via popen()) the command:
+#
+# <filter> <input-file>
+#
+# where <filter> is the value of the INPUT_FILTER tag, and <input-file> is the
+# name of an input file. Doxygen will then use the output that the filter
+# program writes to standard output. If FILTER_PATTERNS is specified, this tag
+# will be ignored.
+#
+# Note that the filter must not add or remove lines; it is applied before the
+# code is scanned, but not when the output code is generated. If lines are added
+# or removed, the anchors will not be placed correctly.
+#
+# Note that for custom extensions or not directly supported extensions you also
+# need to set EXTENSION_MAPPING for the extension otherwise the files are not
+# properly processed by doxygen.
+
+INPUT_FILTER =
+
+# The FILTER_PATTERNS tag can be used to specify filters on a per file pattern
+# basis. Doxygen will compare the file name with each pattern and apply the
+# filter if there is a match. The filters are a list of the form: pattern=filter
+# (like *.cpp=my_cpp_filter). See INPUT_FILTER for further information on how
+# filters are used. If the FILTER_PATTERNS tag is empty or if none of the
+# patterns match the file name, INPUT_FILTER is applied.
+#
+# Note that for custom extensions or not directly supported extensions you also
+# need to set EXTENSION_MAPPING for the extension otherwise the files are not
+# properly processed by doxygen.
+
+FILTER_PATTERNS =
+
+# If the FILTER_SOURCE_FILES tag is set to YES, the input filter (if set using
+# INPUT_FILTER) will also be used to filter the input files that are used for
+# producing the source files to browse (i.e. when SOURCE_BROWSER is set to YES).
+# The default value is: NO.
+
+FILTER_SOURCE_FILES = NO
+
+# The FILTER_SOURCE_PATTERNS tag can be used to specify source filters per file
+# pattern. A pattern will override the setting for FILTER_PATTERN (if any) and
+# it is also possible to disable source filtering for a specific pattern using
+# *.ext= (so without naming a filter).
+# This tag requires that the tag FILTER_SOURCE_FILES is set to YES.
+
+FILTER_SOURCE_PATTERNS =
+
+# If the USE_MDFILE_AS_MAINPAGE tag refers to the name of a markdown file that
+# is part of the input, its contents will be placed on the main page
+# (index.html). This can be useful if you have a project on for instance GitHub
+# and want to reuse the introduction page also for the doxygen output.
+
+USE_MDFILE_AS_MAINPAGE =
+
+#---------------------------------------------------------------------------
+# Configuration options related to source browsing
+#---------------------------------------------------------------------------
+
+# If the SOURCE_BROWSER tag is set to YES then a list of source files will be
+# generated. Documented entities will be cross-referenced with these sources.
+#
+# Note: To get rid of all source code in the generated output, make sure that
+# also VERBATIM_HEADERS is set to NO.
+# The default value is: NO.
+
+SOURCE_BROWSER = NO
+
+# Setting the INLINE_SOURCES tag to YES will include the body of functions,
+# classes and enums directly into the documentation.
+# The default value is: NO.
+
+INLINE_SOURCES = NO
+
+# Setting the STRIP_CODE_COMMENTS tag to YES will instruct doxygen to hide any
+# special comment blocks from generated source code fragments. Normal C, C++ and
+# Fortran comments will always remain visible.
+# The default value is: YES.
+
+STRIP_CODE_COMMENTS = YES
+
+# If the REFERENCED_BY_RELATION tag is set to YES then for each documented
+# function all documented functions referencing it will be listed.
+# The default value is: NO.
+
+REFERENCED_BY_RELATION = NO
+
+# If the REFERENCES_RELATION tag is set to YES then for each documented function
+# all documented entities called/used by that function will be listed.
+# The default value is: NO.
+
+REFERENCES_RELATION = NO
+
+# If the REFERENCES_LINK_SOURCE tag is set to YES and SOURCE_BROWSER tag is set
+# to YES then the hyperlinks from functions in REFERENCES_RELATION and
+# REFERENCED_BY_RELATION lists will link to the source code. Otherwise they will
+# link to the documentation.
+# The default value is: YES.
+
+REFERENCES_LINK_SOURCE = YES
+
+# If SOURCE_TOOLTIPS is enabled (the default) then hovering a hyperlink in the
+# source code will show a tooltip with additional information such as prototype,
+# brief description and links to the definition and documentation. Since this
+# will make the HTML file larger and loading of large files a bit slower, you
+# can opt to disable this feature.
+# The default value is: YES.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+SOURCE_TOOLTIPS = YES
+
+# If the USE_HTAGS tag is set to YES then the references to source code will
+# point to the HTML generated by the htags(1) tool instead of doxygen built-in
+# source browser. The htags tool is part of GNU's global source tagging system
+# (see http://www.gnu.org/software/global/global.html). You will need version
+# 4.8.6 or higher.
+#
+# To use it do the following:
+# - Install the latest version of global
+# - Enable SOURCE_BROWSER and USE_HTAGS in the config file
+# - Make sure the INPUT points to the root of the source tree
+# - Run doxygen as normal
+#
+# Doxygen will invoke htags (and that will in turn invoke gtags), so these
+# tools must be available from the command line (i.e. in the search path).
+#
+# The result: instead of the source browser generated by doxygen, the links to
+# source code will now point to the output of htags.
+# The default value is: NO.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+USE_HTAGS = NO
+
+# If the VERBATIM_HEADERS tag is set the YES then doxygen will generate a
+# verbatim copy of the header file for each class for which an include is
+# specified. Set to NO to disable this.
+# See also: Section \class.
+# The default value is: YES.
+
+VERBATIM_HEADERS = YES
+
+# If the CLANG_ASSISTED_PARSING tag is set to YES then doxygen will use the
+# clang parser (see: http://clang.llvm.org/) for more accurate parsing at the
+# cost of reduced performance. This can be particularly helpful with template
+# rich C++ code for which doxygen's built-in parser lacks the necessary type
+# information.
+# Note: The availability of this option depends on whether or not doxygen was
+# generated with the -Duse-libclang=ON option for CMake.
+# The default value is: NO.
+
+CLANG_ASSISTED_PARSING = NO
+
+# If clang assisted parsing is enabled you can provide the compiler with command
+# line options that you would normally use when invoking the compiler. Note that
+# the include paths will already be set by doxygen for the files and directories
+# specified with INPUT and INCLUDE_PATH.
+# This tag requires that the tag CLANG_ASSISTED_PARSING is set to YES.
+
+CLANG_OPTIONS =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the alphabetical class index
+#---------------------------------------------------------------------------
+
+# If the ALPHABETICAL_INDEX tag is set to YES, an alphabetical index of all
+# compounds will be generated. Enable this if the project contains a lot of
+# classes, structs, unions or interfaces.
+# The default value is: YES.
+
+ALPHABETICAL_INDEX = YES
+
+# The COLS_IN_ALPHA_INDEX tag can be used to specify the number of columns in
+# which the alphabetical index list will be split.
+# Minimum value: 1, maximum value: 20, default value: 5.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+COLS_IN_ALPHA_INDEX = 5
+
+# In case all classes in a project start with a common prefix, all classes will
+# be put under the same header in the alphabetical index. The IGNORE_PREFIX tag
+# can be used to specify a prefix (or a list of prefixes) that should be ignored
+# while generating the index headers.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+IGNORE_PREFIX =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the HTML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_HTML tag is set to YES, doxygen will generate HTML output
+# The default value is: YES.
+
+GENERATE_HTML = NO
+
+# The HTML_OUTPUT tag is used to specify where the HTML docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_OUTPUT = html
+
+# The HTML_FILE_EXTENSION tag can be used to specify the file extension for each
+# generated HTML page (for example: .htm, .php, .asp).
+# The default value is: .html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FILE_EXTENSION = .html
+
+# The HTML_HEADER tag can be used to specify a user-defined HTML header file for
+# each generated HTML page. If the tag is left blank doxygen will generate a
+# standard header.
+#
+# To get valid HTML the header file that includes any scripts and style sheets
+# that doxygen needs, which is dependent on the configuration options used (e.g.
+# the setting GENERATE_TREEVIEW). It is highly recommended to start with a
+# default header using
+# doxygen -w html new_header.html new_footer.html new_stylesheet.css
+# YourConfigFile
+# and then modify the file new_header.html. See also section "Doxygen usage"
+# for information on how to generate the default header that doxygen normally
+# uses.
+# Note: The header is subject to change so you typically have to regenerate the
+# default header when upgrading to a newer version of doxygen. For a description
+# of the possible markers and block names see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_HEADER =
+
+# The HTML_FOOTER tag can be used to specify a user-defined HTML footer for each
+# generated HTML page. If the tag is left blank doxygen will generate a standard
+# footer. See HTML_HEADER for more information on how to generate a default
+# footer and what special commands can be used inside the footer. See also
+# section "Doxygen usage" for information on how to generate the default footer
+# that doxygen normally uses.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FOOTER =
+
+# The HTML_STYLESHEET tag can be used to specify a user-defined cascading style
+# sheet that is used by each HTML page. It can be used to fine-tune the look of
+# the HTML output. If left blank doxygen will generate a default style sheet.
+# See also section "Doxygen usage" for information on how to generate the style
+# sheet that doxygen normally uses.
+# Note: It is recommended to use HTML_EXTRA_STYLESHEET instead of this tag, as
+# it is more robust and this tag (HTML_STYLESHEET) will in the future become
+# obsolete.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_STYLESHEET =
+
+# The HTML_EXTRA_STYLESHEET tag can be used to specify additional user-defined
+# cascading style sheets that are included after the standard style sheets
+# created by doxygen. Using this option one can overrule certain style aspects.
+# This is preferred over using HTML_STYLESHEET since it does not replace the
+# standard style sheet and is therefore more robust against future updates.
+# Doxygen will copy the style sheet files to the output directory.
+# Note: The order of the extra style sheet files is of importance (e.g. the last
+# style sheet in the list overrules the setting of the previous ones in the
+# list). For an example see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_STYLESHEET =
+
+# The HTML_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the HTML output directory. Note
+# that these files will be copied to the base HTML output directory. Use the
+# $relpath^ marker in the HTML_HEADER and/or HTML_FOOTER files to load these
+# files. In the HTML_STYLESHEET file, use the file name only. Also note that the
+# files will be copied as-is; there are no commands or markers available.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_FILES =
+
+# The HTML_COLORSTYLE_HUE tag controls the color of the HTML output. Doxygen
+# will adjust the colors in the style sheet and background images according to
+# this color. Hue is specified as an angle on a colorwheel, see
+# http://en.wikipedia.org/wiki/Hue for more information. For instance the value
+# 0 represents red, 60 is yellow, 120 is green, 180 is cyan, 240 is blue, 300
+# purple, and 360 is red again.
+# Minimum value: 0, maximum value: 359, default value: 220.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_HUE = 220
+
+# The HTML_COLORSTYLE_SAT tag controls the purity (or saturation) of the colors
+# in the HTML output. For a value of 0 the output will use grayscales only. A
+# value of 255 will produce the most vivid colors.
+# Minimum value: 0, maximum value: 255, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_SAT = 100
+
+# The HTML_COLORSTYLE_GAMMA tag controls the gamma correction applied to the
+# luminance component of the colors in the HTML output. Values below 100
+# gradually make the output lighter, whereas values above 100 make the output
+# darker. The value divided by 100 is the actual gamma applied, so 80 represents
+# a gamma of 0.8, The value 220 represents a gamma of 2.2, and 100 does not
+# change the gamma.
+# Minimum value: 40, maximum value: 240, default value: 80.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_GAMMA = 80
+
+# If the HTML_TIMESTAMP tag is set to YES then the footer of each generated HTML
+# page will contain the date and time when the page was generated. Setting this
+# to YES can help to show when doxygen was last run and thus if the
+# documentation is up to date.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_TIMESTAMP = NO
+
+# If the HTML_DYNAMIC_SECTIONS tag is set to YES then the generated HTML
+# documentation will contain sections that can be hidden and shown after the
+# page has loaded.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_DYNAMIC_SECTIONS = NO
+
+# With HTML_INDEX_NUM_ENTRIES one can control the preferred number of entries
+# shown in the various tree structured indices initially; the user can expand
+# and collapse entries dynamically later on. Doxygen will expand the tree to
+# such a level that at most the specified number of entries are visible (unless
+# a fully collapsed tree already exceeds this amount). So setting the number of
+# entries 1 will produce a full collapsed tree by default. 0 is a special value
+# representing an infinite number of entries and will result in a full expanded
+# tree by default.
+# Minimum value: 0, maximum value: 9999, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_INDEX_NUM_ENTRIES = 100
+
+# If the GENERATE_DOCSET tag is set to YES, additional index files will be
+# generated that can be used as input for Apple's Xcode 3 integrated development
+# environment (see: http://developer.apple.com/tools/xcode/), introduced with
+# OSX 10.5 (Leopard). To create a documentation set, doxygen will generate a
+# Makefile in the HTML output directory. Running make will produce the docset in
+# that directory and running make install will install the docset in
+# ~/Library/Developer/Shared/Documentation/DocSets so that Xcode will find it at
+# startup. See http://developer.apple.com/tools/creatingdocsetswithdoxygen.html
+# for more information.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_DOCSET = NO
+
+# This tag determines the name of the docset feed. A documentation feed provides
+# an umbrella under which multiple documentation sets from a single provider
+# (such as a company or product suite) can be grouped.
+# The default value is: Doxygen generated docs.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_FEEDNAME = "Doxygen generated docs"
+
+# This tag specifies a string that should uniquely identify the documentation
+# set bundle. This should be a reverse domain-name style string, e.g.
+# com.mycompany.MyDocSet. Doxygen will append .docset to the name.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_BUNDLE_ID = org.doxygen.Project
+
+# The DOCSET_PUBLISHER_ID tag specifies a string that should uniquely identify
+# the documentation publisher. This should be a reverse domain-name style
+# string, e.g. com.mycompany.MyDocSet.documentation.
+# The default value is: org.doxygen.Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_ID = org.doxygen.Publisher
+
+# The DOCSET_PUBLISHER_NAME tag identifies the documentation publisher.
+# The default value is: Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_NAME = Publisher
+
+# If the GENERATE_HTMLHELP tag is set to YES then doxygen generates three
+# additional HTML index files: index.hhp, index.hhc, and index.hhk. The
+# index.hhp is a project file that can be read by Microsoft's HTML Help Workshop
+# (see: http://www.microsoft.com/en-us/download/details.aspx?id=21138) on
+# Windows.
+#
+# The HTML Help Workshop contains a compiler that can convert all HTML output
+# generated by doxygen into a single compiled HTML file (.chm). Compiled HTML
+# files are now used as the Windows 98 help format, and will replace the old
+# Windows help format (.hlp) on all Windows platforms in the future. Compressed
+# HTML files also contain an index, a table of contents, and you can search for
+# words in the documentation. The HTML workshop also contains a viewer for
+# compressed HTML files.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_HTMLHELP = NO
+
+# The CHM_FILE tag can be used to specify the file name of the resulting .chm
+# file. You can add a path in front of the file if the result should not be
+# written to the html output directory.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_FILE =
+
+# The HHC_LOCATION tag can be used to specify the location (absolute path
+# including file name) of the HTML help compiler (hhc.exe). If non-empty,
+# doxygen will try to run the HTML help compiler on the generated index.hhp.
+# The file has to be specified with full path.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+HHC_LOCATION =
+
+# The GENERATE_CHI flag controls if a separate .chi index file is generated
+# (YES) or that it should be included in the master .chm file (NO).
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+GENERATE_CHI = NO
+
+# The CHM_INDEX_ENCODING is used to encode HtmlHelp index (hhk), content (hhc)
+# and project file content.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_INDEX_ENCODING =
+
+# The BINARY_TOC flag controls whether a binary table of contents is generated
+# (YES) or a normal table of contents (NO) in the .chm file. Furthermore it
+# enables the Previous and Next buttons.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+BINARY_TOC = NO
+
+# The TOC_EXPAND flag can be set to YES to add extra items for group members to
+# the table of contents of the HTML help documentation and to the tree view.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+TOC_EXPAND = NO
+
+# If the GENERATE_QHP tag is set to YES and both QHP_NAMESPACE and
+# QHP_VIRTUAL_FOLDER are set, an additional index file will be generated that
+# can be used as input for Qt's qhelpgenerator to generate a Qt Compressed Help
+# (.qch) of the generated HTML documentation.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_QHP = NO
+
+# If the QHG_LOCATION tag is specified, the QCH_FILE tag can be used to specify
+# the file name of the resulting .qch file. The path specified is relative to
+# the HTML output folder.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QCH_FILE =
+
+# The QHP_NAMESPACE tag specifies the namespace to use when generating Qt Help
+# Project output. For more information please see Qt Help Project / Namespace
+# (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#namespace).
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_NAMESPACE = org.doxygen.Project
+
+# The QHP_VIRTUAL_FOLDER tag specifies the namespace to use when generating Qt
+# Help Project output. For more information please see Qt Help Project / Virtual
+# Folders (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#virtual-
+# folders).
+# The default value is: doc.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_VIRTUAL_FOLDER = doc
+
+# If the QHP_CUST_FILTER_NAME tag is set, it specifies the name of a custom
+# filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_NAME =
+
+# The QHP_CUST_FILTER_ATTRS tag specifies the list of the attributes of the
+# custom filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_ATTRS =
+
+# The QHP_SECT_FILTER_ATTRS tag specifies the list of the attributes this
+# project's filter section matches. Qt Help Project / Filter Attributes (see:
+# http://qt-project.org/doc/qt-4.8/qthelpproject.html#filter-attributes).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_SECT_FILTER_ATTRS =
+
+# The QHG_LOCATION tag can be used to specify the location of Qt's
+# qhelpgenerator. If non-empty doxygen will try to run qhelpgenerator on the
+# generated .qhp file.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHG_LOCATION =
+
+# If the GENERATE_ECLIPSEHELP tag is set to YES, additional index files will be
+# generated, together with the HTML files, they form an Eclipse help plugin. To
+# install this plugin and make it available under the help contents menu in
+# Eclipse, the contents of the directory containing the HTML and XML files needs
+# to be copied into the plugins directory of eclipse. The name of the directory
+# within the plugins directory should be the same as the ECLIPSE_DOC_ID value.
+# After copying Eclipse needs to be restarted before the help appears.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_ECLIPSEHELP = NO
+
+# A unique identifier for the Eclipse help plugin. When installing the plugin
+# the directory name containing the HTML and XML files should also have this
+# name. Each documentation set should have its own identifier.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_ECLIPSEHELP is set to YES.
+
+ECLIPSE_DOC_ID = org.doxygen.Project
+
+# If you want full control over the layout of the generated HTML pages it might
+# be necessary to disable the index and replace it with your own. The
+# DISABLE_INDEX tag can be used to turn on/off the condensed index (tabs) at top
+# of each HTML page. A value of NO enables the index and the value YES disables
+# it. Since the tabs in the index contain the same information as the navigation
+# tree, you can set this option to YES if you also set GENERATE_TREEVIEW to YES.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+DISABLE_INDEX = NO
+
+# The GENERATE_TREEVIEW tag is used to specify whether a tree-like index
+# structure should be generated to display hierarchical information. If the tag
+# value is set to YES, a side panel will be generated containing a tree-like
+# index structure (just like the one that is generated for HTML Help). For this
+# to work a browser that supports JavaScript, DHTML, CSS and frames is required
+# (i.e. any modern browser). Windows users are probably better off using the
+# HTML help feature. Via custom style sheets (see HTML_EXTRA_STYLESHEET) one can
+# further fine-tune the look of the index. As an example, the default style
+# sheet generated by doxygen has an example that shows how to put an image at
+# the root of the tree instead of the PROJECT_NAME. Since the tree basically has
+# the same information as the tab index, you could consider setting
+# DISABLE_INDEX to YES when enabling this option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_TREEVIEW = NO
+
+# The ENUM_VALUES_PER_LINE tag can be used to set the number of enum values that
+# doxygen will group on one line in the generated HTML documentation.
+#
+# Note that a value of 0 will completely suppress the enum values from appearing
+# in the overview section.
+# Minimum value: 0, maximum value: 20, default value: 4.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+ENUM_VALUES_PER_LINE = 4
+
+# If the treeview is enabled (see GENERATE_TREEVIEW) then this tag can be used
+# to set the initial width (in pixels) of the frame in which the tree is shown.
+# Minimum value: 0, maximum value: 1500, default value: 250.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+TREEVIEW_WIDTH = 250
+
+# If the EXT_LINKS_IN_WINDOW option is set to YES, doxygen will open links to
+# external symbols imported via tag files in a separate window.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+EXT_LINKS_IN_WINDOW = NO
+
+# Use this tag to change the font size of LaTeX formulas included as images in
+# the HTML documentation. When you change the font size after a successful
+# doxygen run you need to manually remove any form_*.png images from the HTML
+# output directory to force them to be regenerated.
+# Minimum value: 8, maximum value: 50, default value: 10.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_FONTSIZE = 10
+
+# Use the FORMULA_TRANPARENT tag to determine whether or not the images
+# generated for formulas are transparent PNGs. Transparent PNGs are not
+# supported properly for IE 6.0, but are supported on all modern browsers.
+#
+# Note that when changing this option you need to delete any form_*.png files in
+# the HTML output directory before the changes have effect.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_TRANSPARENT = YES
+
+# Enable the USE_MATHJAX option to render LaTeX formulas using MathJax (see
+# http://www.mathjax.org) which uses client side Javascript for the rendering
+# instead of using pre-rendered bitmaps. Use this if you do not have LaTeX
+# installed or if you want to formulas look prettier in the HTML output. When
+# enabled you may also need to install MathJax separately and configure the path
+# to it using the MATHJAX_RELPATH option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+USE_MATHJAX = NO
+
+# When MathJax is enabled you can set the default output format to be used for
+# the MathJax output. See the MathJax site (see:
+# http://docs.mathjax.org/en/latest/output.html) for more details.
+# Possible values are: HTML-CSS (which is slower, but has the best
+# compatibility), NativeMML (i.e. MathML) and SVG.
+# The default value is: HTML-CSS.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_FORMAT = HTML-CSS
+
+# When MathJax is enabled you need to specify the location relative to the HTML
+# output directory using the MATHJAX_RELPATH option. The destination directory
+# should contain the MathJax.js script. For instance, if the mathjax directory
+# is located at the same level as the HTML output directory, then
+# MATHJAX_RELPATH should be ../mathjax. The default value points to the MathJax
+# Content Delivery Network so you can quickly see the result without installing
+# MathJax. However, it is strongly recommended to install a local copy of
+# MathJax from http://www.mathjax.org before deployment.
+# The default value is: http://cdn.mathjax.org/mathjax/latest.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_RELPATH = http://cdn.mathjax.org/mathjax/latest
+
+# The MATHJAX_EXTENSIONS tag can be used to specify one or more MathJax
+# extension names that should be enabled during MathJax rendering. For example
+# MATHJAX_EXTENSIONS = TeX/AMSmath TeX/AMSsymbols
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_EXTENSIONS =
+
+# The MATHJAX_CODEFILE tag can be used to specify a file with javascript pieces
+# of code that will be used on startup of the MathJax code. See the MathJax site
+# (see: http://docs.mathjax.org/en/latest/output.html) for more details. For an
+# example see the documentation.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_CODEFILE =
+
+# When the SEARCHENGINE tag is enabled doxygen will generate a search box for
+# the HTML output. The underlying search engine uses javascript and DHTML and
+# should work on any modern browser. Note that when using HTML help
+# (GENERATE_HTMLHELP), Qt help (GENERATE_QHP), or docsets (GENERATE_DOCSET)
+# there is already a search function so this one should typically be disabled.
+# For large projects the javascript based search engine can be slow, then
+# enabling SERVER_BASED_SEARCH may provide a better solution. It is possible to
+# search using the keyboard; to jump to the search box use <access key> + S
+# (what the <access key> is depends on the OS and browser, but it is typically
+# <CTRL>, <ALT>/<option>, or both). Inside the search box use the <cursor down
+# key> to jump into the search results window, the results can be navigated
+# using the <cursor keys>. Press <Enter> to select an item or <escape> to cancel
+# the search. The filter options can be selected when the cursor is inside the
+# search box by pressing <Shift>+<cursor down>. Also here use the <cursor keys>
+# to select a filter and <Enter> or <escape> to activate or cancel the filter
+# option.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+SEARCHENGINE = YES
+
+# When the SERVER_BASED_SEARCH tag is enabled the search engine will be
+# implemented using a web server instead of a web client using Javascript. There
+# are two flavors of web server based searching depending on the EXTERNAL_SEARCH
+# setting. When disabled, doxygen will generate a PHP script for searching and
+# an index file used by the script. When EXTERNAL_SEARCH is enabled the indexing
+# and searching needs to be provided by external tools. See the section
+# "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SERVER_BASED_SEARCH = NO
+
+# When EXTERNAL_SEARCH tag is enabled doxygen will no longer generate the PHP
+# script for searching. Instead the search results are written to an XML file
+# which needs to be processed by an external indexer. Doxygen will invoke an
+# external search engine pointed to by the SEARCHENGINE_URL option to obtain the
+# search results.
+#
+# Doxygen ships with an example indexer (doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/).
+#
+# See the section "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH = NO
+
+# The SEARCHENGINE_URL should point to a search engine hosted by a web server
+# which will return the search results when EXTERNAL_SEARCH is enabled.
+#
+# Doxygen ships with an example indexer (doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/). See the section "External Indexing and
+# Searching" for details.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHENGINE_URL =
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the unindexed
+# search data is written to a file for indexing by an external tool. With the
+# SEARCHDATA_FILE tag the name of this file can be specified.
+# The default file is: searchdata.xml.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHDATA_FILE = searchdata.xml
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the
+# EXTERNAL_SEARCH_ID tag can be used as an identifier for the project. This is
+# useful in combination with EXTRA_SEARCH_MAPPINGS to search through multiple
+# projects and redirect the results back to the right project.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH_ID =
+
+# The EXTRA_SEARCH_MAPPINGS tag can be used to enable searching through doxygen
+# projects other than the one defined by this configuration file, but that are
+# all added to the same external search index. Each project needs to have a
+# unique id set via EXTERNAL_SEARCH_ID. The search mapping then maps the id of
+# to a relative location where the documentation can be found. The format is:
+# EXTRA_SEARCH_MAPPINGS = tagname1=loc1 tagname2=loc2 ...
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTRA_SEARCH_MAPPINGS =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the LaTeX output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_LATEX tag is set to YES, doxygen will generate LaTeX output.
+# The default value is: YES.
+
+GENERATE_LATEX = NO
+
+# The LATEX_OUTPUT tag is used to specify where the LaTeX docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_OUTPUT = latex
+
+# The LATEX_CMD_NAME tag can be used to specify the LaTeX command name to be
+# invoked.
+#
+# Note that when enabling USE_PDFLATEX this option is only used for generating
+# bitmaps for formulas in the HTML output, but not in the Makefile that is
+# written to the output directory.
+# The default file is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_CMD_NAME = latex
+
+# The MAKEINDEX_CMD_NAME tag can be used to specify the command name to generate
+# index for LaTeX.
+# The default file is: makeindex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+MAKEINDEX_CMD_NAME = makeindex
+
+# If the COMPACT_LATEX tag is set to YES, doxygen generates more compact LaTeX
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+COMPACT_LATEX = NO
+
+# The PAPER_TYPE tag can be used to set the paper type that is used by the
+# printer.
+# Possible values are: a4 (210 x 297 mm), letter (8.5 x 11 inches), legal (8.5 x
+# 14 inches) and executive (7.25 x 10.5 inches).
+# The default value is: a4.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PAPER_TYPE = a4
+
+# The EXTRA_PACKAGES tag can be used to specify one or more LaTeX package names
+# that should be included in the LaTeX output. The package can be specified just
+# by its name or with the correct syntax as to be used with the LaTeX
+# \usepackage command. To get the times font for instance you can specify :
+# EXTRA_PACKAGES=times or EXTRA_PACKAGES={times}
+# To use the option intlimits with the amsmath package you can specify:
+# EXTRA_PACKAGES=[intlimits]{amsmath}
+# If left blank no extra packages will be included.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+EXTRA_PACKAGES =
+
+# The LATEX_HEADER tag can be used to specify a personal LaTeX header for the
+# generated LaTeX document. The header should contain everything until the first
+# chapter. If it is left blank doxygen will generate a standard header. See
+# section "Doxygen usage" for information on how to let doxygen write the
+# default header to a separate file.
+#
+# Note: Only use a user-defined header if you know what you are doing! The
+# following commands have a special meaning inside the header: $title,
+# $datetime, $date, $doxygenversion, $projectname, $projectnumber,
+# $projectbrief, $projectlogo. Doxygen will replace $title with the empty
+# string, for the replacement values of the other commands the user is referred
+# to HTML_HEADER.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HEADER =
+
+# The LATEX_FOOTER tag can be used to specify a personal LaTeX footer for the
+# generated LaTeX document. The footer should contain everything after the last
+# chapter. If it is left blank doxygen will generate a standard footer. See
+# LATEX_HEADER for more information on how to generate a default footer and what
+# special commands can be used inside the footer.
+#
+# Note: Only use a user-defined footer if you know what you are doing!
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_FOOTER =
+
+# The LATEX_EXTRA_STYLESHEET tag can be used to specify additional user-defined
+# LaTeX style sheets that are included after the standard style sheets created
+# by doxygen. Using this option one can overrule certain style aspects. Doxygen
+# will copy the style sheet files to the output directory.
+# Note: The order of the extra style sheet files is of importance (e.g. the last
+# style sheet in the list overrules the setting of the previous ones in the
+# list).
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_EXTRA_STYLESHEET =
+
+# The LATEX_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the LATEX_OUTPUT output
+# directory. Note that the files will be copied as-is; there are no commands or
+# markers available.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_EXTRA_FILES =
+
+# If the PDF_HYPERLINKS tag is set to YES, the LaTeX that is generated is
+# prepared for conversion to PDF (using ps2pdf or pdflatex). The PDF file will
+# contain links (just like the HTML output) instead of page references. This
+# makes the output suitable for online browsing using a PDF viewer.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PDF_HYPERLINKS = YES
+
+# If the USE_PDFLATEX tag is set to YES, doxygen will use pdflatex to generate
+# the PDF file directly from the LaTeX files. Set this option to YES, to get a
+# higher quality PDF documentation.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+USE_PDFLATEX = YES
+
+# If the LATEX_BATCHMODE tag is set to YES, doxygen will add the \batchmode
+# command to the generated LaTeX files. This will instruct LaTeX to keep running
+# if errors occur, instead of asking the user for help. This option is also used
+# when generating formulas in HTML.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BATCHMODE = NO
+
+# If the LATEX_HIDE_INDICES tag is set to YES then doxygen will not include the
+# index chapters (such as File Index, Compound Index, etc.) in the output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HIDE_INDICES = NO
+
+# If the LATEX_SOURCE_CODE tag is set to YES then doxygen will include source
+# code with syntax highlighting in the LaTeX output.
+#
+# Note that which sources are shown also depends on other settings such as
+# SOURCE_BROWSER.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_SOURCE_CODE = NO
+
+# The LATEX_BIB_STYLE tag can be used to specify the style to use for the
+# bibliography, e.g. plainnat, or ieeetr. See
+# http://en.wikipedia.org/wiki/BibTeX and \cite for more info.
+# The default value is: plain.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BIB_STYLE = plain
+
+# If the LATEX_TIMESTAMP tag is set to YES then the footer of each generated
+# page will contain the date and time when the page was generated. Setting this
+# to NO can help when comparing the output of multiple runs.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_TIMESTAMP = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the RTF output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_RTF tag is set to YES, doxygen will generate RTF output. The
+# RTF output is optimized for Word 97 and may not look too pretty with other RTF
+# readers/editors.
+# The default value is: NO.
+
+GENERATE_RTF = NO
+
+# The RTF_OUTPUT tag is used to specify where the RTF docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: rtf.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_OUTPUT = rtf
+
+# If the COMPACT_RTF tag is set to YES, doxygen generates more compact RTF
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+COMPACT_RTF = NO
+
+# If the RTF_HYPERLINKS tag is set to YES, the RTF that is generated will
+# contain hyperlink fields. The RTF file will contain links (just like the HTML
+# output) instead of page references. This makes the output suitable for online
+# browsing using Word or some other Word compatible readers that support those
+# fields.
+#
+# Note: WordPad (write) and others do not support links.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_HYPERLINKS = NO
+
+# Load stylesheet definitions from file. Syntax is similar to doxygen's config
+# file, i.e. a series of assignments. You only have to provide replacements,
+# missing definitions are set to their default value.
+#
+# See also section "Doxygen usage" for information on how to generate the
+# default style sheet that doxygen normally uses.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_STYLESHEET_FILE =
+
+# Set optional variables used in the generation of an RTF document. Syntax is
+# similar to doxygen's config file. A template extensions file can be generated
+# using doxygen -e rtf extensionFile.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_EXTENSIONS_FILE =
+
+# If the RTF_SOURCE_CODE tag is set to YES then doxygen will include source code
+# with syntax highlighting in the RTF output.
+#
+# Note that which sources are shown also depends on other settings such as
+# SOURCE_BROWSER.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_SOURCE_CODE = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the man page output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_MAN tag is set to YES, doxygen will generate man pages for
+# classes and files.
+# The default value is: NO.
+
+GENERATE_MAN = NO
+
+# The MAN_OUTPUT tag is used to specify where the man pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it. A directory man3 will be created inside the directory specified by
+# MAN_OUTPUT.
+# The default directory is: man.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_OUTPUT = man
+
+# The MAN_EXTENSION tag determines the extension that is added to the generated
+# man pages. In case the manual section does not start with a number, the number
+# 3 is prepended. The dot (.) at the beginning of the MAN_EXTENSION tag is
+# optional.
+# The default value is: .3.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_EXTENSION = .3
+
+# The MAN_SUBDIR tag determines the name of the directory created within
+# MAN_OUTPUT in which the man pages are placed. If defaults to man followed by
+# MAN_EXTENSION with the initial . removed.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_SUBDIR =
+
+# If the MAN_LINKS tag is set to YES and doxygen generates man output, then it
+# will generate one additional man file for each entity documented in the real
+# man page(s). These additional files only source the real man page, but without
+# them the man command would be unable to find the correct page.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_LINKS = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the XML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_XML tag is set to YES, doxygen will generate an XML file that
+# captures the structure of the code including all documentation.
+# The default value is: NO.
+
+GENERATE_XML = YES
+
+# The XML_OUTPUT tag is used to specify where the XML pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: xml.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_OUTPUT = xml
+
+# If the XML_PROGRAMLISTING tag is set to YES, doxygen will dump the program
+# listings (including syntax highlighting and cross-referencing information) to
+# the XML output. Note that enabling this will significantly increase the size
+# of the XML output.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_PROGRAMLISTING = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to the DOCBOOK output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_DOCBOOK tag is set to YES, doxygen will generate Docbook files
+# that can be used to generate PDF.
+# The default value is: NO.
+
+GENERATE_DOCBOOK = NO
+
+# The DOCBOOK_OUTPUT tag is used to specify where the Docbook pages will be put.
+# If a relative path is entered the value of OUTPUT_DIRECTORY will be put in
+# front of it.
+# The default directory is: docbook.
+# This tag requires that the tag GENERATE_DOCBOOK is set to YES.
+
+DOCBOOK_OUTPUT = docbook
+
+# If the DOCBOOK_PROGRAMLISTING tag is set to YES, doxygen will include the
+# program listings (including syntax highlighting and cross-referencing
+# information) to the DOCBOOK output. Note that enabling this will significantly
+# increase the size of the DOCBOOK output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_DOCBOOK is set to YES.
+
+DOCBOOK_PROGRAMLISTING = NO
+
+#---------------------------------------------------------------------------
+# Configuration options for the AutoGen Definitions output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_AUTOGEN_DEF tag is set to YES, doxygen will generate an
+# AutoGen Definitions (see http://autogen.sf.net) file that captures the
+# structure of the code including all documentation. Note that this feature is
+# still experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_AUTOGEN_DEF = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the Perl module output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_PERLMOD tag is set to YES, doxygen will generate a Perl module
+# file that captures the structure of the code including all documentation.
+#
+# Note that this feature is still experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_PERLMOD = NO
+
+# If the PERLMOD_LATEX tag is set to YES, doxygen will generate the necessary
+# Makefile rules, Perl scripts and LaTeX code to be able to generate PDF and DVI
+# output from the Perl module output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_LATEX = NO
+
+# If the PERLMOD_PRETTY tag is set to YES, the Perl module output will be nicely
+# formatted so it can be parsed by a human reader. This is useful if you want to
+# understand what is going on. On the other hand, if this tag is set to NO, the
+# size of the Perl module output will be much smaller and Perl will parse it
+# just the same.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_PRETTY = YES
+
+# The names of the make variables in the generated doxyrules.make file are
+# prefixed with the string contained in PERLMOD_MAKEVAR_PREFIX. This is useful
+# so different doxyrules.make files included by the same Makefile don't
+# overwrite each other's variables.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_MAKEVAR_PREFIX =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the preprocessor
+#---------------------------------------------------------------------------
+
+# If the ENABLE_PREPROCESSING tag is set to YES, doxygen will evaluate all
+# C-preprocessor directives found in the sources and include files.
+# The default value is: YES.
+
+ENABLE_PREPROCESSING = YES
+
+# If the MACRO_EXPANSION tag is set to YES, doxygen will expand all macro names
+# in the source code. If set to NO, only conditional compilation will be
+# performed. Macro expansion can be done in a controlled way by setting
+# EXPAND_ONLY_PREDEF to YES.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+MACRO_EXPANSION = NO
+
+# If the EXPAND_ONLY_PREDEF and MACRO_EXPANSION tags are both set to YES then
+# the macro expansion is limited to the macros specified with the PREDEFINED and
+# EXPAND_AS_DEFINED tags.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_ONLY_PREDEF = NO
+
+# If the SEARCH_INCLUDES tag is set to YES, the include files in the
+# INCLUDE_PATH will be searched if a #include is found.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SEARCH_INCLUDES = YES
+
+# The INCLUDE_PATH tag can be used to specify one or more directories that
+# contain include files that are not input files but should be processed by the
+# preprocessor.
+# This tag requires that the tag SEARCH_INCLUDES is set to YES.
+
+INCLUDE_PATH =
+
+# You can use the INCLUDE_FILE_PATTERNS tag to specify one or more wildcard
+# patterns (like *.h and *.hpp) to filter out the header-files in the
+# directories. If left blank, the patterns specified with FILE_PATTERNS will be
+# used.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+INCLUDE_FILE_PATTERNS =
+
+# The PREDEFINED tag can be used to specify one or more macro names that are
+# defined before the preprocessor is started (similar to the -D option of e.g.
+# gcc). The argument of the tag is a list of macros of the form: name or
+# name=definition (no spaces). If the definition and the "=" are omitted, "=1"
+# is assumed. To prevent a macro definition from being undefined via #undef or
+# recursively expanded use the := operator instead of the = operator.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+PREDEFINED =
+
+# If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then this
+# tag can be used to specify a list of macro names that should be expanded. The
+# macro definition that is found in the sources will be used. Use the PREDEFINED
+# tag if you want to use a different macro definition that overrules the
+# definition found in the source code.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_AS_DEFINED =
+
+# If the SKIP_FUNCTION_MACROS tag is set to YES then doxygen's preprocessor will
+# remove all references to function-like macros that are alone on a line, have
+# an all uppercase name, and do not end with a semicolon. Such function macros
+# are typically used for boiler-plate code, and will confuse the parser if not
+# removed.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SKIP_FUNCTION_MACROS = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to external references
+#---------------------------------------------------------------------------
+
+# The TAGFILES tag can be used to specify one or more tag files. For each tag
+# file the location of the external documentation should be added. The format of
+# a tag file without this location is as follows:
+# TAGFILES = file1 file2 ...
+# Adding location for the tag files is done as follows:
+# TAGFILES = file1=loc1 "file2 = loc2" ...
+# where loc1 and loc2 can be relative or absolute paths or URLs. See the
+# section "Linking to external documentation" for more information about the use
+# of tag files.
+# Note: Each tag file must have a unique name (where the name does NOT include
+# the path). If a tag file is not located in the directory in which doxygen is
+# run, you must also specify the path to the tagfile here.
+
+TAGFILES =
+
+# When a file name is specified after GENERATE_TAGFILE, doxygen will create a
+# tag file that is based on the input files it reads. See section "Linking to
+# external documentation" for more information about the usage of tag files.
+
+GENERATE_TAGFILE =
+
+# If the ALLEXTERNALS tag is set to YES, all external class will be listed in
+# the class index. If set to NO, only the inherited external classes will be
+# listed.
+# The default value is: NO.
+
+ALLEXTERNALS = NO
+
+# If the EXTERNAL_GROUPS tag is set to YES, all external groups will be listed
+# in the modules index. If set to NO, only the current project's groups will be
+# listed.
+# The default value is: YES.
+
+EXTERNAL_GROUPS = YES
+
+# If the EXTERNAL_PAGES tag is set to YES, all external pages will be listed in
+# the related pages index. If set to NO, only the current project's pages will
+# be listed.
+# The default value is: YES.
+
+EXTERNAL_PAGES = YES
+
+# The PERL_PATH should be the absolute path and name of the perl script
+# interpreter (i.e. the result of 'which perl').
+# The default file (with absolute path) is: /usr/bin/perl.
+
+PERL_PATH = /usr/bin/perl
+
+#---------------------------------------------------------------------------
+# Configuration options related to the dot tool
+#---------------------------------------------------------------------------
+
+# If the CLASS_DIAGRAMS tag is set to YES, doxygen will generate a class diagram
+# (in HTML and LaTeX) for classes with base or super classes. Setting the tag to
+# NO turns the diagrams off. Note that this option also works with HAVE_DOT
+# disabled, but it is recommended to install and use dot, since it yields more
+# powerful graphs.
+# The default value is: YES.
+
+CLASS_DIAGRAMS = NO
+
+# You can define message sequence charts within doxygen comments using the \msc
+# command. Doxygen will then run the mscgen tool (see:
+# http://www.mcternan.me.uk/mscgen/)) to produce the chart and insert it in the
+# documentation. The MSCGEN_PATH tag allows you to specify the directory where
+# the mscgen tool resides. If left empty the tool is assumed to be found in the
+# default search path.
+
+MSCGEN_PATH =
+
+# You can include diagrams made with dia in doxygen documentation. Doxygen will
+# then run dia to produce the diagram and insert it in the documentation. The
+# DIA_PATH tag allows you to specify the directory where the dia binary resides.
+# If left empty dia is assumed to be found in the default search path.
+
+DIA_PATH =
+
+# If set to YES the inheritance and collaboration graphs will hide inheritance
+# and usage relations if the target is undocumented or is not a class.
+# The default value is: YES.
+
+HIDE_UNDOC_RELATIONS = YES
+
+# If you set the HAVE_DOT tag to YES then doxygen will assume the dot tool is
+# available from the path. This tool is part of Graphviz (see:
+# http://www.graphviz.org/), a graph visualization toolkit from AT&T and Lucent
+# Bell Labs. The other options in this section have no effect if this option is
+# set to NO
+# The default value is: YES.
+
+HAVE_DOT = NO
+
+# The DOT_NUM_THREADS specifies the number of dot invocations doxygen is allowed
+# to run in parallel. When set to 0 doxygen will base this on the number of
+# processors available in the system. You can set it explicitly to a value
+# larger than 0 to get control over the balance between CPU load and processing
+# speed.
+# Minimum value: 0, maximum value: 32, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_NUM_THREADS = 0
+
+# When you want a differently looking font in the dot files that doxygen
+# generates you can specify the font name using DOT_FONTNAME. You need to make
+# sure dot is able to find the font, which can be done by putting it in a
+# standard location or by setting the DOTFONTPATH environment variable or by
+# setting DOT_FONTPATH to the directory containing the font.
+# The default value is: Helvetica.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTNAME = Helvetica
+
+# The DOT_FONTSIZE tag can be used to set the size (in points) of the font of
+# dot graphs.
+# Minimum value: 4, maximum value: 24, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTSIZE = 10
+
+# By default doxygen will tell dot to use the default font as specified with
+# DOT_FONTNAME. If you specify a different font using DOT_FONTNAME you can set
+# the path where dot can find it using this tag.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTPATH =
+
+# If the CLASS_GRAPH tag is set to YES then doxygen will generate a graph for
+# each documented class showing the direct and indirect inheritance relations.
+# Setting this tag to YES will force the CLASS_DIAGRAMS tag to NO.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CLASS_GRAPH = YES
+
+# If the COLLABORATION_GRAPH tag is set to YES then doxygen will generate a
+# graph for each documented class showing the direct and indirect implementation
+# dependencies (inheritance, containment, and class references variables) of the
+# class with other documented classes.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+COLLABORATION_GRAPH = YES
+
+# If the GROUP_GRAPHS tag is set to YES then doxygen will generate a graph for
+# groups, showing the direct groups dependencies.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GROUP_GRAPHS = YES
+
+# If the UML_LOOK tag is set to YES, doxygen will generate inheritance and
+# collaboration diagrams in a style similar to the OMG's Unified Modeling
+# Language.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LOOK = NO
+
+# If the UML_LOOK tag is enabled, the fields and methods are shown inside the
+# class node. If there are many fields or methods and many nodes the graph may
+# become too big to be useful. The UML_LIMIT_NUM_FIELDS threshold limits the
+# number of items for each type to make the size more manageable. Set this to 0
+# for no limit. Note that the threshold may be exceeded by 50% before the limit
+# is enforced. So when you set the threshold to 10, up to 15 fields may appear,
+# but if the number exceeds 15, the total amount of fields shown is limited to
+# 10.
+# Minimum value: 0, maximum value: 100, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LIMIT_NUM_FIELDS = 10
+
+# If the TEMPLATE_RELATIONS tag is set to YES then the inheritance and
+# collaboration graphs will show the relations between templates and their
+# instances.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+TEMPLATE_RELATIONS = NO
+
+# If the INCLUDE_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are set to
+# YES then doxygen will generate a graph for each documented file showing the
+# direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDE_GRAPH = YES
+
+# If the INCLUDED_BY_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are
+# set to YES then doxygen will generate a graph for each documented file showing
+# the direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDED_BY_GRAPH = YES
+
+# If the CALL_GRAPH tag is set to YES then doxygen will generate a call
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable call graphs for selected
+# functions only using the \callgraph command. Disabling a call graph can be
+# accomplished by means of the command \hidecallgraph.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALL_GRAPH = NO
+
+# If the CALLER_GRAPH tag is set to YES then doxygen will generate a caller
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable caller graphs for selected
+# functions only using the \callergraph command. Disabling a caller graph can be
+# accomplished by means of the command \hidecallergraph.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALLER_GRAPH = NO
+
+# If the GRAPHICAL_HIERARCHY tag is set to YES then doxygen will graphical
+# hierarchy of all classes instead of a textual one.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GRAPHICAL_HIERARCHY = YES
+
+# If the DIRECTORY_GRAPH tag is set to YES then doxygen will show the
+# dependencies a directory has on other directories in a graphical way. The
+# dependency relations are determined by the #include relations between the
+# files in the directories.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DIRECTORY_GRAPH = YES
+
+# The DOT_IMAGE_FORMAT tag can be used to set the image format of the images
+# generated by dot. For an explanation of the image formats see the section
+# output formats in the documentation of the dot tool (Graphviz (see:
+# http://www.graphviz.org/)).
+# Note: If you choose svg you need to set HTML_FILE_EXTENSION to xhtml in order
+# to make the SVG files visible in IE 9+ (other browsers do not have this
+# requirement).
+# Possible values are: png, png:cairo, png:cairo:cairo, png:cairo:gd, png:gd,
+# png:gd:gd, jpg, jpg:cairo, jpg:cairo:gd, jpg:gd, jpg:gd:gd, gif, gif:cairo,
+# gif:cairo:gd, gif:gd, gif:gd:gd, svg, png:gd, png:gd:gd, png:cairo,
+# png:cairo:gd, png:cairo:cairo, png:cairo:gdiplus, png:gdiplus and
+# png:gdiplus:gdiplus.
+# The default value is: png.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_IMAGE_FORMAT = png
+
+# If DOT_IMAGE_FORMAT is set to svg, then this option can be set to YES to
+# enable generation of interactive SVG images that allow zooming and panning.
+#
+# Note that this requires a modern browser other than Internet Explorer. Tested
+# and working are Firefox, Chrome, Safari, and Opera.
+# Note: For IE 9+ you need to set HTML_FILE_EXTENSION to xhtml in order to make
+# the SVG files visible. Older versions of IE do not have SVG support.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INTERACTIVE_SVG = NO
+
+# The DOT_PATH tag can be used to specify the path where the dot tool can be
+# found. If left blank, it is assumed the dot tool can be found in the path.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_PATH =
+
+# The DOTFILE_DIRS tag can be used to specify one or more directories that
+# contain dot files that are included in the documentation (see the \dotfile
+# command).
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOTFILE_DIRS =
+
+# The MSCFILE_DIRS tag can be used to specify one or more directories that
+# contain msc files that are included in the documentation (see the \mscfile
+# command).
+
+MSCFILE_DIRS =
+
+# The DIAFILE_DIRS tag can be used to specify one or more directories that
+# contain dia files that are included in the documentation (see the \diafile
+# command).
+
+DIAFILE_DIRS =
+
+# When using plantuml, the PLANTUML_JAR_PATH tag should be used to specify the
+# path where java can find the plantuml.jar file. If left blank, it is assumed
+# PlantUML is not used or called during a preprocessing step. Doxygen will
+# generate a warning when it encounters a \startuml command in this case and
+# will not generate output for the diagram.
+
+PLANTUML_JAR_PATH =
+
+# When using plantuml, the PLANTUML_CFG_FILE tag can be used to specify a
+# configuration file for plantuml.
+
+PLANTUML_CFG_FILE =
+
+# When using plantuml, the specified paths are searched for files specified by
+# the !include statement in a plantuml block.
+
+PLANTUML_INCLUDE_PATH =
+
+# The DOT_GRAPH_MAX_NODES tag can be used to set the maximum number of nodes
+# that will be shown in the graph. If the number of nodes in a graph becomes
+# larger than this value, doxygen will truncate the graph, which is visualized
+# by representing a node as a red box. Note that doxygen if the number of direct
+# children of the root node in a graph is already larger than
+# DOT_GRAPH_MAX_NODES then the graph will not be shown at all. Also note that
+# the size of a graph can be further restricted by MAX_DOT_GRAPH_DEPTH.
+# Minimum value: 0, maximum value: 10000, default value: 50.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_GRAPH_MAX_NODES = 50
+
+# The MAX_DOT_GRAPH_DEPTH tag can be used to set the maximum depth of the graphs
+# generated by dot. A depth value of 3 means that only nodes reachable from the
+# root by following a path via at most 3 edges will be shown. Nodes that lay
+# further from the root node will be omitted. Note that setting this option to 1
+# or 2 may greatly reduce the computation time needed for large code bases. Also
+# note that the size of a graph can be further restricted by
+# DOT_GRAPH_MAX_NODES. Using a depth of 0 means no depth restriction.
+# Minimum value: 0, maximum value: 1000, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+MAX_DOT_GRAPH_DEPTH = 0
+
+# Set the DOT_TRANSPARENT tag to YES to generate images with a transparent
+# background. This is disabled by default, because dot on Windows does not seem
+# to support this out of the box.
+#
+# Warning: Depending on the platform used, enabling this option may lead to
+# badly anti-aliased labels on the edges of a graph (i.e. they become hard to
+# read).
