mirror of
https://github.com/NVIDIA/cuda-samples.git
synced 2025-07-01 12:00:49 +08:00
Fixing correctness of bf16TensorCoreGemm
This commit is contained in:
parent
5f97d7d0df
commit
fb9f77575a
@ -592,8 +592,8 @@ __global__ void simple_wmma_bf16gemm(__nv_bfloat16 *a, __nv_bfloat16 *b, float *
|
||||
int aCol = i;
|
||||
int aRow = warpM * M;
|
||||
|
||||
int bCol = i;
|
||||
int bRow = warpN * N;
|
||||
int bCol = warpN * N;
|
||||
int bRow = i;
|
||||
|
||||
// Bounds checking
|
||||
if (aRow < m_ld && aCol < k_ld && bRow < k_ld && bCol < n_ld) {
|
||||
|
Loading…
x
Reference in New Issue
Block a user