From 8e5b8bac243f2734d716c09fb24c3853d2e75150 Mon Sep 17 00:00:00 2001
From: Yaman Umuroglu <maltanar@gmail.com>
Date: Thu, 5 Dec 2019 23:03:16 +0000
Subject: [PATCH] [Transform] fix InferBinaryStreamingFCLayer shapes and attrs
---
.../fpgadataflow/convert_to_hls_layers.py | 54 ++++++++++---------
.../test_convert_to_hls_layers.py | 18 +++----
2 files changed, 38 insertions(+), 34 deletions(-)
diff --git a/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py b/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py
index 60c3fa4d7..f75cf2e91 100644
--- a/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py
+++ b/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py
@@ -1,5 +1,6 @@
-import onnx.helper as oh
+from onnx import helper
+from finn.core.datatype import DataType
from finn.transformation import Transformation
@@ -35,40 +36,43 @@ class InferBinaryStreamingFCLayer(Transformation):
# create node with no parallelization first
pe = 1
simd = 1
- wmem = int(mw * mh)
- # extract threshold shape
- tmem = int(mh / pe)
- n_thres = T.shape[1]
+ assert mh % pe == 0
+ assert mw % simd == 0
+ wmem = mw * mh // (pe * simd)
+ assert mw * mh == wmem * pe * simd
+ nf = mh // pe
+ tmem = nf
assert T.shape[0] == 1 or T.shape[0] == mh
- assert n_thres == 1
- # W is expected to be (PE, WMEM, SIMD)
- # transpose first to meet finn-hlslib assumptions
- W_new = W.transpose().reshape(pe, wmem, simd)
- model.set_initializer(mm_weight, W_new)
- # T is expected to be (NF, PE, n_thres)
- # TODO need to double-check the threshold shape here
- T_new = T.reshape(pe, tmem, n_thres)
- model.set_initializer(mt_thres, T_new)
- # reshape input and output tensors to expected shape
- # input is expected to be (1, mw/simd, simd)
- # output is expected to be (1, mh/pe, pe)
- in_shape = [1, int(mw / simd), simd]
- out_shape = [1, int(mh / pe), pe]
+ idt = DataType.BINARY
+ wdt = DataType.BINARY
+ odt = model.get_tensor_datatype(mt_output)
+ if odt.bitwidth() == 1:
+ # covers both bipolar and binary
+ actval = 0
+ else:
+ actval = odt.min()
+ in_shape = [1, mw]
+ out_shape = [1, mh]
model.set_tensor_shape(mm_input, in_shape)
model.set_tensor_shape(mt_output, out_shape)
# create and insert new StreamingFCLayer node
- new_node = oh.make_node(
+ new_node = helper.make_node(
"StreamingFCLayer_Batch",
[mm_input, mm_weight, mt_thres],
[mt_output],
domain="finn",
backend="fpgadataflow",
- MH=mh,
- MW=mw,
- PE=1,
- SIMD=1,
- resDataType="Recast<XnorMul>",
resType="ap_resource_lut()",
+ MW=mw,
+ MH=mh,
+ SIMD=simd,
+ PE=pe,
+ WMEM=wmem,
+ TMEM=tmem,
+ inputDataType=idt.name,
+ weightDataType=wdt.name,
+ outputDataType=odt.name,
+ ActVal=actval,
)
graph.node.insert(node_ind, new_node)
# remove old nodes
diff --git a/tests/fpgadataflow/test_convert_to_hls_layers.py b/tests/fpgadataflow/test_convert_to_hls_layers.py
index 49eccd4fe..27a0dfe00 100644
--- a/tests/fpgadataflow/test_convert_to_hls_layers.py
+++ b/tests/fpgadataflow/test_convert_to_hls_layers.py
@@ -37,17 +37,17 @@ def test_convert_to_hls_layers_lfc_w1a1():
model = model.transform(to_hls.InferBinaryStreamingFCLayer())
fc0 = model.graph.node[2]
assert fc0.op_type == "StreamingFCLayer_Batch"
- assert model.get_tensor_shape(fc0.input[0]) == [1, 784, 1]
- assert model.get_tensor_shape(fc0.input[1]) == [1, 784 * 1024, 1]
- assert model.get_tensor_shape(fc0.input[2]) == [1, 1024, 1]
+ assert model.get_tensor_shape(fc0.input[0]) == [1, 784]
+ assert model.get_tensor_shape(fc0.input[1]) == [784, 1024]
+ assert model.get_tensor_shape(fc0.input[2]) == [1024, 1]
fc1 = model.graph.node[3]
assert fc1.op_type == "StreamingFCLayer_Batch"
- assert model.get_tensor_shape(fc1.input[0]) == [1, 1024, 1]
- assert model.get_tensor_shape(fc1.input[1]) == [1, 1024 * 1024, 1]
- assert model.get_tensor_shape(fc1.input[2]) == [1, 1024, 1]
+ assert model.get_tensor_shape(fc1.input[0]) == [1, 1024]
+ assert model.get_tensor_shape(fc1.input[1]) == [1024, 1024]
+ assert model.get_tensor_shape(fc1.input[2]) == [1024, 1]
fc2 = model.graph.node[4]
assert fc2.op_type == "StreamingFCLayer_Batch"
- assert model.get_tensor_shape(fc2.input[0]) == [1, 1024, 1]
- assert model.get_tensor_shape(fc2.input[1]) == [1, 1024 * 1024, 1]
- assert model.get_tensor_shape(fc2.input[2]) == [1, 1024, 1]
+ assert model.get_tensor_shape(fc2.input[0]) == [1, 1024]
+ assert model.get_tensor_shape(fc2.input[1]) == [1024, 1024]
+ assert model.get_tensor_shape(fc2.input[2]) == [1024, 1]
os.remove(export_onnx_path)
--
GitLab