[oneDNN] Pool softmax and LRN access to cache optimized #32922
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
LRN added - lint
|
Thanks for your contribution! |
jczaja
requested review from
jakpiase,
arogowie-intel,
lidanqing-intel and
arlesniak
arogowie-intel
previously approved these changes
May 17, 2021
jakpiase
suggested changes
May 17, 2021
| class LRNMKLDNNHandler : public platform::MKLDNNHandlerT<T, mkldnn::lrn_forward, | ||
| mkldnn::lrn_backward> { | ||
| public: | ||
| LRNMKLDNNHandler(const paddle::framework::ExecutionContext& ctx, |
There was a problem hiding this comment.
Suggested change
| LRNMKLDNNHandler(const paddle::framework::ExecutionContext& ctx, | |
| LRNMKLDNNHandler(const framework::ExecutionContext& ctx, |
| const float k = ctx.Attr<float>("k"); | ||
| bool is_test = ctx.Attr<bool>("is_test"); | ||
|
|
||
| auto dims = paddle::framework::vectorize(input->dims()); |
There was a problem hiding this comment.
Suggested change
| auto dims = paddle::framework::vectorize(input->dims()); | |
| auto dims = framework::vectorize(input->dims()); |
| } | ||
| } | ||
|
|
||
| LRNMKLDNNHandler(const paddle::framework::ExecutionContext& ctx, |
There was a problem hiding this comment.
Suggested change
| LRNMKLDNNHandler(const paddle::framework::ExecutionContext& ctx, | |
| LRNMKLDNNHandler(const framework::ExecutionContext& ctx, |
| } | ||
|
|
||
| LRNMKLDNNHandler(const paddle::framework::ExecutionContext& ctx, | ||
| const platform::MKLDNNDeviceContext& dev_ctx, |
There was a problem hiding this comment.
Suggested change
| const platform::MKLDNNDeviceContext& dev_ctx, | |
| const MKLDNNDeviceContext& dev_ctx, |
because you have "using paddle::platform::MKLDNNDeviceContext;" in line 30
| const float beta = ctx.Attr<float>("beta"); | ||
| const float k = ctx.Attr<float>("k"); | ||
|
|
||
| auto dims = paddle::framework::vectorize<int64_t>(in_x->dims()); |
There was a problem hiding this comment.
Suggested change
| auto dims = paddle::framework::vectorize<int64_t>(in_x->dims()); | |
| auto dims = framework::vectorize<int64_t>(in_x->dims()); |
arogowie-intel
approved these changes
May 18, 2021
|
@luotao1 Could you please start your review? |
luotao1
approved these changes
May 19, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR types
Function optimization
PR changes
OPs
Describe
This PR reduces some mutexes and references to oneDNN cache for three operators(more to come). General idea is that instead of relying on cache to get FWD_PD when BWD_PD is to be created we recreate FWD_PD from scratch in BWD_PD if the one in cache does not exist.