apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 00:17:25 +08:00
Code Issues Actions 21 Packages Projects Releases Wiki Activity

[Metax] modify wrapSize to WARP_SIZE (#5442)

Browse Source
This commit is contained in:
xiaozude
2025-12-09 17:44:02 +08:00
committed by GitHub
parent e397c4fba6
commit df67379bc3
4 changed files with 406 additions and 228 deletions
Download Patch File Download Diff File Expand all files Collapse all files
custom_ops/gpu_ops/speculate_decoding/top_p_candidates.cu
+1 -1
View File
@@ -18,7 +18,7 @@
template <typename T>
__forceinline__ __device__ T
CudaShuffleDownSync(unsigned mask, T val, int delta, int width = warpSize) {
CudaShuffleDownSync(unsigned mask, T val, int delta, int width = WARP_SIZE) {
return __shfl_down_sync(mask, val, static_cast<unsigned>(delta), width);
}