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

Updated documents for swizzle support #1704

Open
wants to merge 2 commits into
base: develop
Choose a base branch
from

Conversation

Serge45
Copy link
Collaborator

@Serge45 Serge45 commented Feb 26, 2025

Added more descriptions on special data orderings and their supported combinations of datatype and orderings.

@Serge45 Serge45 added the ci:docs-only Only run Read the Docs CI on this PR label Feb 26, 2025
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.

should we remove TT from supported list since we haven't put solutions into hibplaslt yet

@@ -148,15 +148,23 @@ For FP8 type Matmul, hipBLASLt supports the type combinations shown in the follo
| | | BF8 | BF8 | | | | FP32, FP16 | FP16 |
+-------+-------+-------+-------+-------------+----------+----------+------------+-----------+

To use FP16-specific data ordering `HIPBLASLT_ORDER_COL16_4R8` in `hipblasLtMatmul` for the gfx94x architecture, choose one of these valid combinations of transposes and orders of input and output matrices:
To use special data orderings `HIPBLASLT_ORDER_COL16_4R8` and `HIPBLASLT_ORDER_COL16_4R16` in `hipblasLtMatmul` for the gfx94x architecture, choose one of these valid combinations of transposes and orders of input and output matrices:

+-------+-------+-------+-------+-----------------------------+-----------------------------+---------------------+---------------------+
| Atype | Btype | opA | opB | orderA | orderB | orderC | orderD |
Copy link
Collaborator

Choose a reason for hiding this comment

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

Here should be A/Btype | C/D type

@Serge45 Serge45 force-pushed the feature/swizzle-doc-cont branch 2 times, most recently from 171bd2e to 48cdc6f Compare February 26, 2025 05:03
+-------+-------+-------+-------+------+-------+-----------------------------+-----------------------------+---------------------+---------------------+
| Atype | Btype | CType | DTyoe | opA | opB | orderA | orderB | orderC | orderD |
+=======+=======+=======+=======+======+=======+=============================+=============================+=====================+=====================+
| FP8 | FP16 | FP16 | FP16 | T | N | HIPBLASLT_ORDER_COL16_4R16 | HIPBLASLT_ORDER_COL | HIPBLASLT_ORDER_COL | HIPBLASLT_ORDER_COL |
Copy link
Collaborator

Choose a reason for hiding this comment

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

FP8 | FP8 | FP16 | FP16

@Serge45 Serge45 force-pushed the feature/swizzle-doc-cont branch 2 times, most recently from abc5b7e to d6dd5f8 Compare February 26, 2025 08:34
amd-jnovotny
amd-jnovotny previously approved these changes Feb 26, 2025
Copy link
Contributor

@amd-jnovotny amd-jnovotny left a comment

Choose a reason for hiding this comment

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

Docs OK. One rewording suggestion.

Co-authored-by: Jeffrey Novotny <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ci:docs-only Only run Read the Docs CI on this PR
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants