Skip to content

Fix bf16->f32 conversion for NEOVERSEV1 and NEOVERSEN2 targets#5483

Merged
martin-frbg merged 2 commits intoOpenMathLib:developfrom
Mousius:bgemm-correctness
Oct 7, 2025
Merged

Fix bf16->f32 conversion for NEOVERSEV1 and NEOVERSEN2 targets#5483
martin-frbg merged 2 commits intoOpenMathLib:developfrom
Mousius:bgemm-correctness

Conversation

@Mousius
Copy link
Contributor

@Mousius Mousius commented Oct 6, 2025

This fixes an issue originally introduced with the BGEMM kernel when I was tweaking it. #5287 didn't suffer from this bug.

I've updated the tests to run with beta=1.0 so as to test loading and updating from C.

Alongside this, the tests now return sensible return values to reduce the risk of them being ignored.

Also fixed a bug in generic/gemv_t.c resulting in weird outputs for bgemv with updates.

@Mousius Mousius force-pushed the bgemm-correctness branch 2 times, most recently from 5da3bd0 to 5b54afe Compare October 6, 2025 16:48
This fixes an issue originally introduced with the BGEMM kernel.

I've updated the tests to run with `beta=1.0` so as to test loading and
updating from C.

Alongside this, the tests now return sensible return values to reduce
the risk of them being ignored.

Also fixed a bug in `generic/gemv_t.c` resulting in weird outputs for
`bgemv`.
@Mousius Mousius force-pushed the bgemm-correctness branch from 5b54afe to 578e7da Compare October 6, 2025 18:07
@Mousius Mousius changed the title Fix bf16->f32 conversion for NEOVERSEV1 target Fix bf16->f32 conversion for NEOVERSEV1 and NEOVERSEN2 targets Oct 6, 2025
@martin-frbg martin-frbg added this to the 0.3.31 milestone Oct 7, 2025
@martin-frbg martin-frbg merged commit 8918247 into OpenMathLib:develop Oct 7, 2025
79 of 88 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants