ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-03 13:11:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: add support for bf16 for grouped_gemm & grouped_gemm_preshuffle… (#3225)

Browse Source 
* feat: add support for bf16 for grouped_gemm & grouped_gemm_preshuffle kernel(s) along with unit test

* docs: Update CHANGELOG.MD
This commit is contained in:
Aviral Goel
2025-11-18 09:32:27 -05:00
committed by GitHub
parent 3ede8e2a6e
commit ac70206b2c
7 changed files with 48 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
test/ck_tile/grouped_gemm_preshuffle/test_grouped_gemm_preshuffle.cpp
View File

@@ -10,6 +10,7 @@
using F16 = ck_tile::half_t;
using F8 = ck_tile::fp8_t;
using BF16 = ck_tile::bf16_t;
using F32 = float;
using Row = ck_tile::tensor_layout::gemm::RowMajor;
using Col = ck_tile::tensor_layout::gemm::ColumnMajor;
@@ -57,7 +58,17 @@ using KernelTypes = ::testing::Types<
KernelConfig< Row, Col, Row, F16, F16, F32, F16, True, 16, 64, 256, 1>,
KernelConfig< Row, Col, Row, F8, F8, F32, F16, True, 16, 64, 256, 1>,
KernelConfig< Row, Col, Row, F16, F16, F32, F16, True, 128, 128, 128, 2>,
KernelConfig< Row, Col, Row, F8, F8, F32, F16, True, 128, 128, 128, 2>
KernelConfig< Row, Col, Row, F8, F8, F32, F16, True, 128, 128, 128, 2>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, False, 16, 64, 256, 1>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, False, 16, 64, 256, 1>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, False, 128, 128, 128, 2>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, False, 128, 128, 128, 2>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, True, 16, 64, 256, 1>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, True, 16, 64, 256, 1>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, True, 128, 128, 128, 2>,
KernelConfig< Row, Col, Row, BF16, BF16, F32, BF16, True, 128, 128, 128, 2>
>;
// clang-format on