Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

kludge workaround for AOTI fail on x86 Linux #243

Merged
merged 1 commit into from
Apr 17, 2024
Merged

kludge workaround for AOTI fail on x86 Linux #243

merged 1 commit into from
Apr 17, 2024

Conversation

mikekgfb
Copy link
Contributor

kludge workaround for AOTI fail on x86 Linux

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Meta Open Source bot. label Apr 17, 2024
@mikekgfb mikekgfb requested review from malfet and mergennachin April 17, 2024 20:13
@swolchok
Copy link
Contributor

swolchok commented Apr 17, 2024

FWIW this is int4pack_mm_kernel segfaulting for bfloat16; note that export to float16 also doesn't work right now

@mikekgfb
Copy link
Contributor Author

FWIW this is int4pack_mm_kernel segfaulting for bfloat16; note that export to float16 also doesn't work right now

on x86 or ARM. Maybe we should ping Intel?

Also, as a footnote, there are separate kernels for AVX2 and AVX512 with different data layout that are being tested.

Regarding float16 support, I thought @malfet had added that? But it's been a blur how fast development works on this repo!

@swolchok
Copy link
Contributor

It looks like _weight_int4pack_mm_cpu doesn't TORCH_CHECK that all the input sizes line up? @malfet

@mikekgfb mikekgfb merged commit 170dde6 into main Apr 17, 2024
15 of 18 checks passed
@mikekgfb mikekgfb deleted the aoti_kludge branch April 17, 2024 20:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
CLA Signed This label is managed by the Meta Open Source bot.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants