Blockwise and Groupwise GEMM for Blackwell and Improvements for Hopper (#2139)

- Blockwise and Groupwise GEMM improvements for Hopper.
- Blockwise and Groupwise GEMM for Blackwell.
- Blockwise Grouped GEMM for Hopper.
- Static ScalePromotionInterval for Hopper FP8 GEMMs.

Co-authored-by: dePaul Miller <23461061+depaulmillz@users.noreply.github.com>
This commit is contained in:
dePaul Miller
2025-02-26 12:44:58 -05:00
committed by GitHub
co-authored by dePaul Miller
parent eefa171318
commit ca4fdbea70
28 changed files with 6860 additions and 71 deletions
@@ -65,6 +65,7 @@ struct IsCutlass3ArrayKernel<ProblemShape, cute::void_t<typename ProblemShape::U
#include "cutlass/gemm/kernel/sm90_gemm_array_tma_warpspecialized_cooperative.hpp"
#include "cutlass/gemm/kernel/sm100_gemm_tma_warpspecialized.hpp"
#include "cutlass/gemm/kernel/sm100_gemm_tma_warpspecialized_mma_transform.hpp"
#include "cutlass/gemm/kernel/sm100_gemm_array_tma_warpspecialized.hpp"
#include "cutlass/gemm/kernel/sm100_gemm_tma_warpspecialized_input_transform.hpp"
#include "cutlass/gemm/kernel/sm100_gemm_array_tma_warpspecialized_input_transform.hpp"