Merge pull request #579 from canonizer/rank-warp-mask

Mask of all ones in `SHFL_IDX_SYNC` in block radix rank
NVIDIA · Oct 12, 2022 · 1fc34df · 1fc34df
2 parents dcd5b06 + f710ef8
commit 1fc34df
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/cub/block/block_radix_rank.cuh b/cub/block/block_radix_rank.cuh
@@ -1075,7 +1075,7 @@ struct BlockRadixRankMