diff --git a/fetch-repos.sh b/fetch-repos.sh
index ec737729276680bd444756521e84b1eb883abb3b..88c67e8bed0c73038e3d11cfcd518b1e4cf95291 100755
--- a/fetch-repos.sh
+++ b/fetch-repos.sh
@@ -32,7 +32,7 @@ FINN_EXP_COMMIT="9cbd2787b5160e2b44e0e8164a0df1457dbd5366"
BREVITAS_COMMIT="a5b71d6de1389d3e7db898fef72e014842670f03"
PYVERILATOR_COMMIT="64b8294ff1afebb47be76fcad6ae87027e0402c2"
CNPY_COMMIT="4e8810b1a8637695171ed346ce68f6984e585ef4"
-HLSLIB_COMMIT="36e6c8cb1019ba0307e1886011692a58e02f3bfa"
+HLSLIB_COMMIT="bb43a97f799b63f536885919f03ecdfcfb04f405"
OMX_COMMIT="d1065a788219ca0eb54d5e57600b1f9d7f67d4cc"
AVNET_BDF_COMMIT="2d49cfc25766f07792c0b314489f21fe916b639b"
XIL_BDF_COMMIT="8cf4bb674a919ac34e3d99d8d71a9e60af93d14e"
diff --git a/src/finn/custom_op/fpgadataflow/vectorvectoractivation.py b/src/finn/custom_op/fpgadataflow/vectorvectoractivation.py
index 27b23dd32835c265759a8cabfd2a3412844077ca..b0c05d1ad6c74ceaaaa2c932f4add3f0076bda51 100644
--- a/src/finn/custom_op/fpgadataflow/vectorvectoractivation.py
+++ b/src/finn/custom_op/fpgadataflow/vectorvectoractivation.py
@@ -29,6 +29,7 @@
import math
import numpy as np
import os
+import textwrap
import warnings
from qonnx.core.datatype import DataType
from qonnx.util.basic import (
@@ -41,6 +42,7 @@ from finn.custom_op.fpgadataflow.hlscustomop import HLSCustomOp
from finn.util.data_packing import (
npy_to_rtlsim_input,
numpy_to_hls_code,
+ pack_innermost_dim_as_hex_string,
rtlsim_output_to_npy,
)
@@ -67,6 +69,36 @@ class VectorVectorActivation(HLSCustomOp):
"accDataType": ("s", False, "INT32"),
# no-activation mode (produce accumulators)
"noActivation": ("i", False, 0, {0, 1}),
+ # memory mode for the layer weights
+ # const -- embedded weights, default, long compile/synth times
+ # decoupled -- streaming weights with weight streamer packaged inside IP
+ # external -- streaming weights with external streamer
+ "mem_mode": ("s", False, "const", {"const", "decoupled", "external"}),
+ # (mem_mode = decoupled only) whether weights will be writable through
+ # an AXI-lite interface during runtime
+ # 1 for enabled, 0 for disabled.
+ # see finn-rtllib/memstream/doc/README for more about the memory
+ # address map used for writable weights
+ # IMPORTANT: After using AXI lite to either read or write the weights,
+ # always "flush" the accelerator by first passing a dummy input
+ # vector through the accelerator. This will get rid of any old
+ # weight data from the weight FIFOs.
+ "runtime_writeable_weights": ("i", False, 0, {0, 1}),
+ # FPGA resource type for memories in decoupled mode
+ # auto -- let Vivado decide
+ # block -- use BRAM
+ # distributed -- use LUTRAM
+ # ultra -- use UltraRAM (URAM), must have runtime_writeable_weights=1
+ # see also https://www.xilinx.com/support/answers/38070.html
+ "ram_style": (
+ "s",
+ False,
+ "auto",
+ {"auto", "block", "distributed", "ultra"},
+ ),
+ # use xnor-popcount for binary weights/inputs, thus treating them
+ # as bipolar
+ "binaryXnorMode": ("i", False, 0, {0, 1}),
}
my_attrs.update(super().get_nodeattr_types())
return my_attrs
@@ -198,14 +230,23 @@ class VectorVectorActivation(HLSCustomOp):
out_width = o_bits * self.get_nodeattr("PE")
return out_width
- def get_folded_input_shape(self):
+ def get_folded_input_shape(self, ind=0):
k_h, k_w = self.get_nodeattr("Kernel")
sf = k_h * k_w
dim_h, dim_w = self.get_nodeattr("Dim")
ch = self.get_nodeattr("Channels")
pe = self.get_nodeattr("PE")
nf = ch // pe
- folded_input_shape = tuple([1, dim_h, dim_w, sf * nf, pe])
+
+ if ind == 0:
+ # calculate shape of input 0
+ folded_input_shape = tuple([1, dim_h, dim_w, sf * nf, pe])
+ elif ind == 1 and self.get_nodeattr("mem_mode") == "external":
+ # calculate shape of input 1 (weights)
+ folded_input_shape = tuple([1, sf * nf, pe])
+ else:
+ raise Exception("Undefined input shape for requested input")
+
return folded_input_shape
def get_folded_output_shape(self):
@@ -251,13 +292,31 @@ class VectorVectorActivation(HLSCustomOp):
ret = dict()
inp_hls_str = self.get_input_datatype().get_hls_datatype_str()
out_hls_str = self.get_output_datatype().get_hls_datatype_str()
+ inp_is_binary = self.get_input_datatype() == DataType["BINARY"]
+ # out_is_binary = self.get_output_datatype() == DataType["BINARY"]
+ wt_is_binary = self.get_weight_datatype() == DataType["BINARY"]
+ bin_xnor_mode = self.get_nodeattr("binaryXnorMode") == 1
+ if (inp_is_binary or wt_is_binary) and (not bin_xnor_mode):
+ raise Exception("True binary (non-bipolar) inputs not yet supported")
inp_is_bipolar = self.get_input_datatype() == DataType["BIPOLAR"]
+ # out_is_bipolar = self.get_output_datatype() == DataType["BIPOLAR"]
wt_is_bipolar = self.get_weight_datatype() == DataType["BIPOLAR"]
+ # reinterpret inp/wt as bipolar if bin_xnor_mode is iset
+ inp_is_bipolar = inp_is_bipolar or (inp_is_binary and bin_xnor_mode)
+ wt_is_bipolar = wt_is_bipolar or (wt_is_binary and bin_xnor_mode)
# fill in TSrcI and TWeightI
- # TODO handle bipolar inputs
- if inp_is_bipolar or wt_is_bipolar:
- raise Exception("VVAU node doesn't support bipolar values yet.")
- else:
+ # TODO check these with Giulio
+ # TODO handle non-bipolar binary inputs
+ if inp_is_bipolar and wt_is_bipolar:
+ ret["TSrcI"] = "Recast<XnorMul>"
+ ret["TWeightI"] = "Identity"
+ elif (not inp_is_bipolar) and wt_is_bipolar:
+ ret["TSrcI"] = "Slice<%s>" % inp_hls_str
+ ret["TWeightI"] = "Recast<Binary>"
+ elif inp_is_bipolar and (not wt_is_bipolar):
+ ret["TSrcI"] = "Recast<Binary>"
+ ret["TWeightI"] = "Identity"
+ elif (not inp_is_bipolar) and (not wt_is_bipolar):
ret["TSrcI"] = "Slice<%s>" % inp_hls_str
ret["TWeightI"] = "Identity"
@@ -286,6 +345,13 @@ class VectorVectorActivation(HLSCustomOp):
return ret
def get_hls_compatible_threshold_tensor(self, orig_thres_matrix):
+ """Convert the original numpy weight matrix orig_weight_matrix into
+ a form suitable for passing to the hlslib call:
+ * ensure MH % PE == 0
+ * for bipolar weights&inputs, ensure thresholds are positive
+ * interleave rows between PEs
+ * reshape into (PE, TMEM, n_thres_steps) and return
+ """
ch = self.get_nodeattr("Channels")
pe = self.get_nodeattr("PE")
tmem = self.calc_tmem()
@@ -295,14 +361,33 @@ class VectorVectorActivation(HLSCustomOp):
), """Threshold matrix dimension is
not as expected (2)."""
n_thres_steps = orig_thres_matrix.shape[1]
+ inp_is_bipolar = self.get_input_datatype() == DataType["BIPOLAR"]
+ wt_is_bipolar = self.get_weight_datatype() == DataType["BIPOLAR"]
+ # reinterpret inp/wt as bipolar if bin_xnor_mode is iset
+ inp_is_binary = self.get_input_datatype() == DataType["BINARY"]
+ wt_is_binary = self.get_weight_datatype() == DataType["BINARY"]
+ bin_xnor_mode = self.get_nodeattr("binaryXnorMode") == 1
+ inp_is_bipolar = inp_is_bipolar or (inp_is_binary and bin_xnor_mode)
+ wt_is_bipolar = wt_is_bipolar or (wt_is_binary and bin_xnor_mode)
+ if inp_is_bipolar and wt_is_bipolar:
+ # ensure all thresholds are nonnegative
+ assert (orig_thres_matrix >= 0).all()
+ # ensure all thresholds are integer
+ assert (orig_thres_matrix.astype(np.int32) == orig_thres_matrix).all()
ret = orig_thres_matrix
# workaround for vivado_hls threshold bug
- if ret[0][0] == 0:
+ if ret[0][0] == 0 and n_thres_steps == 1:
ret = np.copy(ret)
ret[0][0] = 1
warnings.warn(
"Setting 0-valued first threshold to 1 to avoid vivado_hls bug"
)
+ # ensure channels = mh , duplicating if necessary
+ if ret.shape[0] == 1:
+ ret = np.tile(ret, (ch, 1))
+ assert (
+ ret.shape[0] == ch
+ ), "Channels of threshold matrix are not as expected (ch)"
# distribute rows between PEs
ret = interleave_matrix_outer_dim_from_partitions(ret, pe)
assert (
@@ -319,43 +404,173 @@ class VectorVectorActivation(HLSCustomOp):
rows between PEs is not as expected (n_thres_steps)"""
return ret.reshape(1, pe, tmem, n_thres_steps)
- def generate_params(self, model, path):
- # weights
- weights = model.get_initializer(self.onnx_node.input[1])
+ def make_weight_file(self, weights, weight_file_mode, weight_file_name):
+ """Produce a file containing given weights in appropriate format for this
+ layer. This file can be used for either synthesis or run-time reconfig
+ of weights.
+
+ Arguments:
+ * weights : numpy array with weights to be put into the file
+ * weight_file_mode : one of {hls_header, decoupled_verilog_dat,
+ decoupled_runtime}
+ * weight_file_name : filename for the weight file to be generated
+ """
# convert weights into hlslib-compatible format
weight_tensor = self.get_hls_compatible_weight_tensor(weights)
- wdt = self.get_weight_datatype()
- code_gen_dir = path
-
- """Saves weights into params.h"""
- weight_hls_code = numpy_to_hls_code(weight_tensor, wdt, "weights", True, True)
- # write weights into params.h
- f_weights = open("{}/params.h".format(code_gen_dir), "w")
-
- if wdt.bitwidth() != 1:
- f_weights.write(
- "const FixedPointWeights<1,{},{},{}> weights = ".format(
- wdt.get_hls_datatype_str(),
- self.get_nodeattr("PE"),
- self.calc_wmem(),
+ export_wdt = self.get_weight_datatype()
+ # we have converted bipolar weights to binary for export,
+ # so use it as such for weight generation
+ if self.get_weight_datatype() == DataType["BIPOLAR"]:
+ export_wdt = DataType["BINARY"]
+ if weight_file_mode == "hls_header":
+ weight_hls_code = numpy_to_hls_code(
+ weight_tensor, export_wdt, "weights", True, True
+ )
+ # write weights into C++ header file as dictated by finn-hlslib
+ f_weights = open(weight_file_name, "w")
+ if export_wdt.bitwidth() != 1:
+ f_weights.write(
+ "const FixedPointWeights<1,{},{},{}> weights = ".format(
+ export_wdt.get_hls_datatype_str(),
+ self.get_nodeattr("PE"),
+ self.calc_wmem(),
+ )
)
+ else:
+ f_weights.write(
+ "const BinaryWeights<1,{},{}> weights = ".format(
+ self.get_nodeattr("PE"),
+ self.calc_wmem(),
+ )
+ )
+ f_weights.write(weight_hls_code)
+ f_weights.close()
+ elif "decoupled" in weight_file_mode:
+ # create a weight stream for various flavors of decoupled mode:
+ # transpose weight tensor from (1, PE, WMEM, SIMD) to (1, WMEM, PE, SIMD)
+ weight_tensor_unflipped = np.transpose(weight_tensor, (0, 2, 1, 3))
+ # reverse SIMD flip for saving weights in .npy
+ weight_tensor_simd_flipped = np.flip(weight_tensor_unflipped, axis=-1)
+ # PE flip for saving weights in .dat
+ weight_tensor_pe_flipped = np.flip(weight_tensor_unflipped, axis=-2)
+ # reshape weight tensor (simd_flipped and pe_flipped) to desired shape
+ pe = self.get_nodeattr("PE")
+ simd = 1
+ # simd_flipped
+ weight_tensor_simd_flipped = weight_tensor_simd_flipped.reshape(
+ 1, -1, pe * simd
+ )
+ weight_tensor_simd_flipped = weight_tensor_simd_flipped.copy()
+ # flipped
+ weight_tensor_pe_flipped = weight_tensor_pe_flipped.reshape(
+ 1, -1, pe * simd
)
+ weight_tensor_pe_flipped = weight_tensor_pe_flipped.copy()
+ if weight_file_mode == "decoupled_npy":
+ # save weight stream into npy for cppsim
+ np.save(weight_file_name, weight_tensor_simd_flipped)
+ elif weight_file_mode == "decoupled_verilog_dat":
+ # convert weight values into hexstring
+ weight_width = self.get_weightstream_width()
+ # pad to nearest 4 bits to get hex strings
+ weight_width_padded = roundup_to_integer_multiple(weight_width, 4)
+ weight_tensor_pe_flipped = pack_innermost_dim_as_hex_string(
+ weight_tensor_pe_flipped, export_wdt, weight_width_padded, prefix=""
+ )
+ # add zeroes to pad out file to 1024 entries
+ weight_stream = weight_tensor_pe_flipped.flatten()
+ weight_stream = weight_stream.copy()
+ with open(weight_file_name, "w") as f:
+ for val in weight_stream:
+ f.write(val + "\n")
+ elif weight_file_mode == "decoupled_runtime":
+ # memstream axi-lite interface will map each mem line to
+ # one or multiple 32-bit words
+ weight_width = self.get_weightstream_width()
+ words_per_memwidth = 2 ** math.ceil(math.log2(weight_width / 32))
+ if words_per_memwidth < 1:
+ words_per_memwidth = 1
+ weight_width_padded = words_per_memwidth * 32
+ # first, pack and ensure padding to 32 bits
+ weight_tensor_pe_flipped = pack_innermost_dim_as_hex_string(
+ weight_tensor_pe_flipped, export_wdt, weight_width_padded, prefix=""
+ )
+ weight_stream = weight_tensor_pe_flipped.flatten()
+ weight_stream = weight_stream.copy()
+ with open(weight_file_name, "w") as f:
+ for val in weight_stream:
+ # split into groups of 8 hex digits (= 32 bits)
+ words_32b = textwrap.wrap(val, 8)
+ words_32b.reverse()
+ for word_32b in words_32b:
+ f.write(word_32b + "\n")
+ else:
+ raise Exception("Unknown weight_file_mode")
+
else:
- f_weights.write(
- "const BinaryWeights<1,{},{}> weights = ".format(
- self.get_nodeattr("PE"), self.calc_wmem()
+ raise Exception("Unknown weight_file_mode")
+
+ def generate_params(self, model, path):
+ mem_mode = self.get_nodeattr("mem_mode")
+ code_gen_dir = path
+ # weights, if not external
+ weights = model.get_initializer(self.onnx_node.input[1])
+ if mem_mode == "const":
+ # save hlslib-compatible weights in params.h
+ weight_filename = "{}/params.h".format(code_gen_dir)
+ self.make_weight_file(weights, "hls_header", weight_filename)
+ elif mem_mode == "decoupled" or mem_mode == "external":
+ weight_filename_sim = "{}/weights.npy".format(code_gen_dir)
+ # save decoupled weights for cppsim
+ self.make_weight_file(weights, "decoupled_npy", weight_filename_sim)
+ if mem_mode == "decoupled":
+ # also save weights as Verilog .dat file
+ # note that we provide two different .dat files, one for synth
+ # and one for synthesis. this is because URAM-based weights always
+ # need zero weights for synthesis, otherwise they get inferred
+ # as BRAM
+ weight_filename_rtl_synth = "{}/memblock_synth_0.dat".format(
+ code_gen_dir
+ )
+ weight_filename_rtl_sim = "{}/memblock_sim_0.dat".format(code_gen_dir)
+ # sim weights are always the true weights
+ self.make_weight_file(
+ weights, "decoupled_verilog_dat", weight_filename_rtl_sim
)
+ ram_style = self.get_nodeattr("ram_style")
+ if ram_style == "ultra":
+ # UltraRAM must have no memory initializer, or only zeroes
+ # otherwise BRAM will be inferred instead of URAM
+ # as a workaround we provide a zero-weight init here
+ synth_weights = np.zeros_like(weights, dtype=np.float32)
+ else:
+ synth_weights = weights
+ self.make_weight_file(
+ synth_weights, "decoupled_verilog_dat", weight_filename_rtl_synth
+ )
+ else:
+ raise Exception(
+ """Please set mem_mode to "const", "decoupled", or "external",
+ currently no other parameter value is supported!"""
