Skip to content

Add 1x1 specialization for conv with NCHW order#9671

Closed
xiaomengy wants to merge 1 commit intopytorch:masterfrom
xiaomengy:export-D8944686
Closed

Add 1x1 specialization for conv with NCHW order#9671
xiaomengy wants to merge 1 commit intopytorch:masterfrom
xiaomengy:export-D8944686

Conversation

@xiaomengy
Copy link
Contributor

Summary: Add 1x1 specialization for conv with NCHW order

Differential Revision: D8944686

@xiaomengy
Copy link
Contributor Author

xiaomengy commented Jul 21, 2018

Some benchmark on CPU with N = 128, C = 128, H = 32, W = 32, M = 128, G = 4, kernel = 1
original impl: about 250ms
current impl: about 10ms

Summary:
Pull Request resolved: pytorch#9671

Add 1x1 specialization for conv with NCHW order

Differential Revision: D8944686

fbshipit-source-id: 6a96813f3d9f0d47799391172214e3bf9d938f22
Copy link
Contributor

@yinghai yinghai left a comment

Choose a reason for hiding this comment

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

Could you provide some comment on what the difference is between batchedgemm and batchedstridedgemm?

@@ -255,7 +255,7 @@ class BatchMatMulOp final : public Operator<Context> {

// TODO(T23893772): doing this in a loop is likely going to be slow on GPU

This comment was marked as off-topic.

This comment was marked as off-topic.

Copy link
Member

@houseroad houseroad left a comment

Choose a reason for hiding this comment

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

LGTM.

If using OSS, we can always do git clang-format on the staging files.
Mercurial may also support this.

@xiaomengy xiaomengy deleted the export-D8944686 branch July 24, 2018 01:56
jramseyer pushed a commit to jramseyer/pytorch that referenced this pull request Jul 30, 2018
Summary:
Pull Request resolved: pytorch#9671

Add 1x1 specialization for conv with NCHW order

Reviewed By: houseroad

Differential Revision: D8944686

fbshipit-source-id: 94bf44f69498b1934b7dfff4c0e989342c7bb61c
goodlux pushed a commit to goodlux/pytorch that referenced this pull request Aug 15, 2018
Summary:
Pull Request resolved: pytorch#9671

Add 1x1 specialization for conv with NCHW order

Reviewed By: houseroad

Differential Revision: D8944686

fbshipit-source-id: 94bf44f69498b1934b7dfff4c0e989342c7bb61c
@ezyang ezyang added the merged label Jun 26, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants