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

brushed thresholds (to cover rnn_ cases when there are convs)

Browse Source
This commit is contained in:
myshevts 2021-04-06 16:07:43 +03:00
parent 2003fa3f46
commit c4df94d42d
1 changed files with 16 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
inference-engine/src/mkldnn_plugin/mkldnn_plugin.cpp
View File

@ -622,32 +622,21 @@ Engine::LoadExeNetworkImpl(const InferenceEngine::CNNNetwork &network, const std
// Hybrid specific // Hybrid specific
// etc // etc
int num_streams; int num_streams;
bool considerNonLimited = false; if (NetworkToleranceForLowCache.maxMemTolerance == NetworkPerfStats::memThresholdUnknown) {
if (NetworkToleranceForLowCache.maxMemTolerance > NetworkPerfStats::memThresholdNotLimited) {
std::cout << " case 1.0" <<std::endl;
considerNonLimited = true;
}
if ((NetworkToleranceForLowCache.ratio_compute_convs == NetworkPerfStats::ALL) if ((NetworkToleranceForLowCache.ratio_compute_convs == NetworkPerfStats::ALL)
|| (NetworkToleranceForLowCache.ratio_compute_deconvs == NetworkPerfStats::ALL)) { || (NetworkToleranceForLowCache.ratio_compute_deconvs == NetworkPerfStats::ALL)) {
std::cout << " case 1.1" <<std::endl; std::cout << " case 1.1" <<std::endl;
considerNonLimited = true;
}
if (hasAVX512() && NetworkToleranceForLowCache.maxMemTolerance > NetworkPerfStats::memThresholdAssumeLimitedAVX512
&& (NetworkToleranceForLowCache.ratio_mem_limited_convs <= NetworkPerfStats::memLimitedRatioThresholdAVX512)) {
std::cout << " case 1.2" <<std::endl;
considerNonLimited = true;
}
if (considerNonLimited) {
num_streams = num_cores; num_streams = num_cores;
std::cout << "case 1" <<std::endl; } else {
// if (NetworkToleranceForLowCache.ratio_compute_convs > NetworkPerfStats::memComputeConvs) {
// config[PluginConfigParams::KEY_CPU_THREADS_NUM] = std::to_string(
// std::thread::hardware_concurrency());
// std::cout << "ENABLING HT!!!" << std::endl;
// }
} else if (NetworkToleranceForLowCache.maxMemTolerance == NetworkPerfStats::memThresholdUnknown) {
num_streams = default_num_streams; num_streams = default_num_streams;
std::cout << "case 0" <<std::endl; std::cout << "case 0" <<std::endl;
}
} else if ((NetworkToleranceForLowCache.maxMemTolerance > NetworkPerfStats::memThresholdNotLimited)
|| (hasAVX512()
&& NetworkToleranceForLowCache.maxMemTolerance > NetworkPerfStats::memThresholdAssumeLimitedAVX512
&& NetworkToleranceForLowCache.ratio_mem_limited_convs <= NetworkPerfStats::memLimitedRatioThresholdAVX512)) {
std::cout << " case 1.0 or 1.2" <<std::endl;
num_streams = num_cores;
} else if (NetworkToleranceForLowCache.maxMemTolerance > NetworkPerfStats::memThresholdAssumeLimited) { } else if (NetworkToleranceForLowCache.maxMemTolerance > NetworkPerfStats::memThresholdAssumeLimited) {
num_streams = std::max(default_num_streams, num_streams_default_not_ht); num_streams = std::max(default_num_streams, num_streams_default_not_ht);
std::cout << "case 2" <<std::endl; std::cout << "case 2" <<std::endl;