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

[Experimental] Support on arbitrary M & K for siwzzle-A #1521

Open
wants to merge 7 commits into
base: stable3
Choose a base branch
from

Conversation

Serge45
Copy link
Collaborator

@Serge45 Serge45 commented Jan 7, 2025

Implemented

  • removed the restriction of M and K, now the kernels assume tensor A is properly padded
  • CPU implementation of padding op
  • auto-padding for tensilelite client and hipblaslt-bench
  • samples for weight padding.

@Serge45 Serge45 force-pushed the stable3/swizzle-padding branch from 63114c7 to d21a7e6 Compare January 20, 2025 08:18
Copy link
Collaborator

@jichangjichang jichangjichang left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you add some odd size test items into gtest?

@Serge45 Serge45 requested a review from a team as a code owner January 22, 2025 07:42
* Note that for this order, the number of columns(rows) of the tensor has to be multiple of 16(32) or
* pre-padded to 16(32).
*/
HIPBLASLT_ORDER_COL16_4R8 = 2
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Shift the num to a larger value for hipblaslt specific enum type
ex: HIPBLASLT_MATMUL_DESC_COMPUTE_INPUT_TYPE_A_EXT = 100,

@Serge45 Serge45 force-pushed the stable3/swizzle-padding branch from 28b5c44 to d77b0b4 Compare January 22, 2025 08:08
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