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_TRANSPARENT = NO
+
+# Set the DOT_MULTI_TARGETS tag to YES to allow dot to generate multiple output
+# files in one run (i.e. multiple -o and -T options on the command line). This
+# makes dot run faster, but since only newer versions of dot (>1.8.10) support
+# this, this feature is disabled by default.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_MULTI_TARGETS = NO
+
+# If the GENERATE_LEGEND tag is set to YES doxygen will generate a legend page
+# explaining the meaning of the various boxes and arrows in the dot generated
+# graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GENERATE_LEGEND = YES
+
+# If the DOT_CLEANUP tag is set to YES, doxygen will remove the intermediate dot
+# files that are used to generate the various graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_CLEANUP = YES
diff --git a/finn-hlslib/docs/requirements.txt b/finn-hlslib/docs/requirements.txt
new file mode 100644
index 000000000..28bb7dedc
--- /dev/null
+++ b/finn-hlslib/docs/requirements.txt
@@ -0,0 +1,29 @@
+alabaster==0.7.12
+Babel==2.7.0
+breathe==4.13.0.post0
+certifi==2019.3.9
+chardet==3.0.4
+commonmark==0.9.0
+docutils==0.14
+future==0.17.1
+idna==2.8
+imagesize==1.1.0
+Jinja2==2.10.1
+MarkupSafe==1.1.1
+packaging==19.0
+Pygments==2.4.1
+pyparsing==2.4.0
+pytz==2019.1
+recommonmark==0.5.0
+requests==2.22.0
+six==1.12.0
+snowballstemmer==1.2.1
+Sphinx==2.0.1
+sphinx-rtd-theme==0.4.3
+sphinxcontrib-applehelp==1.0.1
+sphinxcontrib-devhelp==1.0.1
+sphinxcontrib-htmlhelp==1.0.2
+sphinxcontrib-jsmath==1.0.1
+sphinxcontrib-qthelp==1.0.2
+sphinxcontrib-serializinghtml==1.1.3
+urllib3==1.25.3
diff --git a/finn-hlslib/docs/source/conf.py b/finn-hlslib/docs/source/conf.py
new file mode 100755
index 000000000..c708b433e
--- /dev/null
+++ b/finn-hlslib/docs/source/conf.py
@@ -0,0 +1,93 @@
+# Copyright (c) 2019, Xilinx, Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice,
+# this list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in the
+# documentation and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+# OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+# ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Configuration file for the Sphinx documentation builder.
+#
+# This file only contains a selection of the most common options. For a full
+# list see the documentation:
+# http://www.sphinx-doc.org/en/master/config
+
+# -- Path setup --------------------------------------------------------------
+
+# If extensions (or modules to document with autodoc) are in another directory,
+# add these directories to sys.path here. If the directory is relative to the
+# documentation root, use os.path.abspath to make it absolute, like shown here.
+#
+
+import sys
+import datetime
+import subprocess, os
+
+read_the_docs_build = os.environ.get('READTHEDOCS', None) == 'True'
+
+if read_the_docs_build:
+
+ subprocess.call('cd ../doxygen; doxygen Doxyfile', shell=True)
+
+# -- Project information -----------------------------------------------------
+
+project = 'finn-hls'
+copyright = '%s, Xilinx' %str(datetime.datetime.now().year)
+author = 'Xilinx'
+
+# The full version, including alpha/beta/rc tags
+release = '0.1'
+
+
+breathe_projects = {"myproject": "../doxygen/xml/"}
+breathe_default_project = "myproject"
+
+# -- General configuration ---------------------------------------------------
+
+# Add any Sphinx extension module names here, as strings. They can be
+# extensions coming with Sphinx (named 'sphinx.ext.*') or your custom
+# ones.
+extensions = ['recommonmark','breathe']
+
+# Add any paths that contain templates here, relative to this directory.
+templates_path = ['_templates']
+
+# List of patterns, relative to source directory, that match files and
+# directories to ignore when looking for source files.
+# This pattern also affects html_static_path and html_extra_path.
+exclude_patterns = []
+
+autoclass_content = 'both'
+
+# -- Options for HTML output -------------------------------------------------
+
+# The theme to use for HTML and HTML Help pages. See the documentation for
+# a list of builtin themes.
+#
+html_theme = 'sphinx_rtd_theme'
+
+# Add any paths that contain custom static files (such as style sheets) here,
+# relative to this directory. They are copied after the builtin static files,
+# so a file named "default.css" will overwrite the builtin "default.css".
+html_static_path = ['_static']
diff --git a/finn-hlslib/docs/source/index.rst b/finn-hlslib/docs/source/index.rst
new file mode 100755
index 000000000..13a65da13
--- /dev/null
+++ b/finn-hlslib/docs/source/index.rst
@@ -0,0 +1,65 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+Introduction
+====================================
+FINN is an experimental framework from Xilinx Research Labs to explore deep neural network inference on FPGAs. It specifically targets quantized neural networks, with emphasis on generating dataflow-style architectures customized for each network. It is not intended to be a generic DNN accelerator like xDNN, but rather a tool for exploring the design space of DNN inference accelerators on FPGAs.
+
+====================================
+FINN-HLS content
+====================================
+
+The FINN-HLS repository contains the C++ description of multiple layers for the implementation of quantized neural networks using dataflow architecture.
+The library serves as a hardware backend for the FINN project, and it integrates in Xilinx Vivado HLS tool.
+
+
+.. toctree::
+ :maxdepth: 2
+
+
+
+ library/activations
+ library/weights
+ library/matrixvector
+ library/dma
+ library/maxpool
+ library/fclayer
+ library/convlayer
+ library/swg
+ library/streamtools
+ library/mac
+ library/mmv
+
+Indices and tables
+==================
+
+* :ref:`genindex`
+* :ref:`modindex`
+* :ref:`search`
diff --git a/finn-hlslib/docs/source/library/activations.rst b/finn-hlslib/docs/source/library/activations.rst
new file mode 100755
index 000000000..ce488532a
--- /dev/null
+++ b/finn-hlslib/docs/source/library/activations.rst
@@ -0,0 +1,51 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+activations.hpp
+====================================
+
+The activations.hpp file lists a set of classes used to implement the activation part of the mvau.
+
+
+.. doxygenclass:: Activation
+ :members:
+ :private-members:
+
+.. doxygenclass:: PassThroughActivation
+ :members:
+ :private-members:
+
+.. doxygenclass:: ThresholdActivation
+ :members:
+ :private-members:
+
+.. doxygenclass:: ThresholdsActivation
+ :members:
+ :private-members:
diff --git a/finn-hlslib/docs/source/library/convlayer.rst b/finn-hlslib/docs/source/library/convlayer.rst
new file mode 100755
index 000000000..a669cfd94
--- /dev/null
+++ b/finn-hlslib/docs/source/library/convlayer.rst
@@ -0,0 +1,42 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+convlayer.h
+====================================
+
+The convlayer.h file describes the HLS implementation of the convolutional layer.
+
+
+.. doxygenfile:: convlayer.h
+
+
+
+
+
diff --git a/finn-hlslib/docs/source/library/dma.rst b/finn-hlslib/docs/source/library/dma.rst
new file mode 100755
index 000000000..222c0895d
--- /dev/null
+++ b/finn-hlslib/docs/source/library/dma.rst
@@ -0,0 +1,41 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+dma.h
+====================================
+
+The dma.h file describes the HLS blocks used to communicate between programmable logic and host arm processor in Zynq and Zynq Ultrascale+ devices.
+
+.. doxygenfile:: dma.h
+
+
+
+
+
diff --git a/finn-hlslib/docs/source/library/fclayer.rst b/finn-hlslib/docs/source/library/fclayer.rst
new file mode 100755
index 000000000..92a874fa8
--- /dev/null
+++ b/finn-hlslib/docs/source/library/fclayer.rst
@@ -0,0 +1,42 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+fclayer.h
+====================================
+
+The fclayer.h file describes the HLS implementation of the fully connected layer.
+
+
+.. doxygenfile:: fclayer.h
+
+
+
+
+
diff --git a/finn-hlslib/docs/source/library/mac.rst b/finn-hlslib/docs/source/library/mac.rst
new file mode 100755
index 000000000..3f31d1a81
--- /dev/null
+++ b/finn-hlslib/docs/source/library/mac.rst
@@ -0,0 +1,38 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+mac.hpp
+====================================
+
+The mac.hpp file collects the definition of the multiply operation used in the matrix-vector block, as well as the used multiplication
+
+
+.. doxygenfile:: mac.hpp
+
diff --git a/finn-hlslib/docs/source/library/matrixvector.rst b/finn-hlslib/docs/source/library/matrixvector.rst
new file mode 100755
index 000000000..913ad0f87
--- /dev/null
+++ b/finn-hlslib/docs/source/library/matrixvector.rst
@@ -0,0 +1,44 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+mvau.hpp
+====================================
+
+The mvau.hpp file describes the core compute block of the library, namely the matrix vector .
+
+
+.. doxygenfile:: mvau.hpp
+
+
+
+
+
+
+
diff --git a/finn-hlslib/docs/source/library/maxpool.rst b/finn-hlslib/docs/source/library/maxpool.rst
new file mode 100755
index 000000000..d22ed2057
--- /dev/null
+++ b/finn-hlslib/docs/source/library/maxpool.rst
@@ -0,0 +1,38 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+maxpool.h
+====================================
+
+The maxpool.h file describes the HLS implementation of the maxpool layer.
+
+
+.. doxygenfile:: maxpool.h
+
diff --git a/finn-hlslib/docs/source/library/mmv.rst b/finn-hlslib/docs/source/library/mmv.rst
new file mode 100755
index 000000000..df33ce437
--- /dev/null
+++ b/finn-hlslib/docs/source/library/mmv.rst
@@ -0,0 +1,38 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+mmv.hpp
+====================================
+
+The mmv.hpp file collects the description of the class used for Matrix-Multi-Vector compute
+
+
+.. doxygenfile:: mmv.hpp
+
diff --git a/finn-hlslib/docs/source/library/streamtools.rst b/finn-hlslib/docs/source/library/streamtools.rst
new file mode 100755
index 000000000..242fd9ac7
--- /dev/null
+++ b/finn-hlslib/docs/source/library/streamtools.rst
@@ -0,0 +1,38 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+streamtools.h
+====================================
+
+The streamtools.h file collects multiple convenience functions for streams manipulation
+
+
+.. doxygenfile:: streamtools.h
+
diff --git a/finn-hlslib/docs/source/library/swg.rst b/finn-hlslib/docs/source/library/swg.rst
new file mode 100755
index 000000000..3f91177db
--- /dev/null
+++ b/finn-hlslib/docs/source/library/swg.rst
@@ -0,0 +1,42 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+slidingwindow.h
+====================================
+
+The slidingwindow.h file describes the HLS implementation of the sliding window unit, implementing the im2col for convolutional layers
+
+
+.. doxygenfile:: slidingwindow.h
+
+
+
+
+
diff --git a/finn-hlslib/docs/source/library/weights.rst b/finn-hlslib/docs/source/library/weights.rst
new file mode 100755
index 000000000..7a415894d
--- /dev/null
+++ b/finn-hlslib/docs/source/library/weights.rst
@@ -0,0 +1,45 @@
+.. Copyright (c) 2019, Xilinx, Inc.
+.. All rights reserved.
+
+.. Redistribution and use in source and binary forms, with or without
+.. modification, are permitted provided that the following conditions are met:
+
+.. 1. Redistributions of source code must retain the above copyright notice,
+.. this list of conditions and the following disclaimer.
+
+.. 2. Redistributions in binary form must reproduce the above copyright
+.. notice, this list of conditions and the following disclaimer in the
+.. documentation and/or other materials provided with the distribution.
+
+.. 3. Neither the name of the copyright holder nor the names of its
+.. contributors may be used to endorse or promote products derived from
+.. this software without specific prior written permission.
+
+.. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+.. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+.. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+.. PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+.. CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+.. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+.. PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+.. OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+.. WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+.. OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+====================================
+weigths.hpp
+====================================
+
+The weigths.hpp file lists a set of classes used to implement the weigths memory at multiple precisions.
+
+
+.. doxygenclass:: BinaryWeights
+ :members:
+ :private-members:
+
+.. doxygenclass:: FixedPointWeights
+ :members:
+ :private-members:
+
+
diff --git a/finn-hlslib/fclayer.h b/finn-hlslib/fclayer.h
new file mode 100755
index 000000000..643011937
--- /dev/null
+++ b/finn-hlslib/fclayer.h
@@ -0,0 +1,113 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *****************************************************************************/
+
+/*****************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog\xilinx.com>
+ * Thomas B. Preusser <thomas.preusser\utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin\xilinx.com>
+ *
+ * \file fclayer.h
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * This file lists a set of convenience funtions used to implement fully
+ * connected layers
+ *
+ *****************************************************************************/
+
+#ifndef FCLAYER_H
+#define FCLAYER_H
+
+#include <ap_int.h>
+#include <hls_stream.h>
+
+#include "streamtools.h"
+#include "mvau.hpp"
+
+/**
+ * \brief Fully connected layer implementation
+ *
+ * The function implements the fully connected layer, and it's basically a thin wrapper around the
+ * Matrix_Vector_Activate_Batch function to perform the stream width conversion.
+ *
+ * \tparam MatrixW Width of the input matrix
+ * \tparam MatrixH Heigth of the input matrix
+ * \tparam SIMD Number of input columns computed in parallel
+ * \tparam PE Number of output rows computed in parallel
+ * \tparam TSrcI DataType of the input activation (as used in the MAC)
+ * \tparam TDstI DataType of the output activation (as generated by the activation)
+ * \tparam TWeightI DataType of the weights (as used in the MAC)
+ * \tparam InStreamW Width of the input stream
+ * \tparam OutStreamW Width of the output stream
+ * \tparam TW DataType of the weights matrix - safely deducible from the paramaters
+ * \tparam TA DataType of the activation class (e.g. thresholds) - safely deducible from the paramaters
+ * \tparam R Datatype for the resource used for FPGA implementation of the MAC - safely deducible from the paramaters
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param weights Weights matrix (currently supports BinaryWeights or FixedPointWeights)
+ * \param activation Activation class
+ * \param reps Number of time the function has to be repeatedly executed (e.g. number of images)
+ * \param r Resource type for the hardware implementation of the MAC block
+ */
+template<
+ unsigned int MatrixW, unsigned int MatrixH, // geometry must be specified
+ unsigned int SIMD, unsigned int PE,
+
+ typename TSrcI = Identity, // redefine I/O interpretation as needed
+ typename TDstI = Identity,
+ typename TWeightI = Identity, // redefine I/O interpretation as needed for weigths
+
+ int InStreamW, int OutStreamW, // safely deducible (stream width must be int though!)
+ typename TW, typename TA, typename R
+>
+void StreamingFCLayer_Batch(hls::stream<ap_uint<InStreamW>> &in,
+ hls::stream<ap_uint<OutStreamW>> &out,
+ TW const &weights,
+ TA const &activation,
+ unsigned const reps,
+ R const &r) {
+#pragma HLS INLINE
+ unsigned const InpPerImage = MatrixW / InStreamW * TSrcI::width;
+ unsigned const OutPerImage = MatrixH / PE;
+
+ WidthAdjustedInputStream <InStreamW, SIMD*TSrcI::width, InpPerImage> wa_in (in, reps);
+ WidthAdjustedOutputStream<PE*TDstI::width, OutStreamW, OutPerImage> wa_out(out, reps);
+
+ Matrix_Vector_Activate_Batch<MatrixW, MatrixH, SIMD, PE, TSrcI, TDstI, TWeightI>
+ (static_cast<hls::stream<ap_uint<SIMD*TSrcI::width>>&>(wa_in),
+ static_cast<hls::stream<ap_uint<PE*TDstI::width>>&> (wa_out),
+ weights, activation, reps, r);
+}
+
+#endif
diff --git a/finn-hlslib/interpret.hpp b/finn-hlslib/interpret.hpp
new file mode 100755
index 000000000..6be17bd8b
--- /dev/null
+++ b/finn-hlslib/interpret.hpp
@@ -0,0 +1,286 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *****************************************************************************/
+
+/*****************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * @file interpret.hpp
+ *
+ * This project has received funding from the European Union's Framework
+ * Programme for Research and Innovation Horizon 2020 (2014-2020) under
+ * the Marie Skłodowska-Curie Grant Agreement No. 751339.
+ *
+ *****************************************************************************/
+
+#ifndef INTERPRET_HPP
+#define INTERPRET_HPP
+
+#include <ap_int.h>
+
+/**
+ * Thin wrapper around an ap_uint<1> redefining multiplication with
+ * another ap_uint<1> as XNOR operation for use in XNOR networks.
+ */
+class XnorMul {
+ ap_uint<1> const m_val;
+ public:
+ XnorMul(ap_uint<1> const val) : m_val(val) {
+#pragma HLS inline
+ }
+
+ public:
+ int operator*(ap_uint<1> const &b) const {
+#pragma HLS inline
+ return m_val == b? 1 : 0;
+ }
+};
+inline int operator*(ap_uint<1> const &a, XnorMul const &b) {
+#pragma HLS inline
+ return b*a;
+}
+
+class Binary {
+ public:
+ ap_uint<1> const m_val;
+ Binary(ap_uint<1> const val) : m_val(val) {
+#pragma HLS inline
+ }
+
+ public:
+ operator ap_int<2> () const {
+ return ap_int<2>(m_val? 1 : -1);
+ }
+ template<typename T>
+ auto operator*(T const &b) const -> decltype(ap_int<2>(1)*b) {
+#pragma HLS inline
+ return m_val? static_cast<decltype(-b)>(b) : -b;
+ }
+ friend std::ostream& operator<<(std::ostream&, Binary const&);
+};
+
+template<typename T>
+inline int operator*(T const &a, Binary const &b) {
+#pragma HLS inline
+ return b*a;
+}
+
+inline int operator*(Binary const &a, Binary const &b) {
+#pragma HLS inline
+ return (ap_int<2>) b* (ap_int<2>)a;
+}
+
+inline std::ostream& operator<<(std::ostream &out, Binary const &b) {
+ out << (b.m_val? "1" : "-1");
+ return out;
+}
+
+struct Identity {
+ static unsigned const width = 1;
+
+ template<typename T>
+ T const &operator()(T const &v) const {
+#pragma HLS inline
+ return v;
+ }
+
+ template<typename T>
+ T operator()() const {
+#pragma HLS inline
+ return T();
+ }
+};
+
+template<typename T>
+class Recast {
+ public:
+ static unsigned const width = 1;
+
+ private:
+ template<typename TV>
+ class Container {
+ TV m_val;
+ public:
+ Container(TV const &val) : m_val(val) {
+#pragma HLS inline
+ }
+
+ public:
+ T operator[](unsigned const idx) const {
+#pragma HLS inline
+ return T(m_val[idx]);
+ }
+ auto operator[](unsigned const idx) -> decltype(m_val[idx]) {
+#pragma HLS inline
+ return m_val[idx];
+ }
+ operator TV const&() const {
+#pragma HLS inline
+ return m_val;
+ }
+ };
+
+ public:
+ template<typename TV>
+ Container<TV> operator()(TV const &val) const {
+#pragma HLS inline
+ return Container<TV>(val);
+ }
+ template<typename TV>
+ Container<TV> operator()() const {
+#pragma HLS inline
+ return Container<TV>();
+ }
+};
+
+template<typename T>
+struct Caster {
+ template<int M>
+ static T cast(ap_int<M> const &arg) { return T(arg); }
+};
+
+template<int W, int I, ap_q_mode Q, ap_o_mode O, int N>
+struct Caster<ap_fixed<W, I, Q, O, N>> {
+ template<int M>
+ static ap_fixed<W, I, Q, O, N> cast(ap_int<M> const &arg) {
+ return *reinterpret_cast<ap_fixed<W, I, Q, O, N> const*>(&arg);
+ }
+};
+
+template<typename T, unsigned STRIDE=T::width>
+class Slice {
+ public:
+ static unsigned const width = STRIDE;
+
+ private:
+ template<typename TV>
+ class Container {
+ TV m_val;
+
+ public:
+ Container() {
+#pragma HLS inline
+ }
+ Container(TV const &val) : m_val(val) {
+#pragma HLS inline
+ }
+ public:
+ auto access(unsigned const mmv) -> decltype(m_val) {
+#pragma HLS inline
+ return m_val;;
+ }
+ auto operator()(unsigned const idx, unsigned const mmv) const -> decltype(m_val(STRIDE, 0)) {
+#pragma HLS inline
+ return m_val((idx+1)*STRIDE-1, idx*STRIDE);
+ }
+ auto operator[](unsigned mmv) const -> decltype(m_val) {
+#pragma HLS inline
+ return m_val;
+ }
+ operator TV const&() const {
+#pragma HLS inline
+ return m_val;
+ }
+ };
+
+ public:
+ template<typename TV>
+ Container<TV> operator()(TV const &val) const {
+#pragma HLS inline
+ return Container<TV>(val);
+ }
+ template<typename TV>
+ Container<TV> operator()() const {
+#pragma HLS inline
+ return Container<TV>();
+ }
+ template<typename TV>
+ Container<TV> operator() (TV const &val, unsigned mmv) const {
+#pragma HLS inline
+ return Container<TV>(val);
+ }
+};
+
+// This class is done for Slicing an MMV container (vector of ap_uint
+template<typename T, unsigned MMV, unsigned STRIDE=T::width>
+class Slice_mmv {
+ public:
+ static unsigned const width = STRIDE;
+ private:
+ template<typename TV>
+ class Container {
+ TV m_val;
+
+ public:
+ Container() {
+#pragma HLS inline
+ }
+ Container(TV const &val, unsigned mmv) : m_val(val){
+#pragma HLS inline
+ }
+ public:
+ operator TV const&() const {
+#pragma HLS inline
+ return m_val;
+ };
+ auto operator()(unsigned const idx, unsigned const mmv) const -> decltype(m_val.data[mmv](STRIDE, 0)) {
+#pragma HLS inline
+ return m_val.data[mmv]((idx+1)*STRIDE-1, idx*STRIDE);
+ };
+ auto operator[](unsigned const mmv) const -> decltype(m_val.data[mmv]) {
+#pragma HLS inline
+ return m_val.data[mmv];
+ }
+ };
+
+ public:
+ template<typename TV>
+ Container<TV> operator()(TV const &val) const {
+#pragma HLS inline
+ return Container<TV>(val);
+ }
+ template<typename TV>
+ Container<TV> operator()() const {
+#pragma HLS inline
+ return Container<TV>();
+ }
+ template<typename TV>
+ Container<TV> operator()(TV const &val, unsigned mmv) {
+#pragma HLS inline
+ return Container<TV>(val,mmv);
+ }
+};
+
+#endif
\ No newline at end of file
diff --git a/finn-hlslib/mac.hpp b/finn-hlslib/mac.hpp
new file mode 100755
index 000000000..38136c3d3
--- /dev/null
+++ b/finn-hlslib/mac.hpp
@@ -0,0 +1,206 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *****************************************************************************/
+
+/*****************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * \file mac.hpp
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * This file lists a set of convenience funtions used to implement
+ * multipliers with selectable implementation resource
+ *
+ * This project has received funding from the European Union's Framework
+ * Programme for Research and Innovation Horizon 2020 (2014-2020) under
+ * the Marie Skłodowska-Curie Grant Agreement No. 751339.
+ *
+ *****************************************************************************/
+
+/*****************************************************************************
+ * MAC operation template:
+ *
+ * mac<N, T, TC, TD>(T a, TC c[N], TD d[N])
+ * = a + SUM_{i=0}^{N-1} c(i)*d(i)
+ *
+ * All template arguments but N can typically be inferred.