)
- f_weights.write(weight_hls_code)
- f_weights.close()
# save thresholds in thresh.h
if len(self.onnx_node.input) > 2:
thresholds = model.get_initializer(self.onnx_node.input[2])
if thresholds is not None:
threshold_tensor = self.get_hls_compatible_threshold_tensor(thresholds)
+ # use UINT32 threshold export for bipolar times bipolar
+ inp_is_bipolar = self.get_input_datatype() == DataType["BIPOLAR"]
+ wt_is_bipolar = self.get_weight_datatype() == DataType["BIPOLAR"]
+ # reinterpret inp/wt as bipolar if bin_xnor_mode is iset
+ inp_is_binary = self.get_input_datatype() == DataType["BINARY"]
+ wt_is_binary = self.get_weight_datatype() == DataType["BINARY"]
+ bin_xnor_mode = self.get_nodeattr("binaryXnorMode") == 1
+ inp_is_bipolar = inp_is_bipolar or (inp_is_binary and bin_xnor_mode)
+ wt_is_bipolar = wt_is_bipolar or (wt_is_binary and bin_xnor_mode)
# get computed threshold datatype from attribute
tdt = DataType[self.get_nodeattr("accDataType")]
+
assert np.vectorize(tdt.allowed)(
threshold_tensor
).all(), "Thresholds in %s can't be expressed with type %s" % (
@@ -368,8 +583,11 @@ class VectorVectorActivation(HLSCustomOp):
# write thresholds into thresh.h
f_thresh = open("{}/thresh.h".format(code_gen_dir), "w")
tdt_hls = tdt.get_hls_datatype_str()
- odt = self.get_output_datatype()
- odt_hls = odt.get_hls_datatype_str()
+ # use binary to export bipolar activations
+ export_odt = self.get_output_datatype()
+ if self.get_output_datatype() == DataType["BIPOLAR"]:
+ export_odt = DataType["BINARY"]
+ odt_hls = export_odt.get_hls_datatype_str()
f_thresh.write(
"static ThresholdsActivation<{},{},{},{},{},{},{}> threshs \
= ".format(
@@ -387,6 +605,7 @@ class VectorVectorActivation(HLSCustomOp):
def execute_node(self, context, graph):
mode = self.get_nodeattr("exec_mode")
+ mem_mode = self.get_nodeattr("mem_mode")
node = self.onnx_node
# TODO ensure codegen dir exists
@@ -440,7 +659,28 @@ class VectorVectorActivation(HLSCustomOp):
inp = npy_to_rtlsim_input("{}/input_0.npy".format(code_gen_dir), idt, nbits)
super().reset_rtlsim(sim)
super().toggle_clk(sim)
- output = self.rtlsim(sim, inp)
+
+ if mem_mode == "external" or mem_mode == "decoupled":
+ wnbits = self.get_weightstream_width()
+ export_wdt = self.get_weight_datatype()
+ # we have converted bipolar weights to binary for export,
+ # so use it as such for weight generation
+ if self.get_weight_datatype() == DataType["BIPOLAR"]:
+ export_wdt = DataType["BINARY"]
+ wei = npy_to_rtlsim_input(
+ "{}/weights.npy".format(code_gen_dir), export_wdt, wnbits
+ )
+ dim_h, dim_w = self.get_nodeattr("Dim")
+ num_w_reps = dim_h * dim_w
+
+ io_dict = {
+ "inputs": {"in0": inp, "weights": wei * num_w_reps},
+ "outputs": {"out": []},
+ }
+ self.rtlsim_multi_io(sim, io_dict)
+ output = io_dict["outputs"]["out"]
+ else:
+ output = self.rtlsim(sim, inp)
odt = self.get_output_datatype()
target_bits = odt.bitwidth()
packed_bits = self.get_outstream_width()
@@ -466,6 +706,12 @@ class VectorVectorActivation(HLSCustomOp):
def global_includes(self):
self.code_gen_dict["$GLOBALS$"] = ['#include "weights.hpp"']
self.code_gen_dict["$GLOBALS$"] += ['#include "activations.hpp"']
+ mem_mode = self.get_nodeattr("mem_mode")
+ if mem_mode not in ["const", "decoupled", "external"]:
+ raise Exception(
+ """Please set mem_mode to "const", "decoupled", or "external",
+ currently no other parameter value is supported!"""
