Improving mamba runtime by using updates #1552
Merged
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
- Move float16 to bfloat16, which has less imprecisions (load test are failing with the update kernels + f16, all working under bf16). Another note, is that we are not respecting the layer norm in f32 defined in the configuration (this is OK in my book, but that could impact the f16 precision) - Moved to update kernels. Triton overhead is super high, removed by switching to cuda graphs works great (update cuda graph is available in TRT-LLM if needed, seems *exactly* like the regular ssm kernel. - Moved inference_params struct in order to make only 2 tensors, to reduce the overhead of copying back and forth to the cuda graphs. - Left over overhead seems entirely in the tokenization bit. (Still 4 copies are paid before launching the graph)
|
wooooo this is awesome thank you for the optimizations @Narsil! Just pulled down and am getting a huge performance increase. ~50 to 15ms 🙏 LGTM just think the snapshots for mamba are out of date |
|
They are updated, just seems noise is quite high in the update kernels (could explain the f16 instabilities) |
kdamaszk
pushed a commit
to kdamaszk/tgi-gaudi
that referenced
this pull request
Apr 29, 2024
- Move float16 to bfloat16, which has less imprecisions (load test are failing with the update kernels + f16, all working under bf16). Another note, is that we are not respecting the layer norm in f32 defined in the configuration (this is OK in my book, but that could impact the f16 precision) - Moved to update kernels. Triton overhead is super high, removed by switching to cuda graphs works great (update cuda graph is available in TRT-LLM if needed, seems *exactly* like the regular ssm kernel. - Moved inference_params struct in order to make only 2 tensors, to reduce the overhead of copying back and forth to the cuda graphs. - Left over overhead seems entirely in the tokenization bit. (Still 4 copies are paid before launching the graph) # What does this PR do? <!-- Congratulations! You've made it this far! You're not quite done yet though. Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution. Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change. Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost. --> <!-- Remove if not applicable --> Fixes # (issue) ## Before submitting - [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). - [ ] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests), Pull Request section? - [ ] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link to it if that's the case. - [ ] Did you make sure to update the documentation with your changes? Here are the [documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and [here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation). - [ ] Did you write any new necessary tests? ## Who can review? Anyone in the community is free to review the PR once the tests have passed. Feel free to tag members/contributors who may be interested in your PR. <!-- Your PR will be replied to more quickly if you can figure out the right person to tag with @ @OlivierDehaene OR @Narsil -->
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.
Move float16 to bfloat16, which has less imprecisions (load test are
failing with the update kernels + f16, all working under bf16).
Another note, is that we are not respecting the layer norm in f32
defined in the configuration (this is OK in my book, but that could
impact the f16 precision)
Moved to update kernels. Triton overhead is super high, removed by
switching to cuda graphs works great (update cuda graph is available
in TRT-LLM if needed, seems exactly like the regular ssm kernel.
Moved inference_params struct in order to make only 2 tensors, to
reduce the overhead of copying back and forth to the cuda graphs.
Left over overhead seems entirely in the tokenization bit. (Still 4
copies are paid before launching the graph)
What does this PR do?
Fixes # (issue)
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.