+ *
+ * mac<ap_uint<14>>(0, c, d)
+ *****************************************************************************/
+
+#ifndef MAC_HPP
+#define MAC_HPP
+
+#include "utils.hpp"
+
+
+/**
+ * \brief Multipliy operation between 2 operands, HLS choose the best resource
+ *
+ * The same multiply operation can be implemented using multiple Vivado HLS pragmas to select the
+ * hardware resource to be used:
+ * ap_resource_dflt will let HLS choose the best one
+ * ap_resource_lut will force HLS to implement the multiplier in LUTs
+ * ap_resource_dsp will force HLS to implement the multiplier in DSP48
+ *
+ * \tparam TC First operand datatype (weights)
+ * \tparam TD Second operand datatype (input)
+ *
+ * \param c First operand (array of weights)
+ * \param d Second operand (array of input activation)
+ * \param r Resource type for the hardware implementation of the MAC block
+ *
+ * \return Result of the multiply operation
+ */
+template<typename TC, typename TD>
+auto mul(TC const &c, TD const &d, ap_resource_dflt const&) -> decltype(c*d) {
+#pragma HLS inline
+ auto r = c*d;
+ return r;
+}
+
+/**
+ * \brief Multiply operation between 2 operands, implemented in LUT
+ *
+ * The same multiply operation can be implemented using multiple Vivado HLS pragmas to select the
+ * hardware resource to be used:
+ * ap_resource_dflt will let HLS choose the best one
+ * ap_resource_lut will force HLS to implement the multiplier in LUTs
+ * ap_resource_dsp will force HLS to implement the multiplier in DSP48
+ *
+ * \tparam TC First operand datatype (weights)
+ * \tparam TD Second operand datatype (input)
+ *
+ * \param c First operand (array of weights)
+ * \param d Second operand (array of input activation)
+ * \param r Resource type for the hardware implementation of the MAC block
+ *
+ * \return Result of the multiply operation
+ */
+template<typename TC, typename TD>
+auto mul(TC const &c, TD const &d, ap_resource_lut const&) -> decltype(c*d) {
+#pragma HLS inline
+ decltype(c*d) const res = c*d;
+#pragma HLS RESOURCE variable=res core=Mul_LUT
+ return res;
+}
+
+/**
+ * \brief Multipliy operation between 2 operands, implemented in a DSP48
+ *
+ * The same multiply operation can be implemented using multiple Vivado HLS pragmas to select the
+ * hardware resource to be used:
+ * ap_resource_dflt will let HLS choose the best one
+ * ap_resource_lut will force HLS to implement the multiplier in LUTs
+ * ap_resource_dsp will force HLS to implement the multiplier in DSP48
+ *
+ * \tparam TC First operand datatype (weights)
+ * \tparam TD Second operand datatype (input)
+ *
+ * \param c First operand (array of weights)
+ * \param d Second operand (array of input activation)
+ * \param r Resource type for the hardware implementation of the MAC block
+ *
+ * \return Result of the multiply operation
+ */
+template<typename TC, typename TD>
+auto mul(TC const &c, TD const &d, ap_resource_dsp const&) -> decltype(c*d) {
+#pragma HLS inline
+ decltype(c*d) const res = c*d;
+#pragma HLS RESOURCE variable=res core=DSP48
+ return res;
+}
+
+/**
+ * \brief MAC with selectable implementation resource, used by Matrix_Vector_Activate_Batch
+ *
+ * \tparam N Number of MAC to be performed (equals to SIMD in mvau)
+ * \tparam T Accumulator datatype
+ * \tparam TC First operand datatype (weights)
+ * \tparam TD Second operand datatype (input)
+ * \tparam R Datatype for the resource used for FPGA implementation of the MAC - safely deducible from the paramaters
+ *
+ * \param a Initialization value of the accumulation
+ * \param c First operand (array of weights)
+ * \param d Second operand (array of input activation)
+ * \param r Resource type for the hardware implementation of the MAC block
+ * \param mmv MMV value to address accumulator and activation
+ *
+ * \return Result of the MAC operation
+ */
+template<unsigned N, typename T, typename TC, typename TD, typename R>
+T mac(T const &a, TC const &c, TD const &d, R const &r, unsigned mmv) {
+#pragma HLS inline
+ T res = a;
+ for(unsigned i = 0; i < N; i++) {
+#pragma HLS unroll
+ res += mul(c[i], d(i,mmv), r);
+ }
+ return res;
+}
+
+/**
+ * \brief MAC with selectable implementation resource, used by Matrix_Vector_Activate_Batch
+ *
+ * \tparam N Number of MAC to be performed (equals to SIMD in mvau)
+ * \tparam T Accumulator datatype
+ * \tparam TC First operand datatype (weights)
+ * \tparam TD Second operand datatype (input)
+ * \tparam R Datatype for the resource used for FPGA implementation of the MAC - safely deducible from the paramaters
+ *
+ * \param a Initialization value of the accumulation
+ * \param c First operand (array of weights)
+ * \param d Second operand (array of input activation)
+ * \param r Resource type for the hardware implementation of the MAC block
+ *
+ * \return Result of the MAC operation
+ */
+template<unsigned N, typename T, typename TC, typename TD, typename R>
+T mac(T const &a, TC const &c, TD const &d, R const &r) {
+#pragma HLS inline
+ T res = a;
+ for(unsigned i = 0; i < N; i++) {
+#pragma HLS unroll
+ res += mul(c[i], d[i], r);
+ }
+ return res;
+}
+template<unsigned N, typename T, typename TC, typename TD>
+inline T mac(T const &a, TC const &c, TD const &d) {
+#pragma HLS inline
+ return mac<N>(a, c, d, ap_resource_dflt());
+}
+
+#endif
diff --git a/finn-hlslib/maxpool.h b/finn-hlslib/maxpool.h
new file mode 100755
index 000000000..37f862d38
--- /dev/null
+++ b/finn-hlslib/maxpool.h
@@ -0,0 +1,387 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * This file implement the BNN maxpool layer.
+ *
+ ******************************************************************************/
+
+#ifndef MAXPOOL_H
+#define MAXPOOL_H
+
+#include <limits>
+
+/**
+ * \brief Max Pool implementation for Binarized values
+ *
+ * \tparam ImgDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam PoolDim Dimension of the Max Pool kernel (assumed square)
+ * \tparam NumChannels Number of Input Feature Maps
+ *
+ * \param in Input stream
+ * \param out Output stream
+ *
+ */
+template<unsigned int ImgDim, unsigned int PoolDim, unsigned int NumChannels>
+void StreamingMaxPool(stream<ap_uint<NumChannels> > & in,
+ stream<ap_uint<NumChannels> > & out) {
+ CASSERT_DATAFLOW(ImgDim % PoolDim == 0);
+ // need buffer space for a single maxpooled row of the image
+ ap_uint<NumChannels> buf[ImgDim / PoolDim];
+ for(unsigned int i = 0; i < ImgDim / PoolDim; i++) {
+#pragma HLS UNROLL
+ buf[i] = 0;
+ }
+
+ for (unsigned int yp = 0; yp < ImgDim / PoolDim; yp++) {
+ for (unsigned int ky = 0; ky < PoolDim; ky++) {
+ for (unsigned int xp = 0; xp < ImgDim / PoolDim; xp++) {
+#pragma HLS PIPELINE II=1
+ ap_uint<NumChannels> acc = 0;
+ for (unsigned int kx = 0; kx < PoolDim; kx++) {
+ acc = acc | in.read();
+ }
+ // pool with old value in row buffer
+ buf[xp] |= acc;
+ }
+ }
+ for (unsigned int outpix = 0; outpix < ImgDim / PoolDim; outpix++) {
+#pragma HLS PIPELINE II=1
+ out.write(buf[outpix]);
+ // get buffer ready for next use
+ buf[outpix] = 0;
+ }
+ }
+}
+
+/**
+ * \brief Max Pool implementation for Binarized values
+ *
+ * \tparam ImgDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam PoolDim Dimension of the Max Pool kernel (assumed square)
+ * \tparam NumChannels Number of Input Feature Maps
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of time the function has to be repeatedly executed (e.g. number of images)
+ *
+ */
+template<unsigned int ImgDim, unsigned int PoolDim, unsigned int NumChannels>
+void StreamingMaxPool_Batch(stream<ap_uint<NumChannels> > & in,
+ stream<ap_uint<NumChannels> > & out, unsigned int numReps) {
+ for (unsigned int rep = 0; rep < numReps; rep++) {
+ StreamingMaxPool<ImgDim, PoolDim, NumChannels>(in, out);
+ }
+}
+
+
+/**
+ * \brief Max Pool implementation for Binarized values
+ *
+ * \tparam ImgDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam PoolDim Dimension of the Max Pool kernel (assumed square)
+ * \tparam NumChannels Number of Input Feature Maps
+ * \tparam ActType DataType of the input activation (as used in the comparison)
+ * \tparam min_value Minimum value possible with the given ActType, used to initialize the value before the comparison
+ * \tparam StreamW Width of the input and output stream
+ *
+ * \param in Input stream
+ * \param out Output stream
+ *
+ */
+template<unsigned int ImgDim, unsigned int PoolDim, unsigned int NumChannels, typename ActType, int min_value,
+ int StreamW
+ >
+void StreamingMaxPool_Precision(stream<ap_uint<StreamW> > & in,
+ stream<ap_uint<StreamW> > & out) {
+ CASSERT_DATAFLOW(ImgDim % PoolDim == 0);
+ // need buffer space for a single maxpooled row of the image
+ ActType buf[ImgDim / PoolDim][NumChannels];
+#pragma HLS ARRAY_PARTITION variable=buf complete dim=2
+ for(unsigned int i = 0; i < ImgDim / PoolDim; i++) {
+ for(unsigned int ch = 0; ch<NumChannels; ch++){
+#pragma HLS UNROLL
+ buf[i][ch] = min_value; //std::numeric_limits<ActType>::min();
+ }
+ }
+ ap_uint<StreamW> inputData,outputData;
+ for (unsigned int yp = 0; yp < ImgDim / PoolDim; yp++) {
+ for (unsigned int ky = 0; ky < PoolDim; ky++) {
+ for (unsigned int xp = 0; xp < ImgDim / PoolDim; xp++) {
+ // Change to comparator
+ for (unsigned int kx = 0; kx < PoolDim; kx++) {
+#pragma HLS PIPELINE II=1
+ inputData = in.read();
+ for(unsigned int ch = 0; ch<NumChannels; ch++){
+#pragma HLS UNROLL
+ unsigned int lowBit = ch * ActType::width;
+ unsigned int highBit = (ch+1) * ActType::width -1;
+ ActType channeldata = inputData(highBit, lowBit);
+ ActType oldMax = buf[xp][ch];
+ if(channeldata > oldMax){
+ buf[xp][ch] = channeldata;
+ }
+ }
+ }
+ }
+ }
+ for (unsigned int outpix = 0; outpix < ImgDim / PoolDim; outpix++) {
+ for(unsigned int ch = 0; ch < NumChannels; ch++){
+#pragma HLS UNROLL
+ unsigned int lowBit = ch * ActType::width;
+ unsigned int highBit = (ch+1) * ActType::width -1;
+ outputData(highBit, lowBit) = buf[outpix][ch];
+ // get buffer ready for next use
+ buf[outpix][ch] = min_value;
+ }
+ out.write(outputData);
+ }
+ }
+}
+/**
+ * \brief Max Pool implementation for Binarized values
+ *
+ * \tparam ImgDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam PoolDim Dimension of the Max Pool kernel (assumed square)
+ * \tparam NumChannels Number of Input Feature Maps
+ * \tparam ActType DataType of the input activation (as used in the comparison)
+ * \tparam min_value Minimum value possible with the given ActType, used to initialize the value before the comparison
+ * \tparam StreamW Width of the input and output stream
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of time the function has to be repeatedly executed (e.g. number of images)
+ *
+ */
+template<unsigned int ImgDim, unsigned int PoolDim, unsigned int NumChannels, typename ActType, int min_value,
+ int InStreamW, int OutStreamW // safely deducible (stream width must be int though!)
+ >
+void StreamingMaxPool_Precision_Batch(stream<ap_uint<InStreamW> > & in,
+ stream<ap_uint<OutStreamW> > & out, unsigned int numReps) {
+#pragma HLS INLINE
+ unsigned const InpPerImage = ImgDim*ImgDim*NumChannels*ActType::width/InStreamW ;
+ unsigned const OutPerImage = ImgDim*ImgDim / (PoolDim*PoolDim);
+ WidthAdjustedInputStream <InStreamW, NumChannels*ActType::width, InpPerImage> wa_in (in, numReps);
+ WidthAdjustedOutputStream<NumChannels*ActType::width, OutStreamW, OutPerImage> wa_out(out, numReps);
+ for (unsigned int rep = 0; rep < numReps; rep++) {
+ StreamingMaxPool_Precision<ImgDim, PoolDim, NumChannels, ActType, min_value>
+ (static_cast<hls::stream<ap_uint<NumChannels*ActType::width>>&>(wa_in),
+ static_cast<hls::stream<ap_uint<NumChannels*ActType::width>>&>(wa_out));
+ }
+}
+
+
+
+/**
+ * \brief ReLU for fixed-point or integer; can accept a bias at input, which it removes
+ *
+ * \tparam ImgDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam NumChannels Number of Input Feature Maps
+ * \tparam ActType DataType of the input activation (as used in the comparison)
+ * \tparam PECount PE parallelism to apply ReLU
+ * \tparam offset Offset to be subtracted before applying ReLU
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of time the function has to be repeatedly executed (e.g. number of images)
+ *
+ */
+template<
+ unsigned int ImgDim,
+ unsigned int NumChannels,
+ typename ActType,
+ unsigned int PECount,
+ int offset = 0>
+void ReLU_Batch(stream<ap_uint<PECount * ActType::width> > & in,
+ stream<ap_uint<PECount * ActType::width> > & out, const unsigned int numReps) {
+
+ ap_uint<PECount * ActType::width> thin;
+ ap_uint<PECount * ActType::width> thout;
+
+ //call to thresholding library function
+ for(unsigned int reps=0; reps<numReps; reps++){
+ for(unsigned int pixel=0; pixel<ImgDim*ImgDim; pixel++){
+ for(unsigned int fold=0; fold<NumChannels/PECount; fold++){
+#pragma HLS PIPELINE II=1
+ thin = in.read();
+ for(unsigned int pe=0; pe<PECount; pe++){
+ #pragma HLS UNROLL
+ // Threshold and assign to right bits of output buffers
+ unsigned int lowBit = pe * ActType::width;
+ unsigned int highBit = (pe+1) * ActType::width - 1;
+ ActType val = thin(highBit,lowBit);
+ ActType result;
+ if(val < offset)
+ result = 0;
+ else
+ result = val - offset;
+ thout(highBit, lowBit) = result;
+ }
+ out.write(thout);
+ }
+ }
+ }
+}
+
+/**
+ * \brief Accumulate-pool - like average pooling over the whole frame, but without the dividion at end
+ *
+ * \tparam ImgDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam NumChannels Number of Input Feature Maps
+ * \tparam ActType DataType of the input activation (as used in the comparison)
+ * \tparam PECount PE parallelism to apply ReLU
+ * \tparam AccType Datatype of the accumulation (e.g. output)
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of time the function has to be repeatedly executed (e.g. number of images)
+ *
+ */
+template<
+ unsigned int ImgDim,
+ unsigned int NumChannels,
+ typename ActType,
+ unsigned int PECount,
+ typename AccType>
+void AccPool_Batch(stream<ap_uint<PECount * ActType::width> > & in,
+ stream<ap_uint<PECount * AccType::width> > & out, const unsigned int numReps) {
+ ap_uint<PECount * ActType::width> thin;
+ ap_uint<PECount * AccType::width> accumulators[NumChannels/PECount];
+#pragma HLS RESOURCE variable=accumulators core=RAM_2P_LUTRAM
+
+ //call to thresholding library function
+ for(unsigned int reps=0; reps<numReps; reps++){
+ for(unsigned int pixel=0; pixel<ImgDim*ImgDim; pixel++){
+ for(unsigned int fold=0; fold<NumChannels/PECount; fold++){
+#pragma HLS PIPELINE II=1
+ thin = in.read();
+ ap_uint<PECount * AccType::width> accbank = accumulators[fold];
+ for(unsigned int pe=0; pe<PECount; pe++){
+ #pragma HLS UNROLL
+ // Threshold and assign to right bits of output buffers
+ unsigned int lowBit = pe * ActType::width;
+ unsigned int highBit = (pe+1) * ActType::width - 1;
+ ActType val = thin((pe+1) * ActType::width - 1,pe * ActType::width);
+ AccType acc = accbank((pe+1) * AccType::width - 1,pe * AccType::width);
+ AccType result;
+ if(pixel == 0)
+ result = val;
+ else
+ result = val+acc;
+ accbank((pe+1) * AccType::width - 1,pe * AccType::width) = result;
+ }
+ accumulators[fold] = accbank;
+ }
+ }
+ for (unsigned int fold = 0; fold < NumChannels / PECount; fold++)
+ {
+ out.write(accumulators[fold]);
+ }
+ }
+}
+
+
+
+/**
+ * \brief LabelSelect_Batch - returns labels of top-5 in stream
+ *
+ * \tparam NumClasses Number of classes of the dataset
+ * \tparam PECount Number of inputs to be processed in parallel
+ * \tparam NumTop Number of top classes to be selected in output
+ * \tparam In_T Datatype of the input
+ * \tparam Out_T Datatype of the output
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of times the function has to be repeatedly executed (e.g. number of images)
+ *
+ */
+template<
+ // tensor size parameters
+ unsigned int NumClasses,
+ unsigned int PECount,
+ unsigned int NumTop,
+ typename In_T,
+ typename Out_T>
+void LabelSelect_Batch(stream<ap_uint<PECount * In_T::width> > & in,
+ stream<ap_uint<32> > & out, const unsigned int numReps) {
+ ap_uint<PECount * In_T::width> inval;
+ Out_T toplabels[NumTop];
+#pragma HLS ARRAY_PARTITION variable=toplabels complete dim=1
+ In_T topval[NumTop];
+#pragma HLS ARRAY_PARTITION variable=topval complete dim=1
+for(unsigned int reps=0; reps<numReps; reps++){
+ unsigned int idx = 0;
+ for(unsigned int topx=0; topx<NumTop; topx++){
+ #pragma HLS UNROLL
+ topval[topx] = 1<<31;
+ }
+ for(unsigned int block=0; block<(NumClasses/PECount); block++){
+ #pragma HLS PIPELINE II=1
+ inval = in.read();
+ for(unsigned int elem=0; elem<PECount; elem++){
+ unsigned int lowBit = elem * In_T::width;
+ unsigned int highBit = (elem+1) * In_T::width - 1;
+ In_T val = inval(highBit,lowBit);
+ for(unsigned int topx=0; topx<NumTop; topx++){
+ #pragma HLS UNROLL
+ if(val > topval[topx]){
+ if(topx==(NumTop-1)){
+ topval[topx] = val;
+ toplabels[topx] = idx;
+ } else if(val > topval[topx+1]){
+ topval[topx] = topval[topx+1];
+ toplabels[topx] = toplabels[topx+1];
+ } else {
+ topval[topx] = val;
+ toplabels[topx] = idx;
+ }
+ }
+ }
+ idx++;
+ }
+ }
+ for(unsigned int topx = 0; topx < NumTop; topx++){
+ out.write(toplabels[NumTop - topx - 1]);
+ }
+ }
+}
+
+#endif
diff --git a/finn-hlslib/mmv.hpp b/finn-hlslib/mmv.hpp
new file mode 100755
index 000000000..89a00afcd
--- /dev/null
+++ b/finn-hlslib/mmv.hpp
@@ -0,0 +1,60 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file
+ *
+ * This file described the MultiChanData class used for MMV, whenever we exploit
+ * the pixel level of parallelism.
+ *
+ ******************************************************************************/
+
+#ifndef MMVCLASS_H
+#define MMVCLASS_H
+
+#include "hls_stream.h"
+
+template <unsigned int NumChannels, unsigned int DataWidth>
+class MultiChanData {
+public: ap_uint<DataWidth> data[NumChannels];
+ auto operator[](unsigned const mm) -> decltype(data[mm]) {
+#pragma HLS inline
+ return data[mm];
+ }
+
+};
+
+
+#endif
\ No newline at end of file
diff --git a/finn-hlslib/mvau.hpp b/finn-hlslib/mvau.hpp
new file mode 100755
index 000000000..3c63abe64
--- /dev/null
+++ b/finn-hlslib/mvau.hpp
@@ -0,0 +1,183 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *******************************************************************************/
+
+/*******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * \file mvau.hpp
+ *
+ * This file lists a templated funtion used to implement
+ * Matrix-Vector-Activation Unit
+ *
+ * This project has received funding from the European Union's Framework
+ * Programme for Research and Innovation Horizon 2020 (2014-2020) under
+ * the Marie Skłodowska-Curie Grant Agreement No. 751339.
+ *
+ *******************************************************************************/
+
+#ifndef MVAU_HPP
+#define MVAU_HPP
+
+#include "hls_stream.h"
+
+#include "mac.hpp"
+#include "interpret.hpp"
+
+/**
+ * \brief Matrix vector activate function
+ *
+ * The function performs the multiplication between a weigth matrix and the input activation vector,
+ * accumulating the results and then applying an activation function on the accumulated result.
