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

v4.1 release update v2. (#2481)

Browse Source
This commit is contained in:
Junkai-Wu
2025-07-22 10:03:55 +08:00
committed by GitHub
parent 9baa06dd57
commit fd6cfe1ed0
179 changed files with 7878 additions and 1286 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/08_turing_tensorop_gemm/turing_tensorop_gemm.cu
View File

@@ -64,7 +64,7 @@ ElementComputeEpilogue (int32_t), ElementInputA (int8_t), ElementInputB (int8_t)
(int32_t). Communicating just the data type is not enough. As the data is laid out linearly in
memory, we have to convey the layout of matrices. We do that by initializing template variable
LayoutInputA to column major cutlass variable, LayoutInputB to row major and LayoutOutput to row
major. Next, we setup rules to comptue alpha * X + beta * C which is called epilogue of the kernel.
major. Next, we setup rules to compute alpha * X + beta * C which is called epilogue of the kernel.
We initialize template variable EpilogueOp, which takes the data type of output ElementOutput
(int32_t), the number of elements per vector memory access (16), data type of accumulator (int32_t)
and data type of computation of linear combination (alpha * X + beta * C).