Skip to content

[GNA] Fix cases then const blob precision is not FP32/FP16 - master #1020

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dorloff merged 1 commit into from
Jun 22, 2020
Merged

[GNA] Fix cases then const blob precision is not FP32/FP16 - master #1020

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 9 additions & 1 deletion inference-engine/src/gna_plugin/frontend/scale_factor_calc.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -199,9 +199,17 @@ class ScaleFactorPerLayer<InferenceEngine::CNNLayer *> {

if (cnnLayer->type == "Const") {
auto blob = cnnLayer->blobs["custom"];
if (blob->getTensorDesc().getPrecision() == InferenceEngine::Precision::FP16) {
auto blob_precision = blob->getTensorDesc().getPrecision();

if (blob_precision != InferenceEngine::Precision::FP32 && blob_precision != InferenceEngine::Precision::FP16) {
quant->_dst_quant.scale = 1.0f;
return true;
}

if (blob_precision == InferenceEngine::Precision::FP16) {
blob = make_fp32_blob(blob);
}

auto max_val = std::numeric_limits<float>::min();
auto min_val = std::numeric_limits<float>::max();

Expand Down
2 changes: 1 addition & 1 deletion ...tests/functional/plugin/gna/shared_tests_instances/subgraph_tests/concat_quantization.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ const std::vector<std::map<std::string, std::string>> configs = {
}
};

INSTANTIATE_TEST_CASE_P(ConcatQuantization, ConcatQuantization,
INSTANTIATE_TEST_CASE_P(smoke_ConcatQuantization, ConcatQuantization,
Copy link
Copy Markdown
Contributor

@iefode iefode Jun 19, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smoke prefix is using as pre-commit gtest filter in old functional tests in CI and will be reused in the future for new functional tests (it is not enabled yet because running of func tests is very fast). I suppose, we shoul not remove smoke_

Copy link
Copy Markdown
Contributor

@esmirno esmirno Jun 19, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@iefode i dint get - we should remove it or not? I though we need to add this to enable tests in precomit.

Copy link
Copy Markdown
Contributor

@dorloff dorloff Jun 22, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@iefode, please clarify, what should we do for a test to be enabled in precommit? Or do you mean this test should have been running already in precommit?

::testing::Combine(
::testing::ValuesIn(netPrecisions),
::testing::Values(CommonTestUtils::DEVICE_GNA),
Expand Down