+ *
+ *
+ * \tparam MatrixW Width of the input matrix
+ * \tparam MatrixH Heigth of the input matrix
+ * \tparam SIMD Number of input columns computed in parallel
+ * \tparam PE Number of output rows computed in parallel
+ * \tparam MMV Number of output pixels computed in parallel
+ * \tparam TSrcI DataType of the input activation (as used in the MAC)
+ * \tparam TDstI DataType of the output activation (as generated by the activation)
+ * \tparam TWeightI DataType of the weights (as used in the MAC)
+ * \tparam TI DataType of the input stream - safely deducible from the paramaters
+ * \tparam TO DataType of the output stream - safely deducible from the paramaters
+ * \tparam TW DataType of the weights matrix - safely deducible from the paramaters
+ * \tparam TA DataType of the activation class (e.g. thresholds) - safely deducible from the paramaters
+ * \tparam R Datatype for the resource used for FPGA implementation of the MAC - safely deducible from the paramaters
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param weights Weights matrix (currently supports BinaryWeights or FixedPointWeights)
+ * \param activation Activation class
+ * \param reps Number of time the function has to be repeatedly executed (e.g. number of images)
+ * \param r Resource type for the hardware implementation of the MAC block
+ */
+template<
+ unsigned MatrixW, unsigned MatrixH, unsigned SIMD, unsigned PE, unsigned MMV,
+ typename TSrcI = Identity, typename TDstI = Identity, typename TWeightI = Identity,
+ typename TI, typename TO, typename TW, typename TA, typename R
+>
+void Matrix_Vector_Activate_Batch(hls::stream<TI> &in,
+ hls::stream<TO> &out,
+ TW const &weights,
+ TA const &activation,
+ int const reps,
+ R const &r) {
+
+ // how many different rows each neuron will compute
+ // alternatively: number of vertical matrix chunks
+ unsigned const NF = MatrixH / PE;
+
+ // how many synapse groups each row is split into
+ // alternatively: number of horizontal matrix chunks
+ unsigned const SF = MatrixW / SIMD;
+
+ // input vector buffers
+ TI inputBuf[SF];
+#pragma HLS ARRAY_PARTITION variable=inputBuf complete dim=0
+
+
+ decltype(activation.init(0,0)) accu[MMV][PE];
+#pragma HLS ARRAY_PARTITION variable=accu complete dim=0
+
+ unsigned nf = 0;
+ unsigned sf = 0;
+ unsigned tile = 0; // invariant: tile = nf*SF + sf
+
+ // everything merged into a common iteration space (one "big" loop instead
+ // of smaller nested loops) to get the pipelinening the way we want
+ unsigned const TOTAL_FOLD = NF * SF;
+ for(unsigned i = 0; i < reps * TOTAL_FOLD; i++) {
+#pragma HLS PIPELINE II=1
+ TI inElem;
+ if(nf == 0) {
+ // read input from stream
+ inElem = in.read();
+ // store in appropriate buffer for reuse
+ inputBuf[sf] = inElem;
+//#pragma HLS ARRAY_PARTITION variable=inputBuf[sf].data complete dim=1
+// TODO: Fix this and including in contructor of inputBuf
+ }
+ else {
+ // reuse buffered input
+ inElem = inputBuf[sf];
+ }
+
+ // Threshold Initialisation
+ if(sf == 0) {
+ for(unsigned pe = 0; pe < PE; pe++) {
+ for(unsigned mmv = 0; mmv < MMV; mmv++) {
+#pragma HLS UNROLL
+ accu[mmv][pe] = activation.init(nf, pe);
+ }
+ }
+ }
+
+ // compute matrix-vector product for each processing element
+ auto const &w = weights.weights(tile);
+ for(unsigned pe = 0; pe < PE; pe++) {
+#pragma HLS UNROLL
+ auto const wgt = TWeightI()(w[pe]);
+ for (unsigned mmv = 0; mmv < MMV; mmv++){
+ auto const act = TSrcI()(inElem, mmv);
+ accu[mmv][pe] = mac<SIMD>(accu[mmv][pe], wgt, act, r, mmv);
+ }
+ }
+
+ // keep track of which folded synapse/neuron we are processing
+ ++tile;
+ if(++sf == SF) {
+ // produce output and clear accumulators
+ auto outElem = TDstI().template operator()<TO>();
+ for (unsigned pe = 0; pe < PE; pe++) {
+#pragma HLS UNROLL
+ for (unsigned mmv = 0; mmv < MMV; mmv++){
+#pragma HLS UNROLL
+ outElem(pe,mmv) = activation.activate(nf, pe, accu[mmv][pe]);
+ }
+ }
+
+ out.write(outElem);
+
+ // next folded neuron or image
+ sf = 0;
+ if(++nf == NF) {
+ nf = 0;
+ tile = 0;
+ }
+ }
+ }
+}
+#endif
diff --git a/finn-hlslib/requirements.txt b/finn-hlslib/requirements.txt
new file mode 100755
index 000000000..751f77f4b
--- /dev/null
+++ b/finn-hlslib/requirements.txt
@@ -0,0 +1,59 @@
+# Copyright (c) 2019, Xilinx, Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice,
+# this list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in the
+# documentation and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+# OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+# ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+alabaster==0.7.12
+Babel==2.7.0
+breathe==4.13.0.post0
+certifi==2019.3.9
+chardet==3.0.4
+commonmark==0.9.0
+docutils==0.14
+future==0.17.1
+idna==2.8
+imagesize==1.1.0
+Jinja2==2.10.1
+MarkupSafe==1.1.1
+packaging==19.0
+Pygments==2.4.1
+pyparsing==2.4.0
+pytz==2019.1
+recommonmark==0.5.0
+requests==2.22.0
+six==1.12.0
+snowballstemmer==1.2.1
+Sphinx==2.0.1
+sphinx-rtd-theme==0.4.3
+sphinxcontrib-applehelp==1.0.1
+sphinxcontrib-devhelp==1.0.1
+sphinxcontrib-htmlhelp==1.0.2
+sphinxcontrib-jsmath==1.0.1
+sphinxcontrib-qthelp==1.0.2
+sphinxcontrib-serializinghtml==1.1.3
+urllib3==1.25.3
+numpy==1.16.4
diff --git a/finn-hlslib/slidingwindow.h b/finn-hlslib/slidingwindow.h
new file mode 100755
index 000000000..77496815c
--- /dev/null
+++ b/finn-hlslib/slidingwindow.h
@@ -0,0 +1,331 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *******************************************************************************/
+
+ /******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * \file slidingwindow.h
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * This file lists a set of convenience funtions used to implement
+ * Sliding window generator for convolutions
+ *
+ *****************************************************************************/
+
+#ifndef SLIDINGWINDOW_H
+#define SLIDINGWINDOW_H
+
+
+#define MAX(x, y) (((x) > (y)) ? (x) : (y)) /* \brief Maximum value between x and y*/
+#define MIN(x, y) (((x) > (y)) ? (y) : (x)) /* !< \brief Minimum value between x and y*/
+
+
+/**
+ * \brief Sliding Window unit that produces output vectors for feeding
+ * a Matrix_Vector_Activate_Batch, implementing the im2col algorithm. To be used only if
+ * ConvKernelDim%Stride = 0
+ *
+ * \tparam ConvKernelDim Dimension of the convolutional kernel (assumed square)
+ * \tparam IFMChannels Number of Input Feature Maps
+ * \tparam Input_precision Number bits per pixel
+ * \tparam IFMDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam OFMDim Width and Heigth of the Output Feature Map (assumed square)
+ * \tparam SIMD Number of input columns computed in parallel
+ * \tparam Stride Stride of the convolutional kernel
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of time the function has to be repeatedly executed (e.g. number of images)
+ */
+template<unsigned int ConvKernelDim,
+ unsigned int IFMChannels,
+ unsigned int Input_precision,
+ unsigned int IFMDim,
+ unsigned int OFMDim,
+ unsigned int SIMD,
+ unsigned int Stride = 1>
+void ConvolutionInputGenerator(
+ stream<ap_uint<SIMD*Input_precision> > & in,
+ stream<ap_uint<SIMD*Input_precision> > & out,
+ const unsigned int numReps = 1) {
+ CASSERT_DATAFLOW(IFMChannels % SIMD == 0);
+ CASSERT_DATAFLOW(ConvKernelDim % Stride == 0);
+ const unsigned int multiplying_factor = IFMChannels/SIMD;
+ const unsigned int number_blocks = ConvKernelDim/Stride + 1 ;
+ ap_uint<SIMD*Input_precision> inputBuf[number_blocks][Stride * IFMDim * multiplying_factor];
+
+#pragma HLS ARRAY_PARTITION variable=inputBuf complete dim=1
+#pragma HLS RESOURCE variable inputBuf core=RAM_2P
+ const unsigned int cycles_write_block = (OFMDim * ConvKernelDim * ConvKernelDim * multiplying_factor);
+ const unsigned int cycles_read_block = Stride * IFMDim * multiplying_factor;
+ const unsigned int max_cycles = MAX(cycles_write_block,cycles_read_block);
+ const unsigned int baseIter = IFMDim * ConvKernelDim * multiplying_factor// Initial buffer
+ + OFMDim * MAX(cycles_write_block,cycles_read_block);
+ unsigned int counter_internal_block = 0;
+ unsigned int current_block_write = 0;
+ unsigned int next_block_write = 0;
+ unsigned int current_line = 0;
+ unsigned int read_block = 0;
+ unsigned int inp = 0, ofm_y = 0, ofm_x = 0, k_y = 0, k_x = 0, count_simd =0;
+#pragma HLS reset variable=inp
+ for (unsigned int count_image = 0; count_image < numReps; count_image++) {
+ for (unsigned int i = 0; i < baseIter; i++) {
+#pragma HLS PIPELINE II=1
+ if (inp < IFMDim * ConvKernelDim*multiplying_factor) {// Initial buffer of ConvKernelDim lines
+ ap_uint<SIMD*Input_precision> inElem;
+ inElem = in.read();
+ inputBuf[current_block_write][current_line] = inElem;
+ current_line++;
+ inp++;
+ if (current_line == Stride * IFMDim * multiplying_factor ) {
+ current_line = 0;
+ current_block_write++;
+ if (current_block_write == number_blocks) {
+ current_block_write=0;
+ }
+ read_block++;
+ counter_internal_block = 0;
+ }
+ } else {
+ if (counter_internal_block < cycles_write_block-1) { // We are writing output, MMV IFMChan per cycle
+ unsigned int current_block_read = (current_block_write + 1 + k_y / Stride);
+ if (current_block_read >= number_blocks) {
+ current_block_read-= number_blocks;
+ }
+ unsigned int current_line_in_block = ((k_y%Stride) * IFMDim + ofm_x*Stride + k_x)*multiplying_factor + count_simd;
+ ap_uint<SIMD*Input_precision> outElem = inputBuf[current_block_read][(current_line_in_block)];
+ out.write(outElem);
+ count_simd++;
+ if (count_simd == multiplying_factor) {
+ count_simd=0;
+ k_x++;
+ if (k_x == ConvKernelDim) {
+ k_x = 0;
+ k_y++;
+ if (k_y == ConvKernelDim) {
+ k_y = 0;
+ ofm_x ++;
+ if (ofm_x == OFMDim) {
+ ofm_x = 0;
+ ofm_y++;
+ if (ofm_y == OFMDim) {
+ ofm_y = 0;
+ inp = 0;
+ }
+ }
+ }
+ }
+ }
+ }
+ if ((counter_internal_block < cycles_read_block-1) && (read_block<IFMDim/Stride)) { // In parallel we write in the buffer, in the current block write if we still need to
+ ap_uint<SIMD*Input_precision> inElem;
+ inElem = in.read();
+ inputBuf[current_block_write][current_line] = inElem;
+#pragma AP dependence variable=inputBuf intra false
+#pragma AP dependence variable=inputBuf inter false
+ current_line++;
+ if (current_line == Stride * IFMDim * multiplying_factor) {// We read the whole block, we change the next block in which we want to we
+ // We filled up a block, let's not read until
+ current_line = 0;
+ read_block++;
+ current_block_write++;
+ if (current_block_write == number_blocks) {
+ current_block_write=0;
+ }
+#pragma AP dependence variable=current_block_write intra false
+ }
+ }
+ counter_internal_block++; // = (counter_internal_block +1) % max_cycles;
+ if (counter_internal_block == (max_cycles-1)) {
+ counter_internal_block = 0;
+ }
+ }
+ } // End base_iter
+ read_block = 0;
+ } // End count_image
+} // End generator
+
+/**
+ * \brief Sliding Window unit that produces output vectors for feeding
+ * a Matrix_Vector_Activate_Batch, implementing the im2col algorithm with support to multiple output pixels
+ *
+ *
+ * \tparam ConvKernelDim Dimension of the convolutional kernel (assumed square)
+ * \tparam IFMChannels Number of Input Feature Maps
+ * \tparam Input_precision Number bits per pixel
+ * \tparam IFMDim Width and Heigth of the Input Feature Map (assumed square)
+ * \tparam OFMDim Width and Heigth of the Output Feature Map (assumed square)
+ * \tparam SIMD Number of input columns computed in parallel
+ * \tparam Stride Stride of the convolutional kernel
+ * \tparam MMV Number of pixels that have to be produced in parallel
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of time the function has to be repeatedly executed (e.g. number of images)
+ */
+template<unsigned int ConvKernelDim,
+ unsigned int IFMChannels,
+ unsigned int Input_precision,
+ unsigned int IFMDim,
+ unsigned int OFMDim,
+ unsigned int SIMD,
+ unsigned int Stride = 1,
+ unsigned int MMV=1 >
+void ConvolutionInputGenerator_MMV(
+ stream<ap_uint<SIMD*Input_precision> > & in,
+ stream<MultiChanData<MMV, SIMD*Input_precision> > & out,
+ const unsigned int numReps = 1) {
+ CASSERT_DATAFLOW(IFMChannels % SIMD == 0);
+ CASSERT_DATAFLOW(OFMDim % MMV == 0);
+ CASSERT_DATAFLOW(ConvKernelDim % Stride == 0);
+ CASSERT_DATAFLOW(MMV <= OFMDim);
+ constexpr unsigned int multiplying_factor = IFMChannels/SIMD;
+ constexpr unsigned int number_blocks = ConvKernelDim/Stride + 1 ;
+ ap_uint<SIMD*Input_precision> inputBuf[MMV][number_blocks][Stride * IFMDim * multiplying_factor];
+#pragma HLS DEPENDENCE variable=inputBuf inter false
+#pragma HLS DEPENDENCE variable=inputBuf intra false
+#pragma HLS ARRAY_PARTITION variable=inputBuf complete dim=1
+#pragma HLS ARRAY_PARTITION variable=inputBuf complete dim=2
+#pragma HLS RESOURCE variable inputBuf core=RAM_2P
+ constexpr unsigned int cycles_write_block = (OFMDim * ConvKernelDim * ConvKernelDim * multiplying_factor)/MMV;
+ constexpr unsigned int cycles_read_block = Stride * IFMDim * multiplying_factor;
+ constexpr unsigned int max_cycles = MAX(cycles_write_block,cycles_read_block);
+ const unsigned int baseIter = IFMDim * ConvKernelDim * multiplying_factor// Initial buffer
+ + OFMDim * MAX(cycles_write_block,cycles_read_block);
+ unsigned int counter_internal_block = 0;
+ unsigned int current_block_write = 0;
+ unsigned int next_block_write = 0;
+ unsigned int current_line = 0;
+ unsigned int read_block = 0;
+ unsigned int inp = 0, ofm_y = 0, ofm_x = 0, k_y = 0, k_x = 0, count_simd =0;
+#pragma HLS reset variable=inp
+ for (unsigned int count_image = 0; count_image < numReps; count_image++) {
+ for (unsigned int i = 0; i < baseIter; i++) {
+ #pragma HLS PIPELINE II=1
+ if (inp < IFMDim * ConvKernelDim*multiplying_factor) // Initial buffer of ConvKernelDim lines
+ {
+ ap_uint<SIMD*Input_precision> inElem;
+ inElem = in.read();
+ for(unsigned int v = 0; v < MMV; v++)
+ {
+#pragma HLS UNROLL
+ inputBuf[v][current_block_write][current_line] = inElem;
+ }
+ current_line++;
+ inp++;
+ if (current_line == Stride * IFMDim * multiplying_factor )
+ {
+ current_line = 0;
+ current_block_write++;
+ if (current_block_write == number_blocks)
+ current_block_write=0;
+ read_block++;
+ counter_internal_block = 0;
+ }
+ }
+ else
+ {
+ if (counter_internal_block < cycles_write_block-1) // We are writing output, MMV IFMChan per cycle
+ {
+ unsigned int current_block_read = (current_block_write + 1 + k_y / Stride);
+ if (current_block_read >= number_blocks)
+ current_block_read-= number_blocks;
+ unsigned int current_line_in_block = ((k_y%Stride) * IFMDim + ofm_x*Stride + k_x)*multiplying_factor + count_simd;
+ MultiChanData<MMV, SIMD*Input_precision> outElem;
+ // parallel read from all input buffers
+ for(unsigned int v = 0; v < MMV; v++) {
+#pragma HLS UNROLL
+ // each buffer's read addr is offset by its buffer index
+ ap_uint<SIMD*Input_precision> temp_value = inputBuf[v][current_block_read][(current_line_in_block + v*Stride*multiplying_factor)];
+ outElem.data[v] = temp_value;
+ }
+ out.write(outElem);
+ count_simd++;
+ if (count_simd == multiplying_factor) {
+ count_simd=0;
+ k_x++;
+ if (k_x == ConvKernelDim) {
+ k_x = 0;
+ k_y++;
+ if (k_y == ConvKernelDim) {
+ k_y = 0;
+ ofm_x += MMV;
+ if (ofm_x == OFMDim) {
+ ofm_x = 0;
+ ofm_y++;
+ if (ofm_y == OFMDim) {
+ ofm_y = 0;
+ inp = 0;
+ }
+ }
+ }
+ }
+ }
+ }
+ if ((counter_internal_block < cycles_read_block-1) && (read_block<IFMDim/Stride)) // In parallel we write in the buffer, in the current block write if we still need to
+ {
+ ap_uint<SIMD*Input_precision> inElem;
+ inElem = in.read();
+ for(unsigned int v = 0; v < MMV; v++) {
+#pragma HLS UNROLL
+ inputBuf[v][current_block_write][current_line] = inElem;
+#pragma AP dependence variable=inputBuf intra false
+#pragma AP dependence variable=inputBuf inter false
+ }
+
+ current_line++;
+ if (current_line == Stride * IFMDim * multiplying_factor) // We read the whole block, we change the next block in which we want to we
+ { // We filled up a block, let's not read until
+ current_line = 0;
+ read_block++;
+ current_block_write++;
+ if (current_block_write == number_blocks)
+ current_block_write=0;
+#pragma AP dependence variable=current_block_write intra false
+ }
+ }
+ counter_internal_block++; // = (counter_internal_block +1) % max_cycles;
+ if (counter_internal_block == (max_cycles-1))
+ {
+ counter_internal_block = 0;
+ }
+ }
+ } // End base_iter
+ read_block = 0;
+ } // End count_image
+} // End generator
+
+#endif
diff --git a/finn-hlslib/streamtools.h b/finn-hlslib/streamtools.h
new file mode 100755
index 000000000..ba2bec776
--- /dev/null
+++ b/finn-hlslib/streamtools.h
@@ -0,0 +1,704 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * @file stream-tools.h
+ *
+ * Library of templated HLS functions for BNN deployment.
+ * This file lists a set of convenience funtions used to adapt stream size,
+ * remove unnecessary streams (padding) and casting
+ *
+ ******************************************************************************/
+
+#ifndef STREAMTOOLS_H
+#define STREAMTOOLS_H
+
+
+/**
+ * \brief Stream limiter - limits the number of stream packets
+ *
+ * The block only let the first NumAllowed elements of a stream to pass through, the remainder
+ * (NumTotal-NumAllowed) are consumed from input but not re-emitted from the output.
+ * Useful to remove padding
+ *
+ * \tparam DataWidth Width, in number of bits, of the input and output stream
+ * \tparam NumAllowed Number of words to pass through
+ * \tparam NumTotal Total number of words (NumAllowed+NumDropped)
+ *
+ * \param in Input stream
+ * \param out Output stream
+ *
+ */
+template<unsigned int DataWidth,
+ unsigned int NumAllowed,
+ unsigned int NumTotal
+>
+void StreamLimiter(hls::stream<ap_uint<DataWidth> > & in,
+ hls::stream<ap_uint<DataWidth> > & out) {
+ CASSERT_DATAFLOW(NumTotal >= NumAllowed);
+ unsigned int numLeft = NumAllowed;
+ for (unsigned int i = 0; i < NumTotal; i++) {
+#pragma HLS PIPELINE II=1
+ ap_uint<DataWidth> e = in.read();
+ if (numLeft > 0) {
+ out.write(e);
+ numLeft--;
+ }
+ }
+}
+
+/**
+ * \brief Stream limiter batch - limits the number of stream packets multiple times
+ *
+ * The block only let the first NumAllowed elements of a stream to pass through, the remainder
+ * (NumTotal-NumAllowed) are consumed from input but not re-emitted from the output.
+ * Useful to remove padding on multiple images (numReps)
+ *
+ * \tparam DataWidth Width, in number of bits, of the input and output stream
+ * \tparam NumAllowed Number of words to pass through
+ * \tparam NumTotal Total number of words (NumAllowed+NumDropped)
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of times the StreamLimiter function has to be called
+ *
+ */
+template<unsigned int DataWidth,
+ unsigned int NumAllowed,
+ unsigned int NumTotal
+>
+void StreamLimiter_Batch(hls::stream<ap_uint<DataWidth> > & in,
+ hls::stream<ap_uint<DataWidth> > & out, unsigned int numReps) {
+ for (unsigned int rep = 0; rep < numReps; rep++) {
+ StreamLimiter<DataWidth, NumAllowed, NumTotal>(in, out);
+ }
+}
+
+/**
+ * \brief Stream Padding - Padds the input with zeroes for when the sliding window is
+ * centered on border pixels
+ *
+ * Used to add padding to the input with zeroes in case the sliding window is
+ * centered on border pixels
+ *
+ * \tparam ImgDim Size of the input feature map
+ * \tparam KernelDim Size of the sliding window
+ * \tparam Stride Stride of the sliding window
+ * \tparam NumChannels Amount of channels of the input feature map
+ * \tparam In_t Input datatype
+ * \tparam PaddingStyle Type of padding that will be applied
+ *
+ * \param in Input stream
+ * \param out Output stream
+ *
+ */
+template< unsigned int ImgDim,
+ unsigned int KernelDim,
+ unsigned int Stride,
+ unsigned int NumChannels,
+ typename In_t,
+ unsigned int PaddingStyle=2>
+void SameResize(stream<ap_uint<NumChannels* In_t::width> > &in,
+ stream<ap_uint<NumChannels* In_t::width> > &out){
+
+ // Number of "same" windows over the input data
+ constexpr unsigned int SameWindows = (ImgDim) / Stride + ((ImgDim % Stride) > 0);
+
+ // Number of elements to generate as output per dimension
+ constexpr unsigned int OutputDim = KernelDim + Stride * (SameWindows - 1);
+
+ // Padding
+ constexpr unsigned int Padding = OutputDim - ImgDim;
+
+ // Padding Up and Left
+ constexpr unsigned int PaddingUp = Padding/2 + ((PaddingStyle == 2) ? ((Padding % 2) > 0) : 0);
+ constexpr unsigned int PaddingLeft = Padding/2 + ((PaddingStyle == 2) ? ((Padding % 2) > 0) : 0);
+
+ // Padding Down and Right (might be 1 element more than up and left in case of odd padding)
+ constexpr unsigned int PaddingDown = Padding - PaddingUp;
+ constexpr unsigned int PaddingRight = Padding - PaddingLeft;
+
+ ap_uint<NumChannels* In_t::width> outData, inData;
+
+ for(unsigned int y = 0; y<OutputDim; y++){
+ for(unsigned int x=0; x < OutputDim; x++){
+#pragma HLS PIPELINE II=1
+
+ // Padding Rows
+ if(y < PaddingUp || y >= (OutputDim - PaddingDown)){
+ outData = 0;
+ }
+ // Padding Cols
+ else if(x < PaddingLeft || x >= (OutputDim - PaddingRight)){
+ outData = 0;
+ }
+ // No Padding
+ else{
+ inData = in.read();
+ outData = inData;
+ }
+
+ out.write(outData);
+ }
+ }
+}
+
+/**
+ * \brief Stream Padding - Padds the input of multiple frames with zeroes
+ * for when the sliding window is centered on border pixels
+ *
+ * Used to add padding with zeroes to multiple inputs in case the sliding window is
+ * centered on border pixels
+ *
+ * \tparam ImgDim Size of the input feature map
+ * \tparam KernelDim Size of the sliding window
+ * \tparam Stride Stride of the sliding window
+ * \tparam NumChannels Amount of channels of the input feature map
+ * \tparam In_t Input datatype
+ * \tparam PaddingStyle Type of padding that will be applied
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Amount of frames / images
+ *
+ */
+template< unsigned int ImgDim,
+ unsigned int KernelDim,
+ unsigned int Stride,
+ unsigned int NumChannels,
+ typename In_t,
+ unsigned int PaddingStyle=2>
+void SameResize_Batch(stream<ap_uint<NumChannels* In_t::width> > &in,
+ stream<ap_uint<NumChannels* In_t::width> > &out,
+ const unsigned int numReps) {
+ for (unsigned int rep = 0; rep < numReps; rep++) {
+ SameResize<ImgDim, KernelDim, Stride, NumChannels, In_t, PaddingStyle>(in, out);
+ }
+
+}
+
+
+/**
+ * \brief Stream cast - Casts the input stream to a different datatype (OutT)
+ *
+ * Used to upscale or downscale a stream, enabling loss of information for downscaling or
+ * 0 padding for upscaling
+ *
+ * \tparam InT Width, in number of bits, of the input and output stream
+ * \tparam OutT Number of words to pass through
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of times the StreamLimiter function has to be called
+ *
+ */
+template<typename InT, typename OutT>
+void StreamingCast(hls::stream<InT> & in, hls::stream<OutT> & out, unsigned int numReps) {
+ for(unsigned int i = 0; i < numReps; i++) {
+#pragma HLS PIPELINE II=1
+ out.write((OutT) in.read());
+ }
+}
+
+
+/**
+ * \brief Stream Data Width Converter - Converts the width of the input stream in the output stream
+ *
+ * Used to upscale or downscale a stream, without any loss of data in the procedure.
+ * For downscaling (InWidth > OutWidth), InWidth has to be a multiple of OutWidth.
+ * For upscaling (InWidth < OutWidth), OutWidth has to be a multiple of InWidth.
+ *
+ * \tparam InWidth Width, in number of bits, of the input stream
+ * \tparam OutWidth Width, in number of bits, of the output stream
+ * \tparam NumInWords Number of input words to process
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of times the function has to be called
+ *
+ */
+template<unsigned int InWidth,
+ unsigned int OutWidth,
+ unsigned int NumInWords
+>
+void StreamingDataWidthConverter_Batch(hls::stream<ap_uint<InWidth> > & in,
+ hls::stream<ap_uint<OutWidth> > & out, const unsigned int numReps) {
+ if (InWidth > OutWidth) {
+ // emit multiple output words per input word read
+ CASSERT_DATAFLOW(InWidth % OutWidth == 0);
+ const unsigned int outPerIn = InWidth / OutWidth;
+ const unsigned int totalIters = NumInWords * outPerIn * numReps;
+ unsigned int o = 0;
+ ap_uint<InWidth> ei = 0;
+ for (unsigned int t = 0; t < totalIters; t++) {
+#pragma HLS PIPELINE II=1
+ // read new input word if current out count is zero
+ if (o == 0) {
+ ei = in.read();
+ }
+ // pick output word from the rightmost position
+ ap_uint<OutWidth> eo = ei(OutWidth - 1, 0);
+ out.write(eo);
+ // shift input to get new output word for next iteration
+ ei = ei >> OutWidth;
+ // increment written output count
+ o++;
+ // wraparound indices to recreate the nested loop structure
+ if (o == outPerIn) {
+ o = 0;
+ }
+ }
+ } else if (InWidth == OutWidth) {
+ // straight-through copy
+ for (unsigned int i = 0; i < NumInWords * numReps; i++) {
+#pragma HLS PIPELINE II=1
+ ap_uint<InWidth> e = in.read();
+ out.write(e);
+ }
+ } else { // InWidth < OutWidth
+ // read multiple input words per output word emitted
+ CASSERT_DATAFLOW(OutWidth % InWidth == 0);
+ const unsigned int inPerOut = OutWidth / InWidth;
+ const unsigned int totalIters = NumInWords * numReps;
+ unsigned int i = 0;
+ ap_uint<OutWidth> eo = 0;
+ for (unsigned int t = 0; t < totalIters; t++) {
+#pragma HLS PIPELINE II=1
+ // read input and shift into output buffer
+ ap_uint<InWidth> ei = in.read();
+ eo = eo >> InWidth;
+ eo(OutWidth - 1, OutWidth - InWidth) = ei;
+ // increment read input count
+ i++;
+ // wraparound logic to recreate nested loop functionality
+ if (i == inPerOut) {
+ i = 0;
+ out.write(eo);
+ }
+ }
+ }
+}
+
+/**
+ * \brief Stream Duplicator - Reads in a stream and writes the data into two identical streams
+ *
+ * Used to generate the inputs to the bypass and convolutional branches in Resnet-50
+ *
+ * \tparam DataWidth Width, in number of bits, of the streams
+ * \tparam NumTotal Total number of words in the input stream
+ *
+ * \param in Input stream
+ * \param out1 Output stream I
+ * \param out2 Output stream II
+ *
+ */
+template<unsigned int DataWidth,
+ unsigned int NumTotal
+>
+void DuplicateStreams(stream<ap_uint<DataWidth> > & in, stream<ap_uint<DataWidth> > & out1,
+ stream<ap_uint<DataWidth> > & out2) {
+
+ for (unsigned int i = 0; i < NumTotal; i++) {
+#pragma HLS PIPELINE II=1
+ ap_uint<DataWidth> e = in.read();
+
+ out1.write(e);
+ out2.write(e);
+ }
+}
+
+/**
+ * \brief Batch Stream Duplicator - Reads in a stream multiple times and writes the data into two identical streams
+ *
+ * Used to generate the inputs to the bypass and convolutional branches in Resnet-50 when dealing with multiple 'frames'
+ *
+ * \tparam DataWidth Width, in number of bits, of the streams
+ * \tparam NumTotal Total number of words in the input stream
+ *
+ * \param in Input stream
+ * \param out1 Output stream I
+ * \param out2 Output stream II
+ * \param numReps Number of frames / images
+ *
+ */
+template<unsigned int DataWidth,
+ unsigned int NumTotal
+>
+void DuplicateStreams_Batch(stream<ap_uint<DataWidth> > & in, stream<ap_uint<DataWidth> > & out1,
+ stream<ap_uint<DataWidth> > & out2, const unsigned int numReps) {
+ for (unsigned int image = 0; image < numReps; image++) {
+ DuplicateStreams<DataWidth, NumTotal>(in, out1, out2);
+ }
+}
+
+/**
+ * \brief Element-Wise Addition - Reads in data elements from two streams and writes the sum of these elements to an output
+ *
+ * \tparam NumChannels Amount of channels of the streams
+ * \tparam In1_t First operand datatype
+ * \tparam In2_t Second operand datatype
+ * \tparam Out_t Datatype of the accumulation output
+ * \tparam NumTotal Total number of words in the input streams
+ * \tparam offset Offset value for the accumulation
+ *
+ * \param in1 Input stream I
+ * \param in2 Input stream II
+ * \param out Output stream
+ *
+ */
+
+template <unsigned int NumChannels,
+ typename In1_t,
+ typename In2_t,
+ typename Out_t,
+ unsigned int NumTotal,
+ int offset = 0>
+void AddStreams(stream<ap_uint<NumChannels * In1_t::width>> &in1, stream<ap_uint<NumChannels * In2_t::width>> &in2,
+ stream<ap_uint<NumChannels * Out_t::width>> &out) {
+
+ for (unsigned int i = 0; i < NumTotal; i++) {
+#pragma HLS PIPELINE II = 1
+ ap_uint<NumChannels * In1_t::width> e1 = in1.read();
+ ap_uint<NumChannels * In2_t::width> e2 = in2.read();
+ ap_uint<NumChannels * Out_t::width> e;
+ for (unsigned int j = 0; j < NumChannels; j++) {
+#pragma HLS UNROLL
+ In1_t op1 = e1((j + 1) * In1_t::width - 1, j * In1_t::width);
+ In2_t op2 = e2((j + 1) * In2_t::width - 1, j * In2_t::width);
+ Out_t sum = op1 + op2 + offset;
+ e((j + 1) * Out_t::width - 1, j * Out_t::width) = sum;
+ }
+ out.write(e);
+ }
+}
+
+
+/**
+ * \brief
+ *
+ * Used to implement point-wise addition in Resnet-50 for multiple images
+ *
+ * \tparam NumChannels Amount of channels of the streams
+ * \tparam In1_t First operand datatype
+ * \tparam In2_t Second operand datatype
+ * \tparam Out_t Datatype of the accumulation output
+ * \tparam NumTotal Total number of words in the input streams
+ * \tparam offset Offset value for the accumulation
+ *
+ * \param in1 Input stream I
+ * \param in2 Input stream II
+ * \param out Output stream
+ * \param numReps Number of frames / images
+ *
+ */
+template <unsigned int NumChannels,
+ typename In1_t,
+ typename In2_t,
+ typename Out_t,
+ unsigned int NumTotal,
+ int offset = 0>
+void AddStreams_Batch(stream<ap_uint<NumChannels * In1_t::width>> &in1, stream<ap_uint<NumChannels * In2_t::width>> &in2,
+ stream<ap_uint<NumChannels * Out_t::width>> &out, const unsigned int numReps) {
+ for (unsigned int image = 0; image < numReps; image++) {
+ AddStreams<NumChannels, In1_t, In2_t, Out_t, NumTotal, offset>(in1, in2, out);
+ }
+}
+
+/**
+ * \brief Addition Layer - Reads in two streams and writes the sum of these streams to an output
+ *
+ * Used to merge the outputs of the bypass and convolutional branches in Resnet-50
+ *
+ * \tparam NumChannels Amount of channels of the streams
+ * \tparam In1_t First operand datatype
+ * \tparam In2_t Second operand datatype
+ * \tparam Out_t Datatype of the accumulation output * \tparam NumTotal Total number of words in the input streams
+ * \tparam PECount Amount of processing elements working in parallel
+ * \tparam offset Offset value for the accumulation
+ *
+ * \param in1 Input stream I
+ * \param in2 Input stream II
+ * \param out Output stream
+ * \param numReps Number of frames / images
+ *
+ */
+template <unsigned int NumChannels,
+ typename In1_t,
+ typename In2_t,
+ typename Out_t,
+ unsigned int NumTotal,
+ unsigned int PECount,
+ int offset = 0>
+void AddStreamsLayer_Batch(stream<ap_uint<NumChannels * In1_t::width>> &in1, stream<ap_uint<NumChannels * In2_t::width>> &in2,
+ stream<ap_uint<NumChannels * Out_t::width>> &out, const unsigned int numReps) {
+#pragma HLS INLINE
+ CASSERT_DATAFLOW(NumChannels % PECount == 0);
+ stream<ap_uint<PECount * In1_t::width>> in_folded1;
+ stream<ap_uint<PECount * In2_t::width>> in_folded2;
+ stream<ap_uint<PECount * Out_t::width>> out_folded;
+ StreamingDataWidthConverter_Batch<NumChannels * In1_t::width, PECount * In1_t::width, NumTotal>(in1, in_folded1, numReps);
+ StreamingDataWidthConverter_Batch<NumChannels * In2_t::width, PECount * In2_t::width, NumTotal>(in2, in_folded2, numReps);
+ AddStreams_Batch<PECount, In1_t, In2_t, Out_t, NumTotal *(NumChannels / PECount),offset>(in_folded1, in_folded2, out_folded, numReps);
+ StreamingDataWidthConverter_Batch<PECount * Out_t::width, NumChannels * Out_t::width, NumTotal *(NumChannels / PECount)>(out_folded, out, numReps);
+}
+
+
+/**
+ * \brief Stream Multi Chan Data Width Converter - Converts the width of the input stream in the output stream, working on multiple parallel streams
+ *
+ * Used to upscale or downscale a stream, without any loss of data in the procedure.
