leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 27 Packages Projects Releases Wiki Activity

3.6.0 update (#2005)

Browse Source 
* 3.6.0 update

* doc and swap stuff

---------

Co-authored-by: yuzhai <yuzhai@nvidia.com>
Co-authored-by: Haicheng Wu <haichengw@nvidia.com>
This commit is contained in:
Yujia Zhai
2024-12-25 01:34:40 -05:00
committed by GitHub
co-authored by yuzhai Haicheng Wu
parent e1cd8c7866
commit 3d261a5974
258 changed files with 10863 additions and 3883 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/unit/cute/turing/cooperative_gemm.cu
+7 -9
View File
@@ -38,21 +38,19 @@
using namespace cute;
TEST(SM75_CuTe_Turing, CooperativeGemm1_MixedPrecisionFP16FP32_MMA) {
constexpr uint32_t thread_block_size = 128;
constexpr uint32_t max_vec_bits = 128;
using TA = cutlass::half_t;
using TB = cutlass::half_t;
using TC = float;
constexpr uint32_t m = 64;
constexpr uint32_t n = 64;
constexpr uint32_t k = 64;
constexpr uint32_t thread_block_size = 128;
using tiled_mma_t =
auto shape_mnk = make_shape(_64{}, _64{}, _64{});
auto tiled_mma =
TiledMMA<
MMA_Atom<SM75_16x8x8_F32F16F16F32_TN>,
Layout<Shape<_2, _2, _1>>
>;
>{};
test_cooperative_gemm_col_major_layout<m, n, k, thread_block_size, tiled_mma_t, 128, TA, TB, TC>();
test_cooperative_gemm_col_major_layout<thread_block_size, max_vec_bits, TA, TB, TC>(shape_mnk, tiled_mma);
}