+ )
if self.calc_tmem() != 0:
self.code_gen_dict["$GLOBALS$"] += ['#include "thresh.h"']
@@ -474,6 +720,8 @@ class VectorVectorActivation(HLSCustomOp):
numReps = 1 * dim_h * dim_w
k_h, k_w = self.get_nodeattr("Kernel")
innerProdDim = k_h * k_w
+ mem_mode = self.get_nodeattr("mem_mode")
+
self.code_gen_dict["$DEFINES$"] = [
"""#define Channels1 {}\n #define InnerProdDim {}\n
#define SIMD1 1\n #define PE1 {}\n #define numReps {}""".format(
@@ -483,6 +731,11 @@ class VectorVectorActivation(HLSCustomOp):
numReps,
)
]
+ if mem_mode == "decoupled" or mem_mode == "external":
+ wdt = self.get_weight_datatype()
+ self.code_gen_dict["$DEFINES$"].append(
+ "#define WP1 {}\n".format(wdt.bitwidth())
+ )
def read_npy_data(self):
code_gen_dir = self.get_nodeattr("code_gen_dir_cppsim")
@@ -500,7 +753,23 @@ class VectorVectorActivation(HLSCustomOp):
% (packed_hls_type, elem_hls_type, elem_bits, npy_type, npy_in)
)
+ mem_mode = self.get_nodeattr("mem_mode")
+ if mem_mode == "decoupled" or mem_mode == "external":
+ wdt = self.get_weight_datatype()
+ elem_bits = wdt.bitwidth()
+ packed_bits = self.get_weightstream_width()
+ packed_hls_type = "ap_uint<%d>" % packed_bits
+ elem_hls_type = wdt.get_hls_datatype_str()
+ npy_type = "float"
+ npy_in = "%s/weights.npy" % code_gen_dir
+
+ self.code_gen_dict["$READNPYDATA$"].append(
+ 'npy2apintstream<%s, %s, %d, %s>("%s", weights, false, numReps);'
+ % (packed_hls_type, elem_hls_type, elem_bits, npy_type, npy_in)
+ )
+
def strm_decl(self):
+ mem_mode = self.get_nodeattr("mem_mode")
self.code_gen_dict["$STREAMDECLARATIONS$"] = []
self.code_gen_dict["$STREAMDECLARATIONS$"].append(
'hls::stream<ap_uint<{}>> in0 ("in0");'.format(self.get_instream_width())
@@ -508,8 +777,15 @@ class VectorVectorActivation(HLSCustomOp):
self.code_gen_dict["$STREAMDECLARATIONS$"].append(
'hls::stream<ap_uint<{}>> out ("out");'.format(self.get_outstream_width())
)
+ if mem_mode == "decoupled" or mem_mode == "external":
+ self.code_gen_dict["$STREAMDECLARATIONS$"].append(
+ 'hls::stream<ap_uint<{}>> weights ("weights");'.format(
+ self.get_weightstream_width()
+ )
+ )
def docompute(self):
+ mem_mode = self.get_nodeattr("mem_mode")
map_to_hls_mult_style = {
"auto": "ap_resource_dflt()",
"lut": "ap_resource_lut()",
@@ -521,16 +797,42 @@ class VectorVectorActivation(HLSCustomOp):
threshs = "PassThroughActivation<%s>()" % odtype_hls_str
else:
threshs = "threshs"
- self.code_gen_dict["$DOCOMPUTE$"] = [
- """Vector_Vector_Activate_Batch<Channels1, InnerProdDim, SIMD1, PE1, 1, {}, {}, {}>
- (in0, out, weights, {}, numReps, {});""".format(
- tmpl_args["TSrcI"],
- tmpl_args["TDstI"],
- tmpl_args["TWeightI"],
- threshs,
- map_to_hls_mult_style[self.get_nodeattr("resType")],
+
+ if mem_mode == "const":
+ self.code_gen_dict["$DOCOMPUTE$"] = [
+ """Vector_Vector_Activate_Batch<Channels1, InnerProdDim, SIMD1, PE1, 1, {}, {}, {}>
+ (in0, out, weights, {}, numReps, {});""".format(
+ tmpl_args["TSrcI"],
+ tmpl_args["TDstI"],
+ tmpl_args["TWeightI"],
+ threshs,
+ map_to_hls_mult_style[self.get_nodeattr("resType")],
+ )
+ ]
+ elif mem_mode == "decoupled" or mem_mode == "external":
+ wdt = self.get_weight_datatype()
+ if wdt == DataType["BIPOLAR"]:
+ export_wdt = DataType["BINARY"]
+ else:
+ export_wdt = wdt
+ wdtype_hls_str = export_wdt.get_hls_datatype_str()
+ self.code_gen_dict["$DOCOMPUTE$"] = [
+ """{}<Channels1, InnerProdDim, SIMD1, PE1, 1, {}, {}, {}, {}>
+ (in0, out, weights, {}, numReps, {});""".format(
+ "Vector_Vector_Activate_Stream_Batch",
+ tmpl_args["TSrcI"],
+ tmpl_args["TDstI"],
+ tmpl_args["TWeightI"],
+ wdtype_hls_str,
+ threshs,
+ map_to_hls_mult_style[self.get_nodeattr("resType")],
+ )
+ ]
+ else:
+ raise Exception(
+ """Please set mem_mode to "const", "decoupled", or "external",
+ currently no other parameter value is supported!"""