+ * For downscaling (InWidth > OutWidth), InWidth has to be a multiple of OutWidth.
+ * For upscaling (InWidth < OutWidth), OutWidth has to be a multiple of InWidth.
+ * This version works on the MMV structure, with multiple parallel streams
+ *
+ * \tparam InWidth Width, in number of bits, of the input stream
+ * \tparam OutWidth Width, in number of bits, of the output stream
+ * \tparam NumInWords Number of input words to process
+ * \tparam NumVecs Number of parallel vectors MMV
+ *
+ * \param in Input stream
+ * \param out Output stream
+ * \param numReps Number of times the function has to be called
+ *
+ */
+template<unsigned int InWidth, // width of input stream
+ unsigned int OutWidth, // width of output stream
+ unsigned int NumInWords, // number of input words to process
+ unsigned int NumVecs
+>
+void MultiChanDataWidthConverter_Batch(
+ stream<MultiChanData<NumVecs, InWidth> > & in,
+ stream<MultiChanData<NumVecs, OutWidth> > & out,
+ const unsigned int numReps) {
+ if (InWidth > OutWidth) {
+ // emit multiple output words per input word read
+ CASSERT_DATAFLOW((InWidth % OutWidth) == 0);
+ const unsigned int outPerIn = InWidth / OutWidth;
+ const unsigned int totalIters = NumInWords * outPerIn * numReps;
+ unsigned int o = 0;
+ MultiChanData<NumVecs, InWidth> ei;
+ for (unsigned int t = 0; t < totalIters; t++) {
+#pragma HLS PIPELINE II=1
+ // read new input word if current out count is zero
+ if (o == 0)
+ ei = in.read();
+ // pick output word from the rightmost position
+ MultiChanData<NumVecs, OutWidth> eo;
+ for(unsigned int v = 0; v < NumVecs; v++) {
+#pragma HLS UNROLL
+ eo.data[v] = (ei.data[v])(OutWidth - 1, 0);
+ // shift input to get new output word for next iteration
+ ei.data[v] = ei.data[v] >> OutWidth;
+ }
+ out.write(eo);
+ // increment written output count
+ o++;
+ // wraparound indices to recreate the nested loop structure
+ if (o == outPerIn) {
+ o = 0;
+ }
+ }
+ } else if (InWidth == OutWidth) {
+ // straight-through copy
+ for (unsigned int i = 0; i < NumInWords * numReps; i++) {
+#pragma HLS PIPELINE II=1
+ MultiChanData<NumVecs, InWidth> e = in.read();
+ MultiChanData<NumVecs, OutWidth> eo;
+ // we don't support typecasting between templated types, so explicitly
+ // transfer vector-by-vector here
+ for(unsigned int v=0; v < NumVecs; v++) {
+#pragma HLS UNROLL
+ eo.data[v] = e.data[v];
+ }
+ out.write(eo);
+ }
+ } else { // InWidth < OutWidth
+ // read multiple input words per output word emitted
+ CASSERT_DATAFLOW((OutWidth % InWidth) == 0);
+ const unsigned int inPerOut = OutWidth / InWidth;
+ const unsigned int totalIters = NumInWords * numReps;
+ unsigned int i = 0;
+ MultiChanData<NumVecs, OutWidth> eo;
+ for (unsigned int t = 0; t < totalIters; t++) {
+#pragma HLS PIPELINE II=1
+ // read input and shift into output buffer
+ MultiChanData<NumVecs, InWidth> ei = in.read();
+ for(unsigned int v = 0; v < NumVecs; v++) {
+#pragma HLS UNROLL
+ eo.data[v] = eo.data[v] >> InWidth;
+ (eo.data[v])(OutWidth - 1, OutWidth - InWidth) = ei.data[v];
+ }
+ // increment read input count
+ i++;
+ // wraparound logic to recreate nested loop functionality
+ if (i == inPerOut) {
+ i = 0;
+ out.write(eo);
+ }
+ }
+ }
+}
+
+
+/**
+ * \brief Flatten Multi Chan Data - Converts the parallel input stream in a flatten output stream
+ *
+ * Used to pach a flattened stream into a structure with multiple parallel streams
+ *
+ * \tparam NumChannels Number of channels flattened in the input stream
+ * \tparam DataWidth Width, in number of bits, of each stream
+ *
+ * \param in Input parallel stream
+ * \param out Output stream
+ * \param numReps Number of times the function has to be called
+ *
+ */
+template <unsigned int NumChannels, unsigned int DataWidth>
+void FlattenMultiChanData(
+ stream<MultiChanData<NumChannels, DataWidth> > & in,
+ stream<ap_uint<NumChannels*DataWidth> > & out,
+ const unsigned int numReps
+) {
+ for(unsigned int r = 0; r < numReps; r++) {
+#pragma HLS PIPELINE II=1
+ MultiChanData<NumChannels, DataWidth> e = in.read();
+ ap_uint<NumChannels*DataWidth> o = 0;
+ for(unsigned int v = 0; v < NumChannels; v++) {
+#pragma HLS UNROLL
+ o(DataWidth*(v+1)-1, DataWidth*v) = e.data[v];
+ }
+ out.write(o);
+ }
+}
+
+/**
+ * \brief Pack Multi Chan Data - Converts the flatten input stream into a parallel output stream
+ *
+ * Used to pach a flattened stream into a structure with multiple parallel streams
+ *
+ * \tparam NumChannels Number of channels flattened in the input stream
+ * \tparam DataWidth Width, in number of bits, of each stream
+ *
+ * \param in Input stream
+ * \param out Output parallel stream
+ * \param numReps Number of times the function has to be called
+ *
+ */
+template <unsigned int NumChannels, unsigned int DataWidth>
+void PackMultiChanData(
+ stream<ap_uint<NumChannels*DataWidth> > & in,
+ stream<MultiChanData<NumChannels, DataWidth> > & out,
+ const unsigned int numReps
+) {
+ for(unsigned int r = 0; r < numReps; r++) {
+#pragma HLS PIPELINE II=1
+ ap_uint<NumChannels*DataWidth> e = in.read();
+ MultiChanData<NumChannels, DataWidth> o;
+ for(unsigned int v = 0; v < NumChannels; v++) {
+#pragma HLS UNROLL
+ o.data[v] = e(DataWidth*(v+1)-1, DataWidth*v);
+ }
+ out.write(o);
+ }
+}
+
+
+template<unsigned IW, unsigned OW, unsigned N>
+ class WidthAdjustedInputStream {
+ hls::stream<ap_uint<OW>> m_target;
+
+ public:
+ WidthAdjustedInputStream(hls::stream<ap_uint<IW> >& source, unsigned const reps) {
+ StreamingDataWidthConverter_Batch<IW, OW, N>(source, m_target, reps);
+ }
+ ~WidthAdjustedInputStream() {}
+
+ public:
+ operator hls::stream<ap_uint<OW> >&() {
+ return m_target;
+ }
+};
+template<unsigned W, unsigned N>
+ class WidthAdjustedInputStream<W, W, N> {
+
+ hls::stream<ap_uint<W>> &m_source;
+
+ public:
+ WidthAdjustedInputStream(hls::stream<ap_uint<W> >& source, unsigned const reps) : m_source(source) {}
+ ~WidthAdjustedInputStream() {}
+
+ public:
+ operator hls::stream<ap_uint<W> >&() {
+ return m_source;
+ }
+};
+
+
+template<unsigned IW, unsigned OW, unsigned N>
+class WidthAdjustedOutputStream {
+ hls::stream<ap_uint<IW>> m_buffer;
+ hls::stream<ap_uint<OW>> &m_target;
+ unsigned const m_reps;
+
+ public:
+ WidthAdjustedOutputStream(hls::stream<ap_uint<OW> >& target, unsigned const reps) : m_target(target), m_reps(reps) {}
+ ~WidthAdjustedOutputStream() {
+ StreamingDataWidthConverter_Batch<IW, OW, N>(m_buffer, m_target, m_reps);
+ }
+
+ public:
+ operator hls::stream<ap_uint<IW> >&() {
+ return m_buffer;
+ }
+};
+template<unsigned W, unsigned N>
+ class WidthAdjustedOutputStream<W, W, N> {
+ hls::stream<ap_uint<W>> &m_target;
+
+ public:
+ WidthAdjustedOutputStream(hls::stream<ap_uint<W> >& target, unsigned const reps)
+ : m_target(target) {}
+ ~WidthAdjustedOutputStream() {}
+
+ public:
+ operator hls::stream<ap_uint<W> >&() {
+ return m_target;
+ }
+};
+
+#endif
diff --git a/finn-hlslib/tb/.gitignore b/finn-hlslib/tb/.gitignore
new file mode 100755
index 000000000..1c64c8e33
--- /dev/null
+++ b/finn-hlslib/tb/.gitignore
@@ -0,0 +1,5 @@
+# Ignore synthesis files and logs
+/hls-syn/*
+/vivado_hls.log
+/config.h
+/memdata.h
diff --git a/finn-hlslib/tb/README.md b/finn-hlslib/tb/README.md
new file mode 100755
index 000000000..d3ece08ec
--- /dev/null
+++ b/finn-hlslib/tb/README.md
@@ -0,0 +1,7 @@
+# HLS Library Unit Tests
+
+## Instructions
+1. Generate `config.h` and `memdata.h` by running gen_weights.py (modify it if you need non-default precision for weights/activations)
+1. Set the FINN_HLS_ROOT to the root folder of the repo, e.g. `setenv FINN_HLS_ROOT <path to repo root>`
+1. Run a unit test with Vivado HLS, e.g. `vivado_hls <testname>.tcl`
+
diff --git a/finn-hlslib/tb/add_config.h b/finn-hlslib/tb/add_config.h
new file mode 100755
index 000000000..79f576266
--- /dev/null
+++ b/finn-hlslib/tb/add_config.h
@@ -0,0 +1,7 @@
+#define NUM_CHANNELS 32
+#define INPUT_WIDTH 4
+#define OUTPUT_WIDTH 4
+#define NUM_WORDS 4
+#define NUM_REPEAT 1
+#define OFFSET 2
+#define SAT 1
diff --git a/finn-hlslib/tb/add_tb.cpp b/finn-hlslib/tb/add_tb.cpp
new file mode 100755
index 000000000..cba6a0b9d
--- /dev/null
+++ b/finn-hlslib/tb/add_tb.cpp
@@ -0,0 +1,65 @@
+#include <iostream>
+#include <cmath>
+#include <ctime>
+#include <cstring>
+#include <hls_stream.h>
+#include <cstdlib>
+#define AP_INT_MAX_W 16384
+#include "ap_int.h"
+#include "weights.hpp"
+#include "bnn-library.h"
+
+#include "add_config.h"
+
+#include "activations.hpp"
+#include "interpret.hpp"
+
+using namespace hls;
+using namespace std;
+
+void Testbench_add(stream<ap_uint<NUM_CHANNELS * INPUT_WIDTH>> &in1, stream<ap_uint<NUM_CHANNELS * INPUT_WIDTH>> &in2,
+ stream<ap_uint<NUM_CHANNELS * OUTPUT_WIDTH>> &out, unsigned int numReps);
+
+void sw_add(ap_uint<NUM_CHANNELS * INPUT_WIDTH> val1, ap_uint<NUM_CHANNELS * INPUT_WIDTH> val2, ap_uint<NUM_CHANNELS * OUTPUT_WIDTH> & out) {
+ for (int i = 0; i < NUM_CHANNELS; i++) {
+ ap_uint<INPUT_WIDTH> op1 = val1((i+1)*INPUT_WIDTH-1, i*INPUT_WIDTH);
+ ap_uint<INPUT_WIDTH> op2 = val2((i+1)*INPUT_WIDTH-1, i*INPUT_WIDTH);
+#ifdef SAT
+ ap_ufixed<OUTPUT_WIDTH, OUTPUT_WIDTH, AP_RND, AP_SAT> sum = op1 + op2 + OFFSET;
+#else
+ ap_ufixed<OUTPUT_WIDTH, OUTPUT_WIDTH, AP_TRN> sum = op1 + op2 + OFFSET;
+#endif
+ out((i+1)*OUTPUT_WIDTH-1,i*OUTPUT_WIDTH) = sum;
+
+ }
+}
+
+int main()
+{
+ stream<ap_uint<NUM_CHANNELS * INPUT_WIDTH> > input_stream1("input_stream1");
+ stream<ap_uint<NUM_CHANNELS * INPUT_WIDTH>> input_stream2("input_stream2");
+ stream<ap_uint<NUM_CHANNELS * OUTPUT_WIDTH> > output_stream("output_stream");
+ static ap_uint<NUM_CHANNELS * OUTPUT_WIDTH> expected[NUM_REPEAT*NUM_WORDS];
+ unsigned int count_out = 0;
+ unsigned int count_in = 0;
+ for (unsigned int counter = 0; counter < NUM_REPEAT*NUM_WORDS; counter++) {
+ ap_uint<NUM_CHANNELS * INPUT_WIDTH> value = (ap_uint<NUM_CHANNELS * INPUT_WIDTH>) counter;
+ sw_add(value, value, expected[counter]);
+ input_stream1.write(value);
+ input_stream2.write(value);
+ }
+
+ Testbench_add(input_stream1, input_stream2, output_stream, NUM_REPEAT);
+ for (unsigned int counter = 0; counter < NUM_REPEAT*NUM_WORDS; counter++)
+ {
+ ap_uint<NUM_CHANNELS * OUTPUT_WIDTH> value = output_stream.read();
+ if(value!= expected[counter])
+ {
+ cout << "ERROR with counter " << counter << std::hex << " expected " << expected[counter] << " value " << value << std::dec << endl;
+ return(1);
+ }
+ }
+
+}
+
+
diff --git a/finn-hlslib/tb/add_top.cpp b/finn-hlslib/tb/add_top.cpp
new file mode 100755
index 000000000..d6159e6d4
--- /dev/null
+++ b/finn-hlslib/tb/add_top.cpp
@@ -0,0 +1,15 @@
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+
+
+#include "add_config.h"
+
+void Testbench_add(stream<ap_uint<NUM_CHANNELS * INPUT_WIDTH> > & in1, stream<ap_uint<NUM_CHANNELS * INPUT_WIDTH> > & in2, stream<ap_uint<NUM_CHANNELS * OUTPUT_WIDTH> > & out, const unsigned int numReps){
+
+ if (SAT == 0)
+ AddStreams_Batch<NUM_CHANNELS, ap_uint<INPUT_WIDTH>, ap_uint<INPUT_WIDTH>, ap_ufixed<OUTPUT_WIDTH, OUTPUT_WIDTH, AP_TRN>, NUM_WORDS, OFFSET>(in1, in2, out, numReps);
+ else
+ AddStreams_Batch<NUM_CHANNELS, ap_uint<INPUT_WIDTH>, ap_uint<INPUT_WIDTH>, ap_ufixed<OUTPUT_WIDTH, OUTPUT_WIDTH, AP_RND, AP_SAT>, NUM_WORDS, OFFSET>(in1, in2, out, numReps);
+}
diff --git a/finn-hlslib/tb/cnvlayer_batch_unit.cpp b/finn-hlslib/tb/cnvlayer_batch_unit.cpp
new file mode 100755
index 000000000..b53a13802
--- /dev/null
+++ b/finn-hlslib/tb/cnvlayer_batch_unit.cpp
@@ -0,0 +1,41 @@
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+
+#include "activations.hpp"
+#include "weights.hpp"
+#include "activations.hpp"
+#include "interpret.hpp"
+#include "mvau.hpp"
+#include "conv.hpp"
+#include "memdata.h"
+#include "config.h"
+
+// adheres to new config.h
+#define OFD OFMDim1
+#define OFC OFM_Channels1
+#define IFD IFMDim1
+#define IFC IFM_Channels1
+#define TILES TILE1
+#define SIMD SIMD1
+#define PE PE1
+#define WP WEIGHT_PRECISION
+#define AP ACTIVATION_PRECISION
+#define K KERNEL_DIM
+
+typedef stream<ap_uint<SIMD * PE * WP>> paramS;
+typedef stream<ap_uint<IFC * INPUT_PRECISION>> dataInS;
+typedef stream<ap_uint<OFC * ACTIVATION_PRECISION>> dataOutS;
+
+void ConvLayer_B(dataInS &in, dataOutS &out, int const numReps) {
+#pragma HLS INTERFACE s_axilite port=return bundle=control
+#pragma HLS INTERFACE s_axilite port=numReps bundle=control
+#pragma HLS INTERFACE axis port=&in bundle=hostmem
+#pragma HLS INTERFACE axis port=&out bundle=hostmem
+#pragma HLS DATAFLOW
+
+ConvLayer_Batch<K, IFC, IFD, OFC, OFD, SIMD, PE, Slice<ap_uint<INPUT_PRECISION> >, Slice<ap_int<AP> >, Identity>(
+ in, out, PARAM::weights, PassThroughActivation<ap_uint<AP>>(), numReps, ap_resource_dsp());
+
+}
diff --git a/finn-hlslib/tb/conv.hpp b/finn-hlslib/tb/conv.hpp
new file mode 100755
index 000000000..19e618f02
--- /dev/null
+++ b/finn-hlslib/tb/conv.hpp
@@ -0,0 +1,91 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file conv.hpp
+ *
+ * C++ Implementation of a convolution, used for testbench
+ *
+ *****************************************************************************/
+#ifndef CONV_TB_H
+#define CONV_TB_H
+
+template<int MAX_IMAGE,
+ int IFMDim,
+ int OFMDim,
+ int IFMCh,
+ int OFMCh,
+ typename TI,
+ typename TO,
+ typename TW>
+ void conv_1x1(TI const img[MAX_IMAGE][IFMDim][IFMDim][IFMCh], TW const weights[OFMCh][IFMCh], TO out[MAX_IMAGE][OFMDim][OFMDim][OFMCh]){
+ constexpr int stride= (OFMDim==1)? IFMDim:(IFMDim - 1)/(OFMDim - 1);
+ for(int n=0;n<MAX_IMAGE;n++)
+ for(int x=0;x<OFMDim;x++)
+ for(int y=0;y<OFMDim;y++)
+ for(int h=0;h<OFMCh;h++){
+ TO tmp = 0;
+ for(int w=0;w<IFMCh;w++)
+ tmp+=img[n][x*stride][y*stride][w] * weights[h][w];
+ out[n][x][y][h] = tmp;
+
+ }
+ }
+
+template<int MAX_IMAGE,
+ int IFMDim,
+ int OFMDim,
+ int IFMCh,
+ int OFMCh,
+ int kernel,
+ int stride,
+ typename TI,
+ typename TO,
+ typename TW>
+ void conv(TI const img[MAX_IMAGE][IFMDim*IFMDim][IFMCh], TW const weights[OFMCh][kernel][kernel][IFMCh], TO out[MAX_IMAGE][OFMDim][OFMDim][OFMCh]){
+ for(int n=0;n<MAX_IMAGE;n++)
+ for(int x=0;x<OFMDim;x++)
+ for(int y=0;y<OFMDim;y++)
+ for(int h=0;h<OFMCh;h++){
+ TO tmp = 0;
+ for (int ky=0;ky<kernel;ky++)
+ for (int kx=0;kx<kernel;kx++)
+ for(int w=0;w<IFMCh;w++){
+ tmp+=img[n][(y*stride+ky)*IFMDim+x*stride+kx][w] * weights[h][kx][ky][w];
+ }
+ out[n][x][y][h] = tmp;
+ }
+ }
+
+#endif
diff --git a/finn-hlslib/tb/conv3_tb.cpp b/finn-hlslib/tb/conv3_tb.cpp
new file mode 100755
index 000000000..6657e60f0
--- /dev/null
+++ b/finn-hlslib/tb/conv3_tb.cpp
@@ -0,0 +1,166 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file conv3_tb.cpp
+ *
+ * Testbench for the convolution HLS block
+ *
+ *****************************************************************************/
+#include <iostream>
+#include <cmath>
+#include <ctime>
+#include <cstring>
+#include <hls_stream.h>
+#include <cstdlib>
+#define AP_INT_MAX_W 16384
+#include "ap_int.h"
+#include "weights.hpp"
+#include "bnn-library.h"
+#include "memdata.h"
+#include "config.h"
+#include "activations.hpp"
+#include "weights.hpp"
+#include "activations.hpp"
+#include "interpret.hpp"
+#include "mvau.hpp"
+#include "conv.hpp"
+using namespace hls;
+using namespace std;
+
+#define MAX_IMAGES 1
+void Testbench_conv(stream<ap_uint<IFM_Channels1*INPUT_PRECISION> > & in, stream<ap_uint<OFM_Channels1*ACTIVATION_PRECISION> > & out, unsigned int numReps);
+
+int main()
+{
+ static ap_uint<INPUT_PRECISION> IMAGE[MAX_IMAGES][IFMDim1*IFMDim1][IFM_Channels1];
+ static ap_int<ACTIVATION_PRECISION> TEST[MAX_IMAGES][OFMDim1][OFMDim1][OFM_Channels1];
+ stream<ap_uint<IFM_Channels1*INPUT_PRECISION> > input_stream("input_stream");
+ stream<ap_uint<OFM_Channels1*ACTIVATION_PRECISION> > output_stream("output_stream");
+ unsigned int counter = 0;
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < IFMDim1; oy++) {
+ for (unsigned int ox = 0; ox < IFMDim1; ox++) {
+ ap_uint<INPUT_PRECISION*IFM_Channels1> input_channel = 0;
+ for(unsigned int channel = 0; channel < IFM_Channels1; channel++)
+ {
+ ap_uint<INPUT_PRECISION> input = (ap_uint<INPUT_PRECISION>)(counter);
+ IMAGE[n_image][oy*IFMDim1+ox][channel]= input;
+ input_channel = input_channel >> INPUT_PRECISION;
+ input_channel(IFM_Channels1*INPUT_PRECISION-1,(IFM_Channels1-1)*INPUT_PRECISION)=input;
+
+ counter++;
+ }
+ input_stream.write(input_channel);
+ }
+ }
+ }
+ static ap_int<4> W1[OFM_Channels1][KERNEL_DIM][KERNEL_DIM][IFM_Channels1];
+ // initialize the weights
+ constexpr int TX = (IFM_Channels1*KERNEL_DIM*KERNEL_DIM) / SIMD1;
+ constexpr int TY = OFM_Channels1 / PE1;
+ unsigned int kx=0;
+ unsigned int ky=0;
+ unsigned int chan_count=0;
+ unsigned int out_chan_count=0;
+ for (unsigned int oy = 0; oy < TY; oy++) {
+ for (unsigned int ox = 0; ox <TX; ox++) {
+ for(int pe=0;pe <PE1;pe++){
+ for(int simd=0;simd<SIMD1;simd++){
+ std::cout << " Value: " << PARAM::weights.weights(oy*TX + ox)[pe][simd] <<std::endl;
+ std::cout << "W1[" << out_chan_count << "][" << kx << "][" << ky << "][" << chan_count << "]"<< std::endl;
+ W1[out_chan_count][kx][ky][chan_count] = PARAM::weights.weights(oy*TX + ox)[pe][simd];
+ kx++;
+ if (kx==KERNEL_DIM){
+ kx=0;
+ ky++;
+ if (ky==KERNEL_DIM){
+ ky=0;
+ chan_count++;
+ if (chan_count==IFM_Channels1){
+ chan_count=0;
+ out_chan_count++;
+ if (out_chan_count==OFM_Channels1){
+ out_chan_count=0;
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+ conv<MAX_IMAGES,IFMDim1,OFMDim1,IFM_Channels1,OFM_Channels1, KERNEL_DIM, 1, ap_uint<INPUT_PRECISION> >(IMAGE, W1, TEST);
+ Testbench_conv(input_stream, output_stream, MAX_IMAGES);
+ int err_counter = 0, err_perimage=0;
+ ap_int<ACTIVATION_PRECISION> out_chan;
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < OFMDim1; oy++) {
+ for (unsigned int ox = 0; ox < OFMDim1; ox++) {
+ for(int e=0;e<1;e++){
+ ap_uint<OFM_Channels1*ACTIVATION_PRECISION> outElem = output_stream.read();
+ for(unsigned int channel = 0; channel < OFM_Channels1; channel++){
+ ap_int<ACTIVATION_PRECISION> EXP = TEST[n_image][ox][oy][channel + e * OFM_Channels1];
+ out_chan(ACTIVATION_PRECISION-1,0) = outElem((channel + 1)*ACTIVATION_PRECISION-1,channel*ACTIVATION_PRECISION);
+
+ if (EXP != out_chan){
+ std::cout << "ERROR: Expected["<<oy <<"]["<<ox<<"]["<<channel<<"]=" << EXP << " actual " << out_chan << std::endl;
+ //return 1;
+ err_counter ++;
+ err_perimage++;
+ //if(err_counter>10)
+ //return 1;
+ }
+ }
+ }
+ }
+ }
+ if(err_perimage == 0){
+ std::cout << "Image # " << n_image << " passed the testing."<< std::endl;
+ }
+ else{
+ err_perimage=0;
+ std::cout << "Image # " << n_image << " failed the testing."<< std::endl;
+ }
+ }
+ if(err_counter == 0){
+ return 0;
+ }
+ else{
+ return 1;
+ }
+
+}
+
+
diff --git a/finn-hlslib/tb/conv3mmv_tb.cpp b/finn-hlslib/tb/conv3mmv_tb.cpp
new file mode 100755
index 000000000..4dee0da2c
--- /dev/null
+++ b/finn-hlslib/tb/conv3mmv_tb.cpp
@@ -0,0 +1,164 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file conv3mmv_tb.cpp
+ *
+ * Testbench for the convolution HLS block
+ *
+ *****************************************************************************/
+#include <iostream>
+#include <cmath>
+#include <ctime>
+#include <cstring>
+#include <hls_stream.h>
+#include <cstdlib>
+#define AP_INT_MAX_W 16384
+#include "ap_int.h"
+#include "weights.hpp"
+#include "bnn-library.h"
+#include "memdata.h"
+#include "config.h"
+#include "activations.hpp"
+#include "weights.hpp"
+#include "activations.hpp"
+#include "interpret.hpp"
+#include "mvau.hpp"
+#include "conv.hpp"
+using namespace hls;
+using namespace std;
+
+#define MAX_IMAGES 1
+void Testbench_convmmv(stream<ap_uint<IFM_Channels1*INPUT_PRECISION> > & in, stream<ap_uint<OFM_Channels1*ACTIVATION_PRECISION> > & out, unsigned int numReps);
+
+int main()
+{
+ static ap_uint<INPUT_PRECISION> IMAGE[MAX_IMAGES][IFMDim1*IFMDim1][IFM_Channels1];
+ static ap_int<ACTIVATION_PRECISION> TEST[MAX_IMAGES][OFMDim1][OFMDim1][OFM_Channels1];
+ stream<ap_uint<IFM_Channels1*INPUT_PRECISION> > input_stream("input_stream");
+ stream<ap_uint<OFM_Channels1*ACTIVATION_PRECISION> > output_stream("output_stream");
+ unsigned int counter = 0;
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < IFMDim1; oy++) {
+ for (unsigned int ox = 0; ox < IFMDim1; ox++) {
+ ap_uint<INPUT_PRECISION*IFM_Channels1> input_channel = 0;
+ for(unsigned int channel = 0; channel < IFM_Channels1; channel++)
+ {
+ ap_uint<INPUT_PRECISION> input = (ap_uint<INPUT_PRECISION>)(counter);
+ IMAGE[n_image][oy*IFMDim1+ox][channel]= input;
+ input_channel = input_channel >> INPUT_PRECISION;
+ input_channel(IFM_Channels1*INPUT_PRECISION-1,(IFM_Channels1-1)*INPUT_PRECISION)=input;
+
+ counter++;
+ }
+ input_stream.write(input_channel);
+ }
+ }
+ }
+ static ap_int<4> W1[OFM_Channels1][KERNEL_DIM][KERNEL_DIM][IFM_Channels1];
