Add swizzling params

This commit is contained in:
Chenggang Zhao
2025-04-11 13:15:52 +08:00
parent 2e7e58011b
commit 6078b25424
3 changed files with 44 additions and 19 deletions

View File

@@ -395,6 +395,7 @@ fp8_gemm_kernel(__nv_bfloat16* gmem_d, float* scales_b, int* grouped_layout,
template <uint32_t SHAPE_N, uint32_t SHAPE_K,
uint32_t BLOCK_M, uint32_t BLOCK_N, uint32_t BLOCK_K,
uint32_t BLOCK_N_PADDING,
uint32_t kSwizzleDMode,
uint32_t kNumGroups, uint32_t kNumStages,
uint32_t kNumTMAMulticast, bool kIsTMAMulticastOnA,
GemmType kGemmType>