)
- ]
def dataoutstrm(self):
code_gen_dir = self.get_nodeattr("code_gen_dir_cppsim")
@@ -561,17 +863,38 @@ class VectorVectorActivation(HLSCustomOp):
self.code_gen_dict["$SAVEASCNPY$"] = []
def blackboxfunction(self):
- self.code_gen_dict["$BLACKBOXFUNCTION$"] = [
- """void {}(hls::stream<ap_uint<{}>> &in0,
- hls::stream<ap_uint<{}>> &out
- )""".format(
- self.onnx_node.name,
- self.get_instream_width(),
- self.get_outstream_width(),
+ mem_mode = self.get_nodeattr("mem_mode")
+ if mem_mode == "const":
+ self.code_gen_dict["$BLACKBOXFUNCTION$"] = [
+ """void {}(hls::stream<ap_uint<{}>> &in0,
+ hls::stream<ap_uint<{}>> &out
+ )""".format(
+ self.onnx_node.name,
+ self.get_instream_width(),
+ self.get_outstream_width(),
+ )
+ ]
+ elif mem_mode == "decoupled" or mem_mode == "external":
+ self.code_gen_dict["$BLACKBOXFUNCTION$"] = [
+ """void {}(
+ hls::stream<ap_uint<{}>> &in0,
+ hls::stream<ap_uint<{}>> &weights,
+ hls::stream<ap_uint<{}>> &out
+ )""".format(
+ self.onnx_node.name,
+ self.get_instream_width(),
+ self.get_weightstream_width(),
+ self.get_outstream_width(),
+ )
+ ]
+ else:
+ raise Exception(
+ """Please set mem_mode to "const" or "decoupled", currently no other
+ parameter value is supported!"""
)
- ]
def pragmas(self):
+ mem_mode = self.get_nodeattr("mem_mode")
self.code_gen_dict["$PRAGMAS$"] = [
"#pragma HLS INTERFACE axis port=in0 name=in0_" + self.hls_sname()
]
@@ -593,12 +916,30 @@ class VectorVectorActivation(HLSCustomOp):
"#pragma HLS INTERFACE ap_ctrl_none port=return"
)
- self.code_gen_dict["$PRAGMAS$"].append('#include "params.h"')
- # the weight tensor is ap_uint<ch*prec> [PE][WMEM]
- # partition for parallel access along the PE dimension (dim 1)
- self.code_gen_dict["$PRAGMAS$"].append(
- ("#pragma HLS ARRAY_PARTITION variable=weights.m_weights " "complete dim=1")
- )
+ if mem_mode == "const":
+ self.code_gen_dict["$PRAGMAS$"].append('#include "params.h"')
+ # the weight tensor is ap_uint<ch*prec> [PE][WMEM]
+ # partition for parallel access along the PE dimension (dim 1)
+ self.code_gen_dict["$PRAGMAS$"].append(
+ (
+ "#pragma HLS ARRAY_PARTITION variable=weights.m_weights "
+ "complete dim=1"
+ )
+ )
+ elif mem_mode == "decoupled" or mem_mode == "external":
+ self.code_gen_dict["$PRAGMAS$"].append(
+ "#pragma HLS INTERFACE axis port=weights name=weights_"
+ + self.hls_sname()
+ )
+ self.code_gen_dict["$PRAGMAS$"].append(
+ "#pragma HLS stream depth=8 variable=weights"
+ )
+ else:
+ raise Exception(
+ """Please set mem_mode to "const", "decoupled", or external,
+ currently no other parameter value is supported!"""
+ )
+
if self.calc_tmem() != 0:
# TODO find a better way of checking for no pregenerated thresholds
self.code_gen_dict["$PRAGMAS$"].append(
@@ -614,6 +955,157 @@ class VectorVectorActivation(HLSCustomOp):
)
)
+ def get_verilog_top_module_intf_names(self):
+ intf_names = super().get_verilog_top_module_intf_names()
+ mem_mode = self.get_nodeattr("mem_mode")
+ sname = self.hls_sname()
+ if mem_mode == "external":
+ intf_names["s_axis"].append(
+ ("weights_" + sname, self.get_weightstream_width_padded())
+ )
+ if mem_mode == "decoupled":
+ # only expose axilite interface if attribute is set
+ runtime_writable = self.get_nodeattr("runtime_writeable_weights") == 1
+ if runtime_writable:
+ intf_names["axilite"] = ["s_axilite"]
+ return intf_names
+
+ def code_generation_ipi(self):
+ cmd = []
+ # add streamer if needed
+ mem_mode = self.get_nodeattr("mem_mode")
+ if mem_mode == "decoupled":
+ runtime_writable = self.get_nodeattr("runtime_writeable_weights") == 1
+ if self.get_nodeattr("ram_style") == "ultra":
+ assert (
+ runtime_writable == 1
+ ), "Layer with URAM weights must have runtime_writeable_weights=1"
+ node_name = self.onnx_node.name
+ sname = self.hls_sname()
+ # create a hierarchy for this layer, with the same port names
+ clk_name = self.get_verilog_top_module_intf_names()["clk"][0]
+ rst_name = self.get_verilog_top_module_intf_names()["rst"][0]
+ dout_name = self.get_verilog_top_module_intf_names()["m_axis"][0][0]
+ din_name = self.get_verilog_top_module_intf_names()["s_axis"][0][0]
+ cmd.append("create_bd_cell -type hier %s" % node_name)
+ cmd.append("create_bd_pin -dir I -type clk /%s/%s" % (node_name, clk_name))
+ cmd.append("create_bd_pin -dir I -type rst /%s/%s" % (node_name, rst_name))
+ cmd.append(
+ "create_bd_intf_pin -mode Master "
+ "-vlnv xilinx.com:interface:axis_rtl:1.0 /%s/%s"
+ % (node_name, dout_name)
+ )
+ cmd.append(
+ "create_bd_intf_pin -mode Slave "
+ "-vlnv xilinx.com:interface:axis_rtl:1.0 /%s/%s" % (node_name, din_name)
+ )
+ # instantiate the hls ip
+ cmd.append(
+ "create_bd_cell -type ip -vlnv %s /%s/%s"
+ % (self.get_nodeattr("ip_vlnv"), node_name, node_name)
+ )
+ # instantiate a streamer and connect it to the HLS IP
+ strm_vlnv = "xilinx.com:user:memstream:1.0"
+ strm_inst = node_name + "_wstrm"
+ cmd.append(
+ "create_bd_cell -type ip -vlnv %s /%s/%s"
+ % (strm_vlnv, node_name, strm_inst)
+ )
+ cmd.append(
+ "set_property -dict [list "
+ "CONFIG.NSTREAMS {1} "
+ "CONFIG.MEM_DEPTH {%d} "
+ "CONFIG.MEM_WIDTH {%d} "
+ "CONFIG.MEM_INIT {%s} "
+ "CONFIG.RAM_STYLE {%s} "
+ "CONFIG.STRM0_DEPTH {%d} "
+ "CONFIG.STRM0_WIDTH {%d} "
+ "CONFIG.STRM0_OFFSET {0} "
+ "] [get_bd_cells /%s/%s]"
+ % (
+ self.calc_wmem(),
+ self.get_weightstream_width_padded(),
+ self.get_nodeattr("code_gen_dir_ipgen") + "/",
+ self.get_nodeattr("ram_style"),
+ self.calc_wmem(),
+ self.get_weightstream_width_padded(),
+ node_name,
+ strm_inst,
+ )
+ )
+ cmd.append(
+ "connect_bd_intf_net [get_bd_intf_pins %s/%s/m_axis_0] "
+ "[get_bd_intf_pins %s/%s/weights_%s]"
+ % (node_name, strm_inst, node_name, node_name, sname)
+ )
+ cmd.append(
+ "connect_bd_net [get_bd_pins %s/%s] [get_bd_pins %s/%s/aresetn]"
+ % (node_name, rst_name, node_name, strm_inst)
+ )
+ cmd.append(
+ "connect_bd_net [get_bd_pins %s/%s] [get_bd_pins %s/%s/aclk]"
+ % (node_name, clk_name, node_name, strm_inst)
+ )
+ cmd.append(
+ "connect_bd_net [get_bd_pins %s/%s] [get_bd_pins %s/%s/%s]"
+ % (node_name, rst_name, node_name, node_name, rst_name)
+ )
+ cmd.append(
+ "connect_bd_net [get_bd_pins %s/%s] [get_bd_pins %s/%s/%s]"
+ % (node_name, clk_name, node_name, node_name, clk_name)
+ )
+ cmd.append(
+ "connect_bd_intf_net [get_bd_intf_pins %s/%s] "
+ "[get_bd_intf_pins %s/%s/%s]"
+ % (node_name, din_name, node_name, node_name, din_name)
+ )
+ cmd.append(
+ "connect_bd_intf_net [get_bd_intf_pins %s/%s] "
+ "[get_bd_intf_pins %s/%s/%s]"
+ % (node_name, dout_name, node_name, node_name, dout_name)
+ )
+ if runtime_writable:
+ # expose axi lite interface for writeable weights
+ axilite_name = self.get_verilog_top_module_intf_names()["axilite"][0]
+ cmd.append(
+ "create_bd_intf_pin -mode Slave "
+ "-vlnv xilinx.com:interface:aximm_rtl:1.0 /%s/%s"
+ % (node_name, axilite_name)
+ )
+ cmd.append(
+ "connect_bd_intf_net [get_bd_intf_pins %s/%s] "
+ "[get_bd_intf_pins %s/%s/%s]"
+ % (node_name, axilite_name, node_name, strm_inst, axilite_name)
+ )
+ # TODO calculate and pass in segment size here
+ cmd.append("assign_bd_address")
+ cmd.append("save_bd_design")
+ elif mem_mode == "const" or mem_mode == "external":
+ # base class impl sufficient for const/external modes
+ return super().code_generation_ipi()
+ else:
+ raise Exception("Unrecognized mem_mode for VectorVectorActivation")
+ return cmd
+
+ def uram_estimation(self):
+ P = self.get_nodeattr("PE")
+ Q = 1
+ wdt = self.get_weight_datatype()
+ W = wdt.bitwidth()
+ omega = self.calc_wmem()
+ mem_width = Q * W * P
+ mmode = self.get_nodeattr("mem_mode")
+ mstyle = self.get_nodeattr("ram_style")
+ if (
+ (mmode == "decoupled" and mstyle != "ultra")
+ or (mmode == "const" and self.calc_wmem() <= 128)
+ or (mmode == "external")
+ ):
+ return 0
+ width_multiplier = math.ceil(mem_width / 72)
+ depth_multiplier = math.ceil(omega / 4096)
+ return width_multiplier * depth_multiplier
+
def bram_estimation(self):
"""Calculates resource estimation for BRAM"""
# TODO add in/out FIFO contributions
@@ -624,7 +1116,13 @@ class VectorVectorActivation(HLSCustomOp):
# assuming SDP mode RAMB18s (see UG573 Table 1-10)
# since this is HLS memory, not using the full width of a BRAM
# assuming memories up to 128 deep get implemented in LUTs
- if self.calc_wmem() <= 128:
+ mmode = self.get_nodeattr("mem_mode")
+ mstyle = self.get_nodeattr("ram_style")
+ if (
+ (mmode == "decoupled" and mstyle in ["distributed", "ultra"])
+ or (mmode == "const" and self.calc_wmem() <= 128)
+ or (mmode == "external")
+ ):
return 0
if W == 1:
@@ -671,8 +1169,12 @@ class VectorVectorActivation(HLSCustomOp):
c0 = 300
c1 = 1.1
c2 = 0
- if self.calc_wmem() <= 128:
- c2 = P * W * math.ceil(self.calc_wmem() / 64)
+ mmode = self.get_nodeattr("mem_mode")
+ mstyle = self.get_nodeattr("ram_style")
+ if (mmode == "decoupled" and mstyle == "distributed") or (
+ mmode == "const" and self.calc_wmem() <= 128
+ ):
+ c2 = (P * W) * math.ceil(self.calc_wmem() / 64)
# multiplication
res_type = self.get_nodeattr("resType")
@@ -710,6 +1212,25 @@ class VectorVectorActivation(HLSCustomOp):
mult_dsp = 0
return int(mult_dsp)
+ def get_weightstream_width(self):
+ """Returns weight stream width. Used only in decoupled mode."""