+ // initialize the weights
+ constexpr int TX = (IFM_Channels1*KERNEL_DIM*KERNEL_DIM) / SIMD1;
+ constexpr int TY = OFM_Channels1 / PE1;
+ unsigned int kx=0;
+ unsigned int ky=0;
+ unsigned int chan_count=0;
+ unsigned int out_chan_count=0;
+ for (unsigned int oy = 0; oy < TY; oy++) {
+ for (unsigned int ox = 0; ox <TX; ox++) {
+ for(int pe=0;pe <PE1;pe++){
+ for(int simd=0;simd<SIMD1;simd++){
+ W1[out_chan_count][kx][ky][chan_count] = PARAM::weights.weights(oy*TX + ox)[pe][simd];
+ kx++;
+ if (kx==KERNEL_DIM){
+ kx=0;
+ ky++;
+ if (ky==KERNEL_DIM){
+ ky=0;
+ chan_count++;
+ if (chan_count==IFM_Channels1){
+ chan_count=0;
+ out_chan_count++;
+ if (out_chan_count==OFM_Channels1){
+ out_chan_count=0;
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+ conv<MAX_IMAGES,IFMDim1,OFMDim1,IFM_Channels1,OFM_Channels1, KERNEL_DIM, 1, ap_uint<INPUT_PRECISION> >(IMAGE, W1, TEST);
+ Testbench_convmmv(input_stream, output_stream, MAX_IMAGES);
+ int err_counter = 0, err_perimage=0;
+ ap_int<ACTIVATION_PRECISION> out_chan;
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < OFMDim1; oy++) {
+ for (unsigned int ox = 0; ox < OFMDim1; ox++) {
+ for(int e=0;e<1;e++){
+ ap_uint<OFM_Channels1*ACTIVATION_PRECISION> outElem = output_stream.read();
+ for(unsigned int channel = 0; channel < OFM_Channels1; channel++){
+ ap_int<ACTIVATION_PRECISION> EXP = TEST[n_image][ox][oy][channel + e * OFM_Channels1];
+ out_chan(ACTIVATION_PRECISION-1,0) = outElem((channel + 1)*ACTIVATION_PRECISION-1,channel*ACTIVATION_PRECISION);
+
+ if (EXP != out_chan){
+ std::cout << "ERROR: Expected["<<oy <<"]["<<ox<<"]["<<channel<<"]=" << EXP << " actual " << out_chan << std::endl;
+ return 1;
+ err_counter ++;
+ err_perimage++;
+ //if(err_counter>10)
+ //return 1;
+ }
+ }
+ }
+ }
+ }
+ if(err_perimage == 0){
+ std::cout << "Image # " << n_image << " passed the testing."<< std::endl;
+ }
+ else{
+ err_perimage=0;
+ std::cout << "Image # " << n_image << " failed the testing."<< std::endl;
+ }
+ }
+ if(err_counter == 0){
+ return 0;
+ }
+ else{
+ return 1;
+ }
+
+}
+
+
diff --git a/finn-hlslib/tb/conv_mmv.cpp b/finn-hlslib/tb/conv_mmv.cpp
new file mode 100755
index 000000000..a1707b65c
--- /dev/null
+++ b/finn-hlslib/tb/conv_mmv.cpp
@@ -0,0 +1,59 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file conv_top.cpp
+ *
+ * HLS Top function with a single convolutional layer for unit testing
+ *
+ *****************************************************************************/
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+
+#include "activations.hpp"
+#include "weights.hpp"
+#include "activations.hpp"
+#include "interpret.hpp"
+#include "mvau.hpp"
+#include "conv.hpp"
+#include "memdata.h"
+#include "config.h"
+
+void Testbench_convmmv(stream<ap_uint<IFM_Channels1*INPUT_PRECISION> > & in, stream<ap_uint<OFM_Channels1*ACTIVATION_PRECISION> > & out, unsigned int numReps){
+#pragma HLS DATAFLOW
+ ConvLayer_Batch_MMV<KERNEL_DIM, IFM_Channels1, IFMDim1, OFM_Channels1, OFMDim1, STRIDE, SIMD1, PE1, MMV1, Slice_mmv<ap_uint<INPUT_PRECISION>, MMV1 >, Slice_mmv<ap_int<16>, MMV1 >, Identity >(in, out, PARAM::weights, PassThroughActivation<ap_uint<16>>(), numReps, ap_resource_dsp());
+
+}
diff --git a/finn-hlslib/tb/conv_top.cpp b/finn-hlslib/tb/conv_top.cpp
new file mode 100755
index 000000000..f4279bc21
--- /dev/null
+++ b/finn-hlslib/tb/conv_top.cpp
@@ -0,0 +1,58 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file conv_top.cpp
+ *
+ * HLS Top function with a single convolutional layer for unit testing
+ *
+ *****************************************************************************/
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+
+#include "activations.hpp"
+#include "weights.hpp"
+#include "activations.hpp"
+#include "interpret.hpp"
+#include "mvau.hpp"
+#include "conv.hpp"
+#include "memdata.h"
+#include "config.h"
+
+void Testbench_conv(stream<ap_uint<IFM_Channels1*INPUT_PRECISION> > & in, stream<ap_uint<OFM_Channels1*ACTIVATION_PRECISION> > & out, unsigned int numReps){
+#pragma HLS DATAFLOW
+ ConvLayer_Batch<KERNEL_DIM, IFM_Channels1, IFMDim1, OFM_Channels1, OFMDim1, SIMD1, PE1, Slice<ap_uint<INPUT_PRECISION> >, Slice<ap_int<16> >, Identity >(in, out, PARAM::weights, PassThroughActivation<ap_uint<16>>(), numReps, ap_resource_dsp());
+}
diff --git a/finn-hlslib/tb/dup_stream_config.h b/finn-hlslib/tb/dup_stream_config.h
new file mode 100755
index 000000000..c2be8fd45
--- /dev/null
+++ b/finn-hlslib/tb/dup_stream_config.h
@@ -0,0 +1,2 @@
+#define WIDTH 8
+#define NUM_REPEAT 4
\ No newline at end of file
diff --git a/finn-hlslib/tb/dup_stream_tb.cpp b/finn-hlslib/tb/dup_stream_tb.cpp
new file mode 100755
index 000000000..14a5567f2
--- /dev/null
+++ b/finn-hlslib/tb/dup_stream_tb.cpp
@@ -0,0 +1,50 @@
+#include <iostream>
+#include <cmath>
+#include <ctime>
+#include <cstring>
+#include <hls_stream.h>
+#include <cstdlib>
+#define AP_INT_MAX_W 16384
+#include "ap_int.h"
+#include "weights.hpp"
+#include "bnn-library.h"
+
+#include "dup_stream_config.h"
+
+#include "activations.hpp"
+#include "interpret.hpp"
+
+using namespace hls;
+using namespace std;
+
+#define MAX_IMAGES 1
+void Testbench_dup_stream(stream<ap_uint<WIDTH> > & in, stream<ap_uint<WIDTH> > & out1, stream<ap_uint<WIDTH> > & out2, unsigned int numReps);
+
+int main()
+{
+ stream<ap_uint<WIDTH> > input_stream("input_stream");
+ stream<ap_uint<WIDTH> > output_stream1("output_stream");
+ stream<ap_uint<WIDTH> > output_stream2("output_stream");
+ static ap_uint<WIDTH> expected[NUM_REPEAT*MAX_IMAGES];
+ unsigned int count_out = 0;
+ unsigned int count_in = 0;
+ for (unsigned int counter = 0; counter < NUM_REPEAT*MAX_IMAGES; counter++) {
+ ap_uint<WIDTH> value = (ap_uint<WIDTH>) counter;
+ input_stream.write(value);
+ expected[counter] = value;
+ }
+ Testbench_dup_stream(input_stream, output_stream1, output_stream2, MAX_IMAGES);
+ for (unsigned int counter=0 ; counter < NUM_REPEAT*MAX_IMAGES; counter++)
+ {
+ ap_uint<WIDTH> value1 = output_stream1.read();
+ ap_uint<WIDTH> value2 = output_stream2.read();
+ if((value1!= expected[counter]) || (value1!= expected[counter]))
+ {
+ cout << "ERROR with counter " << counter << std::hex << " expected " << expected[counter] << " value1 " << value1 << " value2 " << value2 << std::dec << endl;
+ return(1);
+ }
+ }
+
+}
+
+
diff --git a/finn-hlslib/tb/dup_stream_top.cpp b/finn-hlslib/tb/dup_stream_top.cpp
new file mode 100755
index 000000000..8090171d3
--- /dev/null
+++ b/finn-hlslib/tb/dup_stream_top.cpp
@@ -0,0 +1,11 @@
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+
+
+#include "dup_stream_config.h"
+
+void Testbench_dup_stream(stream<ap_uint<WIDTH> > & in, stream<ap_uint<WIDTH> > & out1, stream<ap_uint<WIDTH> > & out2, unsigned int numReps){
+ DuplicateStreams_Batch<WIDTH, NUM_REPEAT>(in, out1, out2, numReps);
+}
diff --git a/finn-hlslib/tb/dwc_config.h b/finn-hlslib/tb/dwc_config.h
new file mode 100755
index 000000000..dbdf165ad
--- /dev/null
+++ b/finn-hlslib/tb/dwc_config.h
@@ -0,0 +1,34 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+#define INPUT_WIDTH 8
+#define OUT_WIDTH 4
+#define NUM_REPEAT 4
diff --git a/finn-hlslib/tb/dwc_tb.cpp b/finn-hlslib/tb/dwc_tb.cpp
new file mode 100755
index 000000000..b70151aaf
--- /dev/null
+++ b/finn-hlslib/tb/dwc_tb.cpp
@@ -0,0 +1,112 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file dwc_tb.cpp
+ *
+ * Testbench for the data-width converter HLS block
+ *
+ *****************************************************************************/
+#include <iostream>
+#include <cmath>
+#include <ctime>
+#include <cstring>
+#include <hls_stream.h>
+#include <cstdlib>
+#define AP_INT_MAX_W 16384
+#include "ap_int.h"
+#include "weights.hpp"
+#include "bnn-library.h"
+
+#include "dwc_config.h"
+
+#include "activations.hpp"
+#include "interpret.hpp"
+
+using namespace hls;
+using namespace std;
+
+#define MAX_IMAGES 1
+void Testbench_dwc(stream<ap_uint<INPUT_WIDTH> > & in, stream<ap_uint<OUT_WIDTH> > & out, unsigned int numReps);
+
+int main()
+{
+ stream<ap_uint<INPUT_WIDTH> > input_stream("input_stream");
+ stream<ap_uint<OUT_WIDTH> > output_stream("output_stream");
+ static ap_uint<OUT_WIDTH> expected[NUM_REPEAT*MAX_IMAGES*INPUT_WIDTH/OUT_WIDTH];
+ unsigned int count_out = 0;
+ unsigned int count_in = 0;
+ for (unsigned int counter = 0; counter < NUM_REPEAT*MAX_IMAGES; counter++) {
+ ap_uint<INPUT_WIDTH> value = (ap_uint<INPUT_WIDTH>) counter;
+ input_stream.write(value);
+ if(INPUT_WIDTH < OUT_WIDTH){
+ ap_uint<OUT_WIDTH> val_out = expected[count_out];
+ val_out = val_out >> INPUT_WIDTH;
+ val_out(OUT_WIDTH-1,OUT_WIDTH-INPUT_WIDTH)=value;
+ expected[count_out]=val_out;
+ count_in++;
+ if (count_in == OUT_WIDTH/INPUT_WIDTH){
+ count_out++;
+ count_in=0;
+ }
+ }
+ else if(INPUT_WIDTH == OUT_WIDTH)
+ {
+ expected[counter] = value;
+ } else //INPUT_WIDTH > OUT_WIDTH
+ {
+
+ for (unsigned int word_count=0;word_count< INPUT_WIDTH/OUT_WIDTH; word_count++)
+ {
+ ap_uint<OUT_WIDTH> val_out = value(OUT_WIDTH-1,0);
+ value = value >> OUT_WIDTH;
+ expected[count_out] = val_out;
+ count_out++;
+ }
+ }
+ }
+ Testbench_dwc(input_stream, output_stream, MAX_IMAGES);
+ for (unsigned int counter=0 ; counter < NUM_REPEAT*MAX_IMAGES*INPUT_WIDTH/OUT_WIDTH; counter++)
+ {
+ ap_uint<OUT_WIDTH> value = output_stream.read();
+ if(value!= expected[counter])
+ {
+ cout << "ERROR with counter " << counter << std::hex << " expected " << expected[counter] << " value " << value << std::dec << endl;
+ return(1);
+ }
+ }
+
+}
+
+
diff --git a/finn-hlslib/tb/dwc_top.cpp b/finn-hlslib/tb/dwc_top.cpp
new file mode 100755
index 000000000..900654c15
--- /dev/null
+++ b/finn-hlslib/tb/dwc_top.cpp
@@ -0,0 +1,51 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file dwc_top.cpp
+ *
+ * HLS Top function with a single HLS data-width converter block unit testing
+ *
+ *****************************************************************************/
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+
+
+#include "dwc_config.h"
+
+void Testbench_dwc(stream<ap_uint<INPUT_WIDTH> > & in, stream<ap_uint<OUT_WIDTH> > & out, unsigned int numReps){
+ StreamingDataWidthConverter_Batch<INPUT_WIDTH, OUT_WIDTH, NUM_REPEAT>(in, out, numReps);
+}
diff --git a/finn-hlslib/tb/gen_weigths.py b/finn-hlslib/tb/gen_weigths.py
new file mode 100755
index 000000000..de2a67c11
--- /dev/null
+++ b/finn-hlslib/tb/gen_weigths.py
@@ -0,0 +1,100 @@
+# Copyright (c) 2019, Xilinx, Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice,
+# this list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in the
+# documentation and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+# OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+# ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+import numpy as np
+import os
+import sys
+import random
+import subprocess
+
+outFileWeights = open("memdata.h" , "wt")
+outFileConfig = open("config.h" , "wt")
+
+kernel_dim = 3
+stride = 1
+input_precision = 4
+ifm_channels = 1
+ofm_channels = 1
+ifm_dimension = 8
+ofm_dimension = 6
+
+activation_precision = 16
+expand = 1
+simd = 1
+pe = 1
+w_precision = 1
+mmv=2
+
+tile = ifm_channels *kernel_dim*kernel_dim * ofm_channels // (simd*pe)
+
+outFileConfig.write("#define KERNEL_DIM %d \n" % kernel_dim)
+outFileConfig.write("#define SIMD1 %d \n" % simd)
+outFileConfig.write("#define PE1 %d \n" % pe)
+outFileConfig.write("#define MMV1 %d \n" % mmv)
+outFileConfig.write("#define WIDTH %d \n" % w_precision)
+
+outFileConfig.write("#define IFM_Channels1 %d \n" % ifm_channels)
+outFileConfig.write("#define OFM_Channels1 %d \n" % ofm_channels)
+outFileConfig.write("#define IFMDim1 %d \n" % ifm_dimension)
+outFileConfig.write("#define OFMDim1 %d \n" % ofm_dimension)
+outFileConfig.write("#define STRIDE %d \n" % stride)
+outFileConfig.write("#define INPUT_PRECISION %d \n" % input_precision)
+outFileConfig.write("#define TILE1 %d \n" % tile)
+
+outFileConfig.write("#define ACTIVATION_PRECISION %d \n" % activation_precision)
+
+outFileConfig.close()
+outFileWeights.write("#ifndef PARAMS_HPP\n")
+outFileWeights.write("#define PARAMS_HPP\n")
+
+outFileWeights.write("namespace PARAM{ \n")
+if (w_precision == 1):
+ outFileWeights.write("static BinaryWeights<%d,%d,%d> weights= {\n{\n" %(simd,pe,tile))
+else:
+ outFileWeights.write("static FixedPointWeights<%d,ap_int<%d>,%d,%d> weights= {\n{\n" %(simd,w_precision,pe,tile))
+
+for p in range(pe):
+ outFileWeights.write("{ \n")
+ for t in range(tile):
+ width = simd*w_precision;
+ val = random.randint(0, 1<<width-1)
+ outFileWeights.write("%s" % hex(val))
+ if t!=tile-1:
+ outFileWeights.write(",\n")
+ outFileWeights.write("} \n")
+ if p!=pe-1:
+ outFileWeights.write(",")
+
+
+outFileWeights.write("}\n};\n } \n")
+outFileWeights.write("#endif \n")
+outFileWeights.close()
+
+
+
diff --git a/finn-hlslib/tb/input_gen.cpp b/finn-hlslib/tb/input_gen.cpp
new file mode 100755
index 000000000..beb8049a5
--- /dev/null
+++ b/finn-hlslib/tb/input_gen.cpp
@@ -0,0 +1,58 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file input_gen.cpp
+ *
+ * HLS Top function with a single HLS sliding-window generator block unit testing
+ *
+ *****************************************************************************/
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+#include "input_gen.h"
+
+void Testbench(stream<ap_uint<SIMD*INPUT_PRECISION> > & in, stream<ap_uint<SIMD*INPUT_PRECISION> > & out, unsigned int numReps)
+{
+ConvolutionInputGenerator<KERNEL_DIM,
+ IFM_Channels,
+ INPUT_PRECISION,
+ IFMDim,
+ OFMDim,
+ SIMD,
+ STRIDE>(in, out, numReps);
+
+}
+
diff --git a/finn-hlslib/tb/input_gen.h b/finn-hlslib/tb/input_gen.h
new file mode 100755
index 000000000..c64c4af34
--- /dev/null
+++ b/finn-hlslib/tb/input_gen.h
@@ -0,0 +1,41 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+#define SIMD 2
+#define KERNEL_DIM 2
+#define IFM_Channels 2
+#define IFMDim 6
+#define OFMDim 3
+#define STRIDE 2
+#define INPUT_PRECISION 8
+#define MMV 1
+#define INPUT_MULTIPLIER 1
+
diff --git a/finn-hlslib/tb/maxpool_tb.cpp b/finn-hlslib/tb/maxpool_tb.cpp
new file mode 100755
index 000000000..e1c5d4f14
--- /dev/null
+++ b/finn-hlslib/tb/maxpool_tb.cpp
@@ -0,0 +1,124 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file maxpool_tb.cpp
+ *
+ * Testbench for the maxpool layer HLS block
+ *
+ *****************************************************************************/
+#include <iostream>
+#include <cmath>
+#include <ctime>
+#include <cstring>
+#include <hls_stream.h>
+#include <cstdlib>
+#define AP_INT_MAX_W 16384
+#include "ap_int.h"
+#include "weights.hpp"
+#include "bnn-library.h"
+
+#include "pool_config.h"
+#include "pool.hpp"
+#include "activations.hpp"
+#include "interpret.hpp"
+
+using namespace hls;
+using namespace std;
+
+#define MAX_IMAGES 1
+void Testbench_pool(stream<ap_uint<FM_Channels1*PRECISION> > & in, stream<ap_uint<FM_Channels1*PRECISION> > & out, unsigned int numReps);
+
+int main()
+{
+ static ap_uint<PRECISION> IMAGE[MAX_IMAGES][IFMDim1][IFMDim1][FM_Channels1];
+ static ap_uint<PRECISION> OUTPUT[MAX_IMAGES][OFMDim1][OFMDim1][FM_Channels1];
+ stream<ap_uint<FM_Channels1*PRECISION> > input_stream("input_stream");
+ stream<ap_uint<FM_Channels1*PRECISION> > output_stream("output_stream");
+ unsigned int counter = 0;
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < IFMDim1; oy++) {
+ for (unsigned int ox = 0; ox < IFMDim1; ox++) {
+ ap_uint<PRECISION*FM_Channels1> input_channel = 0;
+ for(unsigned int channel = 0; channel < FM_Channels1; channel++)
+ {
+ ap_uint<PRECISION> input = (ap_uint<PRECISION>)(counter);
+ IMAGE[n_image][oy][ox][channel]= input;
+ input_channel = input_channel >> PRECISION;
+ input_channel(FM_Channels1*PRECISION-1,(FM_Channels1-1)*PRECISION)=input;
+
+ counter++;
+ }
+ input_stream.write(input_channel);
+ }
+ }
+ }
+ pool<MAX_IMAGES,IFMDim1,OFMDim1,FM_Channels1,KERNEL_DIM,KERNEL_DIM,ap_uint<PRECISION> >(IMAGE,OUTPUT);
+ Testbench_pool(input_stream, output_stream, MAX_IMAGES);
+ int err_counter = 0, err_perimage=0;
+ ap_uint<PRECISION> out_chan;
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < OFMDim1; oy++) {
+ for (unsigned int ox = 0; ox < OFMDim1; ox++) {
+ ap_uint<FM_Channels1*PRECISION> outElem = output_stream.read();
+ for(unsigned int channel = 0; channel < FM_Channels1; channel++){
+ ap_uint<PRECISION> EXP = OUTPUT[n_image][ox][oy][channel];
+ out_chan(PRECISION-1,0) = outElem((channel + 1)*PRECISION-1,channel*PRECISION);
+ if (EXP != out_chan){
+ std::cout << "ERROR: Expected["<<oy <<"]["<<ox<<"]["<<channel<<"]=" << EXP << " actual " << out_chan << std::endl;
+ err_counter ++;
+ err_perimage++;
+ }
+
+ }
+ }
+ }
+ if(err_perimage == 0){
+ std::cout << "Image # " << n_image << " passed the testing."<< std::endl;
+ }
+ else{
+ err_perimage=0;
+ std::cout << "Image # " << n_image << " failed the testing."<< std::endl;
+ }
+ }
+ if(err_counter == 0){
+ return 0;
+ }
+ else{
+ return 1;
+ }
+
+}
+
+
diff --git a/finn-hlslib/tb/pool.hpp b/finn-hlslib/tb/pool.hpp
new file mode 100755
index 000000000..73c5dfa8e
--- /dev/null
+++ b/finn-hlslib/tb/pool.hpp
@@ -0,0 +1,67 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file pool.hpp
+ *
+ * C++ Implementation of a max pool layer, used for testbench
+ *
+ *****************************************************************************/
+
+#ifndef POOL_TB_H
+#define POOL_TB_H
+
+template<int MAX_IMAGE,
+ int IFMDim,
+ int OFMDim,
+ int FMCh,
+ int kernel,
+ int stride,
+ typename TI>
+ void pool(TI const img[MAX_IMAGE][IFMDim][IFMDim][FMCh], TI out[MAX_IMAGE][OFMDim][OFMDim][FMCh]){
+ for(int n=0;n<MAX_IMAGE;n++)
+ for(int x=0;x<OFMDim;x++)
+ for(int y=0;y<OFMDim;y++)
+ for(int h=0;h<FMCh;h++){
+ TI tmp = 0;
+ for (int ky=0;ky<kernel;ky++)
+ for (int kx=0;kx<kernel;kx++)
+ if(img[n][(y*stride+ky)][x*stride+kx][h]>tmp){
+ tmp=img[n][(y*stride+ky)][x*stride+kx][h];
+ }
+ out[n][x][y][h] = tmp;
+ }
+ }
+
+#endif
diff --git a/finn-hlslib/tb/pool_config.h b/finn-hlslib/tb/pool_config.h
new file mode 100755
index 000000000..ccf93c08f
--- /dev/null
+++ b/finn-hlslib/tb/pool_config.h
@@ -0,0 +1,38 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+#define KERNEL_DIM 2
+#define WIDTH 4
+#define FM_Channels1 16
+#define IFMDim1 16
+#define OFMDim1 8
+#define PRECISION 3
+
diff --git a/finn-hlslib/tb/pool_top.cpp b/finn-hlslib/tb/pool_top.cpp
new file mode 100755
index 000000000..60b1fa385
--- /dev/null
+++ b/finn-hlslib/tb/pool_top.cpp
@@ -0,0 +1,51 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file pool_top.cpp
+ *
+ * HLS Top function with a single HLS max pool block unit testing
+ *
+ *****************************************************************************/
+#include <hls_stream.h>
+using namespace hls;
+#include "ap_int.h"
+#include "bnn-library.h"
+
+
+#include "pool_config.h"
+
+void Testbench_pool(stream<ap_uint<FM_Channels1*PRECISION> > & in, stream<ap_uint<FM_Channels1*PRECISION> > & out, unsigned int numReps){
+ StreamingMaxPool_Precision_Batch<IFMDim1, KERNEL_DIM, FM_Channels1, ap_uint<PRECISION>, 0>(in, out, numReps);
+}
diff --git a/finn-hlslib/tb/swg_tb.cpp b/finn-hlslib/tb/swg_tb.cpp
new file mode 100755
index 000000000..a0604521d
--- /dev/null
+++ b/finn-hlslib/tb/swg_tb.cpp
@@ -0,0 +1,105 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ ******************************************************************************/
+/******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ *
+ * \file swg_tb.cpp
+ *
+ * Testbench for the sliding window generator HLS block
+ *
+ *****************************************************************************/
+#include <hls_stream.h>
+#include "ap_int.h"
+#include <iostream>
+#include <string>
+#include "input_gen.h"
+#include "math.h"
+using namespace hls;
+using namespace std;
+
+#define MAX_IMAGES 2
+
+void Testbench(stream<ap_uint<IFM_Channels*INPUT_PRECISION> > & in, stream<ap_uint<IFM_Channels*INPUT_PRECISION> > & out, unsigned int numReps);
+
+
+int main()
+{
+static ap_uint<INPUT_PRECISION> INPUT_IMAGES[MAX_IMAGES][IFMDim*IFMDim][IFM_Channels];
+stream<ap_uint<IFM_Channels*INPUT_PRECISION> > input_stream("input_stream");
+stream<ap_uint<IFM_Channels*INPUT_PRECISION> > output_stream("output_stream");
+ unsigned int counter = 0;
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < IFMDim; oy++) {
+ for (unsigned int ox = 0; ox < IFMDim; ox++) {
+ ap_uint<INPUT_PRECISION*IFM_Channels> input_channel = 0;
+ for(unsigned int channel = 0; channel < IFM_Channels; channel++)
+ {
+ ap_uint<INPUT_PRECISION> input = (ap_uint<INPUT_PRECISION>)(counter);
+ INPUT_IMAGES[n_image][oy*IFMDim+ox][channel]= input;
+ input_channel = input_channel >> INPUT_PRECISION;
+ input_channel(IFM_Channels*INPUT_PRECISION-1,(IFM_Channels-1)*INPUT_PRECISION)=input;
+
+ counter++;
+ }
