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

rank_k_asm_d8x4() VMOVAPD alignment problem #2

Open
ChenhanYu opened this issue Nov 3, 2016 · 2 comments
Open

rank_k_asm_d8x4() VMOVAPD alignment problem #2

ChenhanYu opened this issue Nov 3, 2016 · 2 comments
Assignees
Labels

Comments

@ChenhanYu
Copy link
Owner

/kernel/x86_64/sandybridge/rank_k_d8x4.hpp

VMOVAPP requires c to be 32 bytes aligned. To use this kernel ldc must multiples of 4.
Still need to decide how to fix this problem.

It seems that rank_k_int_d8x4() does not have this problem.

@jianyuh
Copy link
Collaborator

jianyuh commented Nov 3, 2016

I have similar issues with /kernel/x86_64/sandybridge/stra_k_d8x4.hpp

Currently it requires that matrix C is allocated with 32 bytes aligned.

@ChenhanYu
Copy link
Owner Author

I think having C allocated as 32 bytes aligned is not enough. You will still need to have ldc as multiples of 4. I tried (m, n, k) = (9, 4, 1) and I still got a segmentation fault even that I allocated C using aligned allocation.

@ChenhanYu ChenhanYu self-assigned this Nov 3, 2016
@ChenhanYu ChenhanYu added the bug label Nov 3, 2016
ChenhanYu added a commit that referenced this issue Jan 26, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

2 participants