+ if (
+ self.get_nodeattr("mem_mode") == "decoupled"
+ or self.get_nodeattr("mem_mode") == "external"
+ ):
+ pe = self.get_nodeattr("PE")
+ wp = self.get_weight_datatype().bitwidth()
+ w_width = pe * wp
+ return w_width
+ else:
+ return 0
+
+ def get_weightstream_width_padded(self):
+ """Returns weight stream width padded to a multiple of 8. This is required
+ by the AXI Stream spec. Used in decoupled mode."""
+ weight_width = self.get_weightstream_width()
+ return roundup_to_integer_multiple(weight_width, 8)
+
def get_op_and_param_counts(self):
k_h, k_w = self.get_nodeattr("Kernel")
fm = self.get_nodeattr("Channels")
diff --git a/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py b/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py
index 381bed143eb5adc5a049cb5b82d74c2af936552d..2c377f23d765ceffc3b8fad83fe0c329de3016df 100644
--- a/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py
+++ b/src/finn/transformation/fpgadataflow/convert_to_hls_layers.py
@@ -910,6 +910,10 @@ class InferVectorVectorActivation(Transformation):
a depthwise convolution. Any immediately following MultiThreshold
layers will also be absorbed into the VVAU."""
+ def __init__(self, mem_mode="const"):
+ super().__init__()
+ self.mem_mode = mem_mode
+
def apply(self, model):
graph = model.graph
node_ind = 0
@@ -1010,6 +1014,7 @@ class InferVectorVectorActivation(Transformation):
ActVal=actval,
noActivation=0,
name="VectorVectorActivation_" + n.name,
+ mem_mode=self.mem_mode,
)
graph.node.insert(node_ind, new_node)
# remove old nodes
diff --git a/tests/fpgadataflow/test_fpgadataflow_vvau.py b/tests/fpgadataflow/test_fpgadataflow_vvau.py
index c48448787d8a3bb926c1e94850be6e99e8c106d3..03ddb1286320b8178276ea53082095106a43d7a1 100644
--- a/tests/fpgadataflow/test_fpgadataflow_vvau.py
+++ b/tests/fpgadataflow/test_fpgadataflow_vvau.py
@@ -75,7 +75,19 @@ def _calculate_dot_prod_range(dt_a, dt_b, len):
def _make_single_vvau_modelwrapper(
- W, pe, k_h, k_w, channels, dim_h, dim_w, wdt, idt, odt, T=None, tdt=None
+ W,
+ pe,
+ k_h,
+ k_w,
+ channels,
+ dim_h,
+ dim_w,
+ wdt,
+ idt,
+ odt,
+ T=None,
+ tdt=None,
+ mem_mode="const",
):
in_shape = [1, dim_h, dim_w, k_h * k_w * channels] # [N, H, W, K*K*CH]
out_shape = [
@@ -113,6 +125,7 @@ def _make_single_vvau_modelwrapper(
weightDataType=wdt.name,
outputDataType=odt.name,
noActivation=no_act,
+ mem_mode=mem_mode,
)
graph = helper.make_graph(
@@ -140,7 +153,7 @@ def prepare_inputs(input_tensor):
return {"inp": input_tensor}
-# mem_mode: const or decoupled
+# input datatype
@pytest.mark.parametrize("idt", [DataType["UINT4"], DataType["UINT8"]])
# weight datatype
@pytest.mark.parametrize("wdt", [DataType["INT4"]])
@@ -156,13 +169,15 @@ def prepare_inputs(input_tensor):
@pytest.mark.parametrize("k_w", [3, 1])
# Number of input and output channels
@pytest.mark.parametrize("channels", [3, 4])
+# memory mode
+@pytest.mark.parametrize("mem_mode", ["const", "decoupled"])
# execution mode
@pytest.mark.parametrize("exec_mode", ["cppsim", "rtlsim"])
@pytest.mark.fpgadataflow
@pytest.mark.slow
@pytest.mark.vivado
def test_fpgadataflow_vvau(
- idt, wdt, act, pe, dim_h, dim_w, k_h, k_w, channels, exec_mode
+ idt, wdt, act, pe, dim_h, dim_w, k_h, k_w, channels, mem_mode, exec_mode
):
if pe == "channels":
pe = channels
@@ -198,7 +213,7 @@ def test_fpgadataflow_vvau(
tdt = DataType["INT32"]
model = _make_single_vvau_modelwrapper(
- W, pe, k_h, k_w, channels, dim_h, dim_w, wdt, idt, odt, T, tdt
+ W, pe, k_h, k_w, channels, dim_h, dim_w, wdt, idt, odt, T, tdt, mem_mode
)
if exec_mode == "cppsim":