[GPU] Support pointwise floor and ceil in GEMM fusions.#11912
Closed
[GPU] Support pointwise floor and ceil in GEMM fusions.#11912
Conversation
Member
|
The change looks good, though the needed Triton change would need to be included as you mentioned. I will cherry-pick your change internally so that it reflects on OpenXLA/Triton and then you can merge this change. |
Member
|
@sergachev The required dependency change now applies on top of OpenXLA/xla so this should be unblocked. |
Moerafaat
approved these changes
May 10, 2024
4a2e65e to
19c87d5
Compare
copybara-service bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
May 10, 2024
Imported from GitHub PR openxla/xla#11912 For this to work with Triton openxla/triton will need to include triton-lang/triton@62706e8. Copybara import of the project: -- 37bc5f77e0eb1772ec0ae1430d54a48c626840be by Ilia Sergachev <isergachev@nvidia.com>: [GPU] Triton: support elementwise floor and ceil. -- 19c87d539078405122cdc9c9cfb2789e696317cc by Ilia Sergachev <isergachev@nvidia.com>: [GPU] cuDNN fusions: support elementwise floor and ceil. Merging this change closes #11912 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#11912 from openxla:pointwise_floor_ceil 19c87d539078405122cdc9c9cfb2789e696317cc PiperOrigin-RevId: 632445295
Contributor
Author
|
//xla/service/gpu:softmax_rewriter_triton_test fails, I'm looking at that |
copybara-service bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
May 10, 2024
Imported from GitHub PR openxla/xla#11912 For this to work with Triton openxla/triton will need to include triton-lang/triton@62706e8. Copybara import of the project: -- 37bc5f77e0eb1772ec0ae1430d54a48c626840be by Ilia Sergachev <isergachev@nvidia.com>: [GPU] Triton: support elementwise floor and ceil. -- 19c87d539078405122cdc9c9cfb2789e696317cc by Ilia Sergachev <isergachev@nvidia.com>: [GPU] cuDNN fusions: support elementwise floor and ceil. Merging this change closes #11912 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#11912 from openxla:pointwise_floor_ceil 19c87d539078405122cdc9c9cfb2789e696317cc PiperOrigin-RevId: 632445295
Member
|
Thanks @sergachev. Please ping me when you fix the issue and I will merge the PR internally. |
19c87d5 to
9731974
Compare
Contributor
Author
|
@dimitar-asenov it's fixed now |
Member
|
@sergachev Actually, one more thing: Could you please add the new ops to this list: https://github.com/openxla/xla/blob/main/xla/service/gpu/triton_support_test.cc#L193-L201 |
9731974 to
7a21cb7
Compare
Contributor
Author
|
OK, I added them, but could not test - this test does not build in OSS because of the known problem with bazel backend tags; if I hack them it fails on unknown EXPECT_OK and so on. |
copybara-service bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
May 10, 2024
Imported from GitHub PR openxla/xla#11912 For this to work with Triton openxla/triton will need to include triton-lang/triton@62706e8. Copybara import of the project: -- 37efe4abc9caae6033ac1aabcf1799f2f3c1b0c2 by Ilia Sergachev <isergachev@nvidia.com>: [GPU] Triton: support elementwise floor and ceil. -- 7a21cb770cdb08bdb4504fb23c4dc1d31126c01f by Ilia Sergachev <isergachev@nvidia.com>: [GPU] cuDNN fusions: support elementwise floor and ceil. Merging this change closes #11912 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#11912 from openxla:pointwise_floor_ceil 7a21cb770cdb08bdb4504fb23c4dc1d31126c01f PiperOrigin-RevId: 632445295
copybara-service bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
May 10, 2024
Imported from GitHub PR openxla/xla#11912 For this to work with Triton openxla/triton will need to include triton-lang/triton@62706e8. Copybara import of the project: -- 37efe4abc9caae6033ac1aabcf1799f2f3c1b0c2 by Ilia Sergachev <isergachev@nvidia.com>: [GPU] Triton: support elementwise floor and ceil. -- 7a21cb770cdb08bdb4504fb23c4dc1d31126c01f by Ilia Sergachev <isergachev@nvidia.com>: [GPU] cuDNN fusions: support elementwise floor and ceil. Merging this change closes #11912 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#11912 from openxla:pointwise_floor_ceil 7a21cb770cdb08bdb4504fb23c4dc1d31126c01f PiperOrigin-RevId: 632445295
copybara-service bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
May 10, 2024
Imported from GitHub PR openxla/xla#11912 For this to work with Triton openxla/triton will need to include triton-lang/triton@62706e8. Copybara import of the project: -- 37efe4abc9caae6033ac1aabcf1799f2f3c1b0c2 by Ilia Sergachev <isergachev@nvidia.com>: [GPU] Triton: support elementwise floor and ceil. -- 7a21cb770cdb08bdb4504fb23c4dc1d31126c01f by Ilia Sergachev <isergachev@nvidia.com>: [GPU] cuDNN fusions: support elementwise floor and ceil. Merging this change closes #11912 PiperOrigin-RevId: 632495980
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 participants
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.
For this to work with Triton openxla/triton will need to include triton-lang/triton@62706e8.