Skip to content

Commit

Permalink
Update dequant_impl_packed.cu
Browse files Browse the repository at this point in the history
  • Loading branch information
YangWang92 authored Oct 28, 2024
1 parent a7a5e2e commit dee1128
Showing 1 changed file with 0 additions and 4 deletions.
4 changes: 0 additions & 4 deletions csrc/dequant_impl_packed.cu
Original file line number Diff line number Diff line change
Expand Up @@ -43,13 +43,9 @@ __global__ void WqA16WithOutliers_PackIndice(
__shared__ float shared_output[GROUPSIZE][cuda::kBlockSize / WARP_SIZE + 1];
scalar_t tmp_output[GROUPSIZE];
const scalar_t zero_value = ZERO_VALUE(scalar_t());
__shared__ float shared_output[GROUPSIZE][cuda::kBlockSize / WARP_SIZE + 1];
scalar_t tmp_output[GROUPSIZE];
const scalar_t zero_value = ZERO_VALUE(scalar_t());
#pragma unroll
for (int i = 0; i < GROUPSIZE; i++) {
tmp_output[i] = zero_value;
tmp_output[i] = zero_value;
}
input_data = input_data + in_features * bidy;
out = out + out_features * bidy * gridDim.z;
Expand Down

0 comments on commit dee1128

Please sign in to comment.