+ input_stream.write(input_channel);
+ }
+ }
+ }
+ Testbench(input_stream, output_stream, MAX_IMAGES);
+ for (unsigned int n_image = 0; n_image < MAX_IMAGES; n_image++) {
+ for (unsigned int oy = 0; oy < OFMDim; oy++) {
+ for (unsigned int ox = 0; ox < OFMDim; ox+=MMV) {
+ for (unsigned int ky = 0; ky < KERNEL_DIM; ky++) {
+ for (unsigned int kx = 0; kx < KERNEL_DIM; kx++) {
+ unsigned int input_base = (oy*STRIDE) * IFMDim + (ox*STRIDE);
+ unsigned int input_ind = input_base + ky * IFMDim + kx;
+ ap_uint<IFM_Channels*INPUT_PRECISION> outElem = output_stream.read();
+ for(unsigned int channel = 0; channel < IFM_Channels; channel++){
+ ap_uint<INPUT_PRECISION> out_chan = 0;
+ out_chan = outElem(INPUT_PRECISION-1,0);
+ if (((INPUT_IMAGES[n_image][input_ind][channel])) != out_chan){
+ std::cout << "ERROR: " << " Expected " << INPUT_IMAGES[n_image][input_ind][channel] << " actual " << out_chan << std::endl;
+ std::cout << "oy= " << oy << " ox= " << ox << " ky= " << ky << " kx= " << kx << std::endl;
+ return 1;
+ }
+ outElem = outElem >> INPUT_PRECISION;
+ }
+ }
+ }
+ }
+ }
+ std::cout << "Image # " << n_image << std::endl;
+ }
+ return 0;
+
+}
diff --git a/finn-hlslib/tb/test_add.tcl b/finn-hlslib/tb/test_add.tcl
new file mode 100755
index 000000000..612cf10da
--- /dev/null
+++ b/finn-hlslib/tb/test_add.tcl
@@ -0,0 +1,11 @@
+open_project hls-syn-add
+add_files add_top.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+add_files -tb add_tb.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+set_top Testbench_add
+open_solution sol1
+set_part {xczu3eg-sbva484-1-i}
+create_clock -period 5 -name default
+csim_design
+csynth_design
+cosim_design
+exit
diff --git a/finn-hlslib/tb/test_conv3.tcl b/finn-hlslib/tb/test_conv3.tcl
new file mode 100755
index 000000000..5d96fb7de
--- /dev/null
+++ b/finn-hlslib/tb/test_conv3.tcl
@@ -0,0 +1,51 @@
+##############################################################################
+ # Copyright (c) 2019, Xilinx, Inc.
+ # All rights reserved.
+ #
+ # Redistribution and use in source and binary forms, with or without
+ # modification, are permitted provided that the following conditions are met:
+ #
+ # 1. Redistributions of source code must retain the above copyright notice,
+ # this list of conditions and the following disclaimer.
+ #
+ # 2. Redistributions in binary form must reproduce the above copyright
+ # notice, this list of conditions and the following disclaimer in the
+ # documentation and/or other materials provided with the distribution.
+ #
+ # 3. Neither the name of the copyright holder nor the names of its
+ # contributors may be used to endorse or promote products derived from
+ # this software without specific prior written permission.
+ #
+ # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ # THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ # OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ # WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ # OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ # ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ #
+###############################################################################
+###############################################################################
+ #
+ # Authors: Giulio Gambardella <giuliog@xilinx.com>
+ #
+ # \file test_conv3.tcl
+ #
+ # Tcl script for HLS csim, synthesis and cosim of the convolutional layer
+ #
+###############################################################################
+open_project hls-syn-conv
+add_files conv_top.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+add_files -tb conv3_tb.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+set_top Testbench_conv
+open_solution sol1
+set_part {xczu3eg-sbva484-1-i}
+create_clock -period 5 -name default
+csim_design
+csynth_design
+cosim_design
+exit
diff --git a/finn-hlslib/tb/test_convmmv.tcl b/finn-hlslib/tb/test_convmmv.tcl
new file mode 100755
index 000000000..aaada6901
--- /dev/null
+++ b/finn-hlslib/tb/test_convmmv.tcl
@@ -0,0 +1,51 @@
+##############################################################################
+ # Copyright (c) 2019, Xilinx, Inc.
+ # All rights reserved.
+ #
+ # Redistribution and use in source and binary forms, with or without
+ # modification, are permitted provided that the following conditions are met:
+ #
+ # 1. Redistributions of source code must retain the above copyright notice,
+ # this list of conditions and the following disclaimer.
+ #
+ # 2. Redistributions in binary form must reproduce the above copyright
+ # notice, this list of conditions and the following disclaimer in the
+ # documentation and/or other materials provided with the distribution.
+ #
+ # 3. Neither the name of the copyright holder nor the names of its
+ # contributors may be used to endorse or promote products derived from
+ # this software without specific prior written permission.
+ #
+ # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ # THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ # OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ # WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ # OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ # ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ #
+###############################################################################
+###############################################################################
+ #
+ # Authors: Giulio Gambardella <giuliog@xilinx.com>
+ #
+ # \file test_convmmv.tcl
+ #
+ # Tcl script for HLS csim, synthesis and cosim of the convolutional layer with MMV
+ #
+###############################################################################
+open_project hls-syn-convmmv
+add_files conv_mmv.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+add_files -tb conv3mmv_tb.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+set_top Testbench_convmmv
+open_solution sol1
+set_part {xczu3eg-sbva484-1-i}
+create_clock -period 5 -name default
+csim_design
+csynth_design
+cosim_design
+exit
diff --git a/finn-hlslib/tb/test_dup_stream.tcl b/finn-hlslib/tb/test_dup_stream.tcl
new file mode 100755
index 000000000..a80fd5f37
--- /dev/null
+++ b/finn-hlslib/tb/test_dup_stream.tcl
@@ -0,0 +1,11 @@
+open_project hls-syn-dup_stream
+add_files dup_stream_top.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+add_files -tb dup_stream_tb.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+set_top Testbench_dup_stream
+open_solution sol1
+set_part {xczu3eg-sbva484-1-i}
+create_clock -period 5 -name default
+csim_design
+csynth_design
+cosim_design
+exit
diff --git a/finn-hlslib/tb/test_dwc.tcl b/finn-hlslib/tb/test_dwc.tcl
new file mode 100755
index 000000000..8faf3a523
--- /dev/null
+++ b/finn-hlslib/tb/test_dwc.tcl
@@ -0,0 +1,51 @@
+##############################################################################
+ # Copyright (c) 2019, Xilinx, Inc.
+ # All rights reserved.
+ #
+ # Redistribution and use in source and binary forms, with or without
+ # modification, are permitted provided that the following conditions are met:
+ #
+ # 1. Redistributions of source code must retain the above copyright notice,
+ # this list of conditions and the following disclaimer.
+ #
+ # 2. Redistributions in binary form must reproduce the above copyright
+ # notice, this list of conditions and the following disclaimer in the
+ # documentation and/or other materials provided with the distribution.
+ #
+ # 3. Neither the name of the copyright holder nor the names of its
+ # contributors may be used to endorse or promote products derived from
+ # this software without specific prior written permission.
+ #
+ # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ # THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ # OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ # WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ # OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ # ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ #
+###############################################################################
+###############################################################################
+ #
+ # Authors: Giulio Gambardella <giuliog@xilinx.com>
+ #
+ # \file test_dwc.tcl
+ #
+ # Tcl script for HLS csim, synthesis and cosim of the datawidth converter block
+ #
+###############################################################################
+open_project hls-syn-dwc
+add_files dwc_top.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+add_files -tb dwc_tb.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+set_top Testbench_dwc
+open_solution sol1
+set_part {xczu3eg-sbva484-1-i}
+create_clock -period 5 -name default
+csim_design
+csynth_design
+cosim_design
+exit
diff --git a/finn-hlslib/tb/test_pool.tcl b/finn-hlslib/tb/test_pool.tcl
new file mode 100755
index 000000000..0aef96333
--- /dev/null
+++ b/finn-hlslib/tb/test_pool.tcl
@@ -0,0 +1,51 @@
+##############################################################################
+ # Copyright (c) 2019, Xilinx, Inc.
+ # All rights reserved.
+ #
+ # Redistribution and use in source and binary forms, with or without
+ # modification, are permitted provided that the following conditions are met:
+ #
+ # 1. Redistributions of source code must retain the above copyright notice,
+ # this list of conditions and the following disclaimer.
+ #
+ # 2. Redistributions in binary form must reproduce the above copyright
+ # notice, this list of conditions and the following disclaimer in the
+ # documentation and/or other materials provided with the distribution.
+ #
+ # 3. Neither the name of the copyright holder nor the names of its
+ # contributors may be used to endorse or promote products derived from
+ # this software without specific prior written permission.
+ #
+ # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ # THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ # OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ # WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ # OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ # ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ #
+###############################################################################
+###############################################################################
+ #
+ # Authors: Giulio Gambardella <giuliog@xilinx.com>
+ #
+ # \file test-pool.tcl
+ #
+ # Tcl script for HLS csim, synthesis and cosim of the max pooling layer
+ #
+###############################################################################
+open_project hls-syn-pool
+add_files pool_top.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+add_files -tb maxpool_tb.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT) -I$::env(FINN_HLS_ROOT)/tb"
+set_top Testbench_pool
+open_solution sol1
+set_part {xczu3eg-sbva484-1-i}
+create_clock -period 5 -name default
+csim_design
+csynth_design
+cosim_design
+exit
diff --git a/finn-hlslib/tb/test_swg.tcl b/finn-hlslib/tb/test_swg.tcl
new file mode 100755
index 000000000..074f0819e
--- /dev/null
+++ b/finn-hlslib/tb/test_swg.tcl
@@ -0,0 +1,51 @@
+##############################################################################
+ # Copyright (c) 2019, Xilinx, Inc.
+ # All rights reserved.
+ #
+ # Redistribution and use in source and binary forms, with or without
+ # modification, are permitted provided that the following conditions are met:
+ #
+ # 1. Redistributions of source code must retain the above copyright notice,
+ # this list of conditions and the following disclaimer.
+ #
+ # 2. Redistributions in binary form must reproduce the above copyright
+ # notice, this list of conditions and the following disclaimer in the
+ # documentation and/or other materials provided with the distribution.
+ #
+ # 3. Neither the name of the copyright holder nor the names of its
+ # contributors may be used to endorse or promote products derived from
+ # this software without specific prior written permission.
+ #
+ # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ # THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ # OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ # WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ # OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ # ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ #
+###############################################################################
+###############################################################################
+ #
+ # Authors: Giulio Gambardella <giuliog@xilinx.com>
+ #
+ # \file test_swg.tcl
+ #
+ # Tcl script for HLS csim, synthesis and cosim of the sliding window generator block
+ #
+###############################################################################
+open_project hls-syn-swg
+add_files input_gen.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT)"
+add_files -tb swg_tb.cpp -cflags "-std=c++0x -I$::env(FINN_HLS_ROOT)"
+set_top Testbench
+open_solution sol1
+set_part {xczu3eg-sbva484-1-i}
+create_clock -period 5 -name default
+csim_design
+csynth_design
+cosim_design
+exit
diff --git a/finn-hlslib/utils.hpp b/finn-hlslib/utils.hpp
new file mode 100755
index 000000000..5ef2fac0b
--- /dev/null
+++ b/finn-hlslib/utils.hpp
@@ -0,0 +1,104 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ *******************************************************************************/
+
+/*******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * @file utils.hpp
+ *
+ * This project has received funding from the European Union's Framework
+ * Programme for Research and Innovation Horizon 2020 (2014-2020) under
+ * the Marie Skłodowska-Curie Grant Agreement No. 751339.
+ *
+ *******************************************************************************/
+
+#ifndef UTILS_HPP
+#define UTILS_HPP
+
+#include <iostream>
+#include <fstream>
+#include <cstddef>
+
+//- Static Evaluation of ceil(log2(x)) ---------------------------------------
+template<size_t N> struct clog2 {
+ static unsigned const value = 1 + ((N&1) == 0? clog2<N/2>::value : clog2<N/2+1>::value);
+};
+template<> struct clog2<0> {};
+template<> struct clog2<1> { static unsigned const value = 0; };
+template<> struct clog2<2> { static unsigned const value = 1; };
+
+//- Helpers to get hold of types ---------------------------------------------
+template<typename T> struct first_param {};
+template<typename R, typename A, typename... Args>
+struct first_param<R (*)(A, Args...)> { typedef A type; };
+template<typename C, typename R, typename A, typename... Args>
+struct first_param<R (C::*)(A, Args...)> { typedef A type; };
+
+//- Resource Representatives -------------------------------------------------
+class ap_resource_dflt {};
+class ap_resource_lut {};
+class ap_resource_dsp {};
+
+/**
+ * \brief Stream logger - Logging call to dump on file - not synthezisable
+ *
+ *
+ * \tparam BitWidth Width, in number of bits, of the input (and output) stream
+ *
+ * \param layer_name File name of the dump
+ * \param log Input (and output) stream
+ *
+ */
+template < unsigned int BitWidth >
+void logStringStream(const char *layer_name, hls::stream<ap_uint<BitWidth> > &log){
+ std::ofstream ofs(layer_name);
+ hls::stream<ap_uint<BitWidth> > tmp_stream;
+
+ while(!log.empty()){
+ ap_uint<BitWidth> tmp = (ap_uint<BitWidth>) log.read();
+ ofs << std::hex << tmp << std::endl;
+ tmp_stream.write(tmp);
+ }
+
+ while(!tmp_stream.empty()){
+ ap_uint<BitWidth> tmp = tmp_stream.read();
+ log.write((ap_uint<BitWidth>) tmp);
+ }
+
+ ofs.close();
+}
+
+#endif
diff --git a/finn-hlslib/weights.hpp b/finn-hlslib/weights.hpp
new file mode 100755
index 000000000..dda1ea435
--- /dev/null
+++ b/finn-hlslib/weights.hpp
@@ -0,0 +1,152 @@
+/******************************************************************************
+ * Copyright (c) 2019, Xilinx, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ * this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION). HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *******************************************************************************/
+
+/*******************************************************************************
+ *
+ * Authors: Giulio Gambardella <giuliog@xilinx.com>
+ * Thomas B. Preusser <thomas.preusser@utexas.edu>
+ * Marie-Curie Fellow, Xilinx Ireland, Grant Agreement No. 751339
+ * Christoph Doehring <cdoehrin@xilinx.com>
+ *
+ * @file weights.hpp
+ *
+ * Library of templated HLS classes for BNN deployment.
+ * This file lists a set of classes used to implement
+ * weights in neural network.
+ *
+ * This project has received funding from the European Union's Framework
+ * Programme for Research and Innovation Horizon 2020 (2014-2020) under
+ * the Marie Skłodowska-Curie Grant Agreement No. 751339.
+ *
+ *******************************************************************************/
+
+#ifndef WEIGHTS_HPP
+#define WEIGHTS_HPP
+
+#include <ap_int.h>
+#include <array>
+
+
+/**
+ * \brief A binary weight storage adapter that translates the internal
+ * organization optimized for storage to the generalized access by the MVAU.
+ *
+ * \tparam SIMD Number of input columns (channels) computed in parallel
+ * \tparam PE Number of output rows (channels) computed in parallel
+ * \tparam TILES 3rd dimension of the weights matrix
+ */
+template<unsigned SIMD, unsigned PE, unsigned TILES>
+class BinaryWeights {
+ public:
+ ap_uint<SIMD> m_weights[PE][TILES];
+
+ private:
+ /**
+ * Temporary container for the tile index to implement the
+ * memory access in pe -> tile order.
+ */
+ class TileIndex {
+ BinaryWeights const &m_par;
+ unsigned const m_idx;
+
+ public:
+ TileIndex(BinaryWeights const &par, unsigned const idx)
+ : m_par(par), m_idx(idx) {
+#pragma HLS inline
+ }
+
+ public:
+ ap_uint<SIMD> operator[](unsigned const pe) const {
+#pragma HLS inline
+ return m_par.m_weights[pe][m_idx];
+ }
+ };
+
+ public:
+ TileIndex weights(unsigned const tile) const {
+#pragma HLS inline
+ return TileIndex(*this, tile);
+ }
+};
+
+
+/**
+ * \brief A fixeed point weight storage adapter that translates the internal
+ * organization optimized for storage to the generalized access by the MVAU.
+ *
+ * \tparam SIMD Number of input columns (channels) computed in parallel
+ * \tparam WT Datatype of the weights
+ * \tparam PE Number of output rows (channels) computed in parallel
+ * \tparam TILES 3rd dimension of the weights matrix
+ */
+template<unsigned SIMD, typename WT ,unsigned PE, unsigned TILES>
+class FixedPointWeights {
+ public:
+ ap_uint<SIMD*WT::width> m_weights[PE][TILES];
+
+ private:
+ /**
+ * Temporary container for the tile index to implement the
+ * memory access in pe -> tile order.
+ */
+ class TileIndex {
+ FixedPointWeights const &m_par;
+ unsigned const m_idx;
+
+ public:
+ TileIndex(FixedPointWeights const &par, unsigned const idx)
+ : m_par(par), m_idx(idx) {
+#pragma HLS inline
+ }
+
+ public:
+ std::array<WT,SIMD> operator[](unsigned const pe) const {
+#pragma HLS inline
+ std::array<WT,SIMD> temp;
+ for(unsigned int i=0; i<SIMD; i++) {
+#pragma HLS unroll
+ ap_int<WT::width> local_temp;
+ local_temp = m_par.m_weights[pe][m_idx]((i+1)*WT::width-1, i*WT::width);
+ WT value = *reinterpret_cast<WT*>(&local_temp);
+ temp[i] = value;
+ }
+ return temp;
+ }
+ };
+
+ public:
+ TileIndex weights(unsigned const tile) const {
+#pragma HLS inline
+ return TileIndex(*this, tile);
+ }
+};
+
+#endif
diff --git a/run-docker.sh b/run-docker.sh
index 6b191376b..2d691b319 100755
--- a/run-docker.sh
+++ b/run-docker.sh
@@ -15,10 +15,12 @@ SCRIPTPATH=$(dirname "$SCRIPT")
BREVITAS_REPO=https://github.com/Xilinx/brevitas.git
EXAMPLES_REPO=https://github.com/maltanar/brevitas_cnv_lfc.git
CNPY_REPO=https://github.com/rogersce/cnpy.git
+FINN_HLS_REPO=https://github.com/Xilinx/finn-hlslib.git
BREVITAS_LOCAL=$SCRIPTPATH/brevitas
EXAMPLES_LOCAL=$SCRIPTPATH/brevitas_cnv_lfc
CNPY_LOCAL=$SCRIPTPATH/cnpy
+FINN_HLS_LOCAL=$SCRIPTPATH/finn-hlslib
# clone dependency repos
git clone --branch feature/finn_onnx_export $BREVITAS_REPO $BREVITAS_LOCAL || git -C "$BREVITAS_LOCAL" pull
@@ -28,6 +30,8 @@ echo "Mounting $SCRIPTPATH into /workspace/finn"
echo "Mounting $SCRIPTPATH/brevitas into /workspace/brevitas"
echo "Mounting $SCRIPTPATH/brevitas_cnv_lfc into /workspace/brevitas_cnv_lfc"
echo "Mounting $SCRIPTPATH/cnpy into /workspace/cnpy"
+echo "Mounting $SCRIPTPATH/finn-hlslib into /workspace/finn-hlslib"
+
# Build the FINN Docker image
docker build --tag=$DOCKER_TAG \
@@ -43,4 +47,5 @@ docker run --rm --name finn_dev -it \
-v $SCRIPTPATH/brevitas:/workspace/brevitas \
-v $SCRIPTPATH/brevitas_cnv_lfc:/workspace/brevitas_cnv_lfc \
-v $SCRIPTPATH/cnpy:/workspace/cnpy \
+-v $SCRIPTPATH/finn-hlslib:/workspace/finn-hlslib \
$DOCKER_TAG bash
--
GitLab