[ExecuTorch] Support BFloat16 in CPUBlas gemm #5122
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
bfloat16.h was a stub. I've filled it out by porting the c10 implementation, added it to ET_SWITCH and ET_FORALL macros, and hooked it up to promoteTypes. I extended the half_to_float argument to promoteTypes to also coerce bfloat16 to float because I figured anybody who wants to ignore half probably also wants to ignore bf16. Differential Revision: [D61981361](https://our.internmc.facebook.com/intern/diff/D61981361/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981356](https://our.internmc.facebook.com/intern/diff/D61981356/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981355](https://our.internmc.facebook.com/intern/diff/D61981355/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981353](https://our.internmc.facebook.com/intern/diff/D61981353/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981357](https://our.internmc.facebook.com/intern/diff/D61981357/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981364](https://our.internmc.facebook.com/intern/diff/D61981364/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981360](https://our.internmc.facebook.com/intern/diff/D61981360/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981359](https://our.internmc.facebook.com/intern/diff/D61981359/) [ghstack-poisoned]
Adding bfloat16 support to important ops for LLMs to start. Differential Revision: [D61981362](https://our.internmc.facebook.com/intern/diff/D61981362/) [ghstack-poisoned]
The LLM runner assumed that the data type could only be float or half. Suport bfloat16 and neaten up the code while we're at it. Differential Revision: [D61981354](https://our.internmc.facebook.com/intern/diff/D61981354/) **NOTE FOR REVIEWERS**: This PR has internal Meta-specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D61981354/)! [ghstack-poisoned]
Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
…n export_llama" Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
…n export_llama" Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
…n export_llama" Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
…n export_llama" Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
…n export_llama" Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
…n export_llama" Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
Support creating bf16 PTEs. Differential Revision: [D61981363](https://our.internmc.facebook.com/intern/diff/D61981363/) [ghstack-poisoned]
Differential Revision: [D62151658](https://our.internmc.facebook.com/intern/diff/D62151658/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/5122
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 9c7e8fb with merge base b69ae0c ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D62151658 |
This was referenced Sep 6, 2024
swolchok
changed the base branch from
gh/swolchok/31/base
to
gh/swolchok/30/head
kimishpatel
approved these changes
Sep 6, 2024
kedarnath03
pushed a commit
to kedarnath03/executorch
that referenced
this pull request
Jun 25, 2025
Differential Revision: [D62151658](https://our.internmc.facebook.com/intern/diff/D62151658/) ghstack-source-id: 241282450 Pull Request resolved: pytorch/executorch#5122
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.
Stack from ghstack (oldest at bottom):
Differential Revision: D62151658