leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 24 Packages Projects Releases Wiki Activity

update 3.8 v2 (#2112)

Browse Source 
* update 3.8 v2

* update 3.8

---------

Co-authored-by: yuzhai <yuzhai@nvidia.com>
This commit is contained in:
Yujia Zhai
2025-02-19 22:03:14 -05:00
committed by GitHub
co-authored by yuzhai
parent e9627ce55b
commit b84e9802d8
166 changed files with 3986 additions and 4037 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/cutlass/gemm/warp/mma_tensor_op_tile_iterator_sm70.h
-2
View File
@@ -142,7 +142,6 @@ class MmaVoltaTensorOpMultiplicandTileIterator<
"Shape of warp-level Mma must be divisible by operator shape.");
// Shape of one individual LDS.128
// TODO: 32 and 4 are hardcoded, 32-by-4 is logical shape
using LdsShape = layout::PitchLinearShape<
32,
4
@@ -458,7 +457,6 @@ class MmaVoltaTensorOpMultiplicandTileIterator<
"Shape of warp-level Mma must be divisible by operator shape.");
// Shape of one individual LDS
// TODO: remove hardcoded 32 and 4
using LdsShape = layout::PitchLinearShape<
32,
4
include/cutlass/gemm/warp/mma_tensor_op_tile_iterator_sm80.h
-1
View File
@@ -995,7 +995,6 @@ public:
CUTLASS_DEVICE
MmaTensorOpMultiplicandTileIterator &add_tile_offset_negative(TensorCoord const &tile_offset) {
// TODO: fix this if it becomes an issue during warp it reset
add_tile_offset(tile_offset);
return *this;