[midend/lib/Conversion/MatMulOptimization] add batchmatmul vectorization pass and add relevant examples and tests. #427
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
…ion pass and add relevant examples and tests.
…ion pass and examples.
…ion pass and examples.
…ion pass and examples.
linalg-batchmatmul-f32.mlir: add operator execution time calculation; batchmatmul-vectorisation.mlir: handwritten mlir file makefile: update related commands. midend/lib/Conversion/MatMulOptimization/: update BatchMatMulOptimize.cpp implementation of vectorisation, now using the standard method in buddy. tests/Conversion/: batchmatmul-vectorisation.mlir: a test file. Before vectorisation optimization: 0.0043869, after vectorization optimization: 0.00018692 (with vectorization size of 128), optimization is about 23.5x. After vectorization size adjustment to 256: 0.000138998, optimization is about 31.6x. Compared to the previous version of BatchMatMulOptimize.cpp: before the update: 0.000586033, after the update: 0.000358105, the optimisation is improved by 1.5x. In addition, the original implementation will even have a negative optimisation.
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Outdated suggestions cannot be applied.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.
Add batchmatmul vectorization pass and add relevant examples and tests.
examples/BuddyMatmul/:
linalg-batchmatmul-f32.mlir: add operator execution time calculation;
batchmatmul-vectorisation.mlir: handwritten mlir file
makefile: update related commands.
midend/lib/Conversion/MatMulOptimization/: update BatchMatMulOptimize.cpp implementation of vectorisation, now using the standard method in buddy.
tests/Conversion/: batchmatmul-vectorisation.mlir: a test file.
Before vectorisation optimization: 0.0043869, after vectorization optimization: 0.00018692 (with vectorization size of 128), speedup ratio about 22.5. After vectorization size adjustment to 256: 0.000138998, speedup ratio about 30.6.
Vectorisation size of 32 compared to the previous version: 0.000586033 before the update, 0.000358105 after the update, with a speedup ratio of about 0.636. In addition, the original implementation may even be negatively optimised.