NVIDIA/cutlass
1
0
Fork 0
mirror of https://github.com/NVIDIA/cutlass.git synced 2026-05-11 17:00:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

Replace std::min with cute::min in sm120 blockwise scaling device functions (#3055)

Browse Source
This commit is contained in:
Vrushtee
2026-04-24 20:43:38 +05:30
committed by GitHub
parent b46b16d003
commit 9135a9bb6d
2 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
include/cutlass/gemm/collective/sm120_mma_array_tma_blockwise_scaling.hpp
View File

@@ -540,8 +540,8 @@ struct CollectiveMma<
Tensor tApA_SFA = make_tensor<bool>(shape(tAsA_SFA(_,_,0)));
Tensor tBpB_SFB = make_tensor<bool>(shape(tBsB_SFB(_,_,0)));
auto scale_m_lim = std::min(scales_m, (m_coord + 1) * ScaleMsPerTile);
auto scale_n_lim = std::min(scales_n, (n_coord + 1) * ScaleNsPerTile);
auto scale_m_lim = cute::min(scales_m, (m_coord + 1) * ScaleMsPerTile);
auto scale_n_lim = cute::min(scales_n, (n_coord + 1) * ScaleNsPerTile);
CUTLASS_PRAGMA_UNROLL
for (int i = 0; i < size(tApA_SFA); ++i)

4
include/cutlass/gemm/collective/sm120_mma_tma_blockwise_scaling.hpp
View File

@@ -472,8 +472,8 @@ struct CollectiveMma<
Tensor tApA_SFA = make_tensor<bool>(shape(tAsA_SFA(_,_,0)));
Tensor tBpB_SFB = make_tensor<bool>(shape(tBsB_SFB(_,_,0)));
auto scale_m_lim = std::min(scales_m, (m_coord + 1) * ScaleMsPerTile);
auto scale_n_lim = std::min(scales_n, (n_coord + 1) * ScaleNsPerTile);
auto scale_m_lim = cute::min(scales_m, (m_coord + 1) * ScaleMsPerTile);
auto scale_n_lim = cute::min(scales_n, (n_coord + 1) * ScaleNsPerTile);
CUTLASS_PRAGMA_UNROLL
for (int i = 0; i < size(tApA_SFA); ++i)