-
Notifications
You must be signed in to change notification settings - Fork 21.5k
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add NEON-accelerated int8mm for bfloat16 (#125290)
As apparently `vshlq_u32` is faster than `vcvt_f32_f16` Refactor NEON `tinygemm_kernel` to rely on `load_as_float32x4` and `load_as_float32x4x2` and implement them for float16 (using vcvt), bfloat16 (using left shift) and plain float32 (not using anything) As result stories110M run at 60 tokens/sec with f16, but at 66 tokens/sec with bf16 and 75 tokens/sec with f32, though more bandwith demand starts to favor reduced floating types as model size gets bigger. Pull Request resolved: #125290 Approved by: https://github.com/mikekgfb
- Loading branch information
1 parent
9fbb4df
commit a271514
Showing
1 changed file
with
82 additions
and
12 deletions.
There are no files selected for viewing
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