Allow specifiying the use of persistent kernel #5129
Conversation
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
Contributor
|
@peiying779 has exported this pull request. If you are a Meta employee, you can view the originating Diff in D86579911. |
peiying779
pushed a commit
to peiying779/FBGEMM-1
that referenced
this pull request
Nov 14, 2025
Summary: X-link: facebookresearch/FBGEMM#2130 Added environment argument "use_persistent" (default is False) to explicitly turn off non-persistent kernel and use persistent kernel. Throws error when both "use_persistent" and "no_use_persistent" are specified in the arguments. Example usage: Persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent Non-persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --no_use_persistent When both specified in the arguments: buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent --no_use_persistent IT WILL THROW ERROR: Cannot specify both '--use_persistent' and '--no_use_persistent' at the same time. These options are mutually exclusive. Please use only one. Reviewed By: jwfromm Differential Revision: D86579911
peiying779
force-pushed
the
export-D86579911
branch
from
e73a4fb to
e2e0abb
Compare
facebook-github-bot
pushed a commit
to meta-pytorch/tritonbench
that referenced
this pull request
Nov 14, 2025
Summary: X-link: pytorch/FBGEMM#5129 X-link: facebookresearch/FBGEMM#2130 Added environment argument "use_persistent" (default is False) to explicitly turn off non-persistent kernel and use persistent kernel. Throws error when both "use_persistent" and "no_use_persistent" are specified in the arguments. Example usage: Persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent Non-persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --no_use_persistent When both specified in the arguments: buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent --no_use_persistent IT WILL THROW ERROR: Cannot specify both '--use_persistent' and '--no_use_persistent' at the same time. These options are mutually exclusive. Please use only one. Reviewed By: jwfromm Differential Revision: D86579911
facebook-github-bot
pushed a commit
to meta-pytorch/tritonbench
that referenced
this pull request
Nov 19, 2025
Summary: X-link: pytorch/FBGEMM#5129 X-link: facebookresearch/FBGEMM#2130 Added environment argument "use_persistent" (default is False) to explicitly turn off non-persistent kernel and use persistent kernel. Throws error when both "use_persistent" and "no_use_persistent" are specified in the arguments. Example usage: Persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent Non-persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --no_use_persistent When both specified in the arguments: buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent --no_use_persistent IT WILL THROW ERROR: Cannot specify both '--use_persistent' and '--no_use_persistent' at the same time. These options are mutually exclusive. Please use only one. Reviewed By: jwfromm Differential Revision: D86579911
peiying779
force-pushed
the
export-D86579911
branch
from
e2e0abb to
40bb52b
Compare
peiying779
pushed a commit
to peiying779/FBGEMM-1
that referenced
this pull request
Nov 19, 2025
Summary: X-link: facebookresearch/FBGEMM#2130 Added environment argument "use_persistent" (default is False) to explicitly turn off non-persistent kernel and use persistent kernel. Throws error when both "use_persistent" and "no_use_persistent" are specified in the arguments. Example usage: Persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent Non-persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --no_use_persistent When both specified in the arguments: buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent --no_use_persistent IT WILL THROW ERROR: Cannot specify both '--use_persistent' and '--no_use_persistent' at the same time. These options are mutually exclusive. Please use only one. Reviewed By: jwfromm Differential Revision: D86579911
Summary: X-link: facebookresearch/FBGEMM#2130 Added environment argument "use_persistent" (default is False) to explicitly turn off non-persistent kernel and use persistent kernel. Throws error when both "use_persistent" and "no_use_persistent" are specified in the arguments. Example usage: Persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent Non-persistent kernel-- buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --no_use_persistent When both specified in the arguments: buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent --no_use_persistent IT WILL THROW ERROR: Cannot specify both '--use_persistent' and '--no_use_persistent' at the same time. These options are mutually exclusive. Please use only one. Reviewed By: jwfromm Differential Revision: D86579911
peiying779
force-pushed
the
export-D86579911
branch
from
40bb52b to
d226d04
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Summary:
Added environment argument "use_persistent" (default is False) to explicitly turn off non-persistent kernel and use persistent kernel.
Throws error when both "use_persistent" and "no_use_persistent" are specified in the arguments.
Example usage:
Persistent kernel--
buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent
Non-persistent kernel--
buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --no_use_persistent
When both specified in the arguments:
buck2 run mode/{dev-nosan,amd-gpu} -c xlog.level=WARNING -m ovr_config//triton:trunk -m rocm7 -c fbcode.nvcc_arch=mi350 -c fbcode.enable_gpu_sections=true pytorch/tritonbench:run -- --op fp8_gemm_rowwise --no_use_tma --use_persistent --no_use_persistent
IT WILL THROW ERROR: Cannot specify both '--use_persistent' and '--no_use_persistent' at the same time. These options are mutually exclusive. Please use only one.
Reviewed By: jwfromm
Differential Revision: D86579911