add comment on reduce cuda

This commit is contained in:
lucasdelimanogueira 2024-06-04 11:21:56 -03:00
parent 12eb38a940
commit b56ed81cd7
2 changed files with 1 additions and 1 deletions

0
.gitmodules vendored
View file

View file

@ -136,7 +136,7 @@ __global__ void sum_tensor_cuda_kernel(float* data, float* result_data, int size
__syncthreads();
// Perform block-wise reduction
for (int s = blockDim.x / 2; s > 0; s >>= 1) {
for (int s = blockDim.x / 2; s > 0; s >>= 1) { // s>>=1 --> s = s/2
if (tid < s) {
partial_sum[tid] += partial_sum[tid + s];
}