Move scaling logic to input generation #338
Open
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
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
force-pushed
the
export-D80571223
branch
from
3484c61 to
a5bc2fa
Compare
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Pull Request resolved: meta-pytorch#338 Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
force-pushed
the
export-D80571223
branch
from
a5bc2fa to
edfdb2a
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Pull Request resolved: meta-pytorch#338 Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
force-pushed
the
export-D80571223
branch
from
edfdb2a to
a609415
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
force-pushed
the
export-D80571223
branch
from
a609415 to
b024198
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 20, 2025
Summary: Pull Request resolved: meta-pytorch#338 Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Reviewed By: NikhilAPatel Differential Revision: D80571223
jananisriram
force-pushed
the
export-D80571223
branch
from
b024198 to
4efc17d
Compare
jananisriram
force-pushed
the
export-D80571223
branch
from
c817c15 to
e3adb1b
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
xuzhao9
requested review from
NikhilAPatel
and removed request for
NikhilAPatel
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 21, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Test Plan: Imported from GitHub, without a `Test Plan:` line. Rollback Plan: Differential Revision: D80571223 Pulled By: jananisriram
jananisriram
force-pushed
the
export-D80571223
branch
from
e3adb1b to
d9e0627
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Error
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 21, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Test Plan: Imported from GitHub, without a `Test Plan:` line. Rollback Plan: Differential Revision: D80571223 Pulled By: jananisriram
jananisriram
force-pushed
the
export-D80571223
branch
from
d9e0627 to
1acc603
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 21, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Test Plan: Imported from GitHub, without a `Test Plan:` line. Rollback Plan: Differential Revision: D80571223 Pulled By: jananisriram
jananisriram
force-pushed
the
export-D80571223
branch
from
1acc603 to
ad791c4
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
jananisriram
added a commit
to jananisriram/tritonbench
that referenced
this pull request
Aug 21, 2025
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Test Plan: Imported from GitHub, without a `Test Plan:` line. Rollback Plan: Differential Revision: D80571223 Pulled By: jananisriram
jananisriram
force-pushed
the
export-D80571223
branch
from
ad791c4 to
da77dd8
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
Summary: Move scaling logic for FP8 benchmarks to `get_input_iter()`. This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (`bfloat16`, `float16`), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (`float8_e4m3fn`). This diff also circumvents performing unsupported operations, like `torch.max` and `torch.abs`, on low-precision data types. Pull Request resolved: meta-pytorch#338 Test Plan: Imported from GitHub, without a `Test Plan:` line. Rollback Plan: Reviewed By: xuzhao9 Differential Revision: D80571223 Pulled By: jananisriram
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80571223 |
jananisriram
force-pushed
the
export-D80571223
branch
from
da77dd8 to
7d67063
Compare
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
jananisriram
had a problem deploying
to
docker-s3-upload
GitHub Actions
Failure
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Summary:
Move scaling logic for FP8 benchmarks to
get_input_iter().This diff aligns our fp8_gemm benchmarking suite with real-world practices: input tensors are of high precision types (
bfloat16,float16), scales are computed on the high-precision input tensors, and input tensors are then casted to a lower precision (float8_e4m3fn).This diff also circumvents performing unsupported operations, like
torch.maxandtorch.abs, on low-precision data types.Reviewed By: NikhilAPatel
Differential Revision: D80571223