ck moe gemm implement (#1936)

Browse Source

* port all moe changes from ck_moe_gemm branch

* refine codes in the pr

* fix tail odd

* fix clang format

* fix clang format2

* make hot loop scheduler compatible with 16x16 and 32x32

* clang format

* fix per token quant

* rename moe example

* clang format

---------

Co-authored-by: coderfeli <coderfeli@163.com>

[ROCm/composable_kernel commit: 3786e16375]

...

This commit is contained in:

feli

2025-03-05 15:56:55 +08:00

committed by GitHub

parent dfd15c220d

commit 6fd94cff45

13 changed files with 6144 additions and 7 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

2144

include/ck/tensor_operation/gpu/grid/gridwise_moe_gemm.hpp Normal file

File diff suppressed because it is too large Load Diff