FP8 Grouped Gemm Optimization #3655
Open
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
Summary: When benchmarking quantize functions, we'd like the overhead to mimic e2e behavior as closely as possible. For example, weights should be quantized ahead of time. The current design of quantize_bench does not allow this. To accomodate it, I've added a new optional preprocess phase that allows some transformations to be applied independently from benchmarking. Here we use it to prepare data for grouped gemm benchmarks to more accurately capture the e2e behavior. Differential Revision: D68964950
Summary: Adds support for the --trace option which will produce gpu traces for each benchmarked operator. This only works internally so if tried in OSS we fall back to nullcontext. Differential Revision: D68980020
Summary: X-link: facebookresearch/FBGEMM#731 While optimizing MOE, we found that small overheads were a major bottleneck for grouped gemm performance. This diff tackles a few of them, specifically overhead from torch.dynamo wrapping `quantize_fp8_row` and having to slice input tensors before calling `f8f8bf16_rowwise_grouped`. To fix the former, we enable `triton_quantize_fp8_row` to be directly called, skipping dynamo compatibility. In cases where AOTI isnt needed, this removes a bit of overhead. To fix the latter, we templatize f8f8fbf16_rowwise_grouped_dynamic to accept at::Tensor instead of lists. We introduce a new wrapper called f8f8bf16_rowwise_grouped_stacked to maintain the behavior where zero_start_index_M isnt provided but a user wants a single contiguous output tensor. In microbenchmarks, we've found these seemingly small changes can improve TFLOPs by 2X for small workloads. Reviewed By: jiawenliu64 Differential Revision: D69072529
|
This pull request was exported from Phabricator. Differential Revision: D69072529 |
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
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.
Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/731
While optimizing MOE, we found that small overheads were a major bottleneck for grouped gemm performance. This diff tackles a few of them, specifically overhead from torch.dynamo wrapping
quantize_fp8_rowand having to slice input tensors before callingf8f8bf16_rowwise_grouped.To fix the former, we enable
triton_quantize_fp8_rowto be directly called, skipping dynamo compatibility. In cases where AOTI isnt needed, this removes a bit of overhead.To fix the latter, we templatize f8f8fbf16_rowwise_grouped_dynamic to accept at::Tensor instead of lists. We introduce a new wrapper called f8f8bf16_rowwise_grouped_stacked to maintain the behavior where zero_start_index_M isnt provided but a user wants a single contiguous output tensor.
In microbenchmarks, we've found these seemingly small changes can improve TFLOPs by 2X for small workloads.
Reviewed By: jiawenliu64
Differential Revision: D69072529