IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[CPU] Optimize processing for FQ + Sum + FQ post ops pattern

Browse Source
This commit is contained in:
dmitrygo
2022-03-22 14:04:22 +03:00
committed by Luo Cheng
parent af4731a1f1
commit eae4782284
1 changed files with 28 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
src/plugins/intel_cpu/src/nodes/conv.cpp
View File

@@ -23,6 +23,7 @@
#include "memory_desc/dnnl_blocked_memory_desc.h"
#include "utils/cpu_utils.hpp"
#include <common/primitive_hashing_utils.hpp>
#include <cpu/cpu_primitive.hpp>
using namespace dnnl;
using namespace InferenceEngine;
@@ -646,6 +647,33 @@ void Convolution::setPostOps(dnnl::primitive_attr &attr, const VectorDims &dims,
}
}
if (node == fusedWith[fusedWith.size() - 1] &&
outputDataType == memory::data_type::u8 &&
fakeQuantizeNode->getAlgorithm() == Algorithm::FQQuantization &&
ops.len() == 1 && ops.kind(0) == primitive::kind::sum
/*levels == 256*/) {
auto &cl = fakeQuantizeNode->getCropLow();
auto &isc = fakeQuantizeNode->getInputScale();
auto &ish = fakeQuantizeNode->getInputShift();
if (std::all_of(cl.cbegin(), cl.cend(), [](float val) { return val == 0.0f; }) &&
std::all_of(isc.cbegin(), isc.cend(), [&](float val) { return val == isc[0]; }) &&
std::all_of(ish.cbegin(), ish.cend(), [&](float val) { return val == 0; })) {
std::vector<float> outScales;
int mask = 1 << 1;
attr.get_output_scales(mask, outScales);
for (int j = 0; j < outScales.size(); j++) {
outScales[j] *= isc[0];
}
attr.set_output_scales(mask, outScales);
ops.get()->entry_[0].sum.scale = isc[0];
continue;
}
}
fakeQuantizeNode->appendBinPostOpsOptimized(ops, getBinPostOpShape(), postOpsArgs,
node == fusedWith[fusedWith.size() - 1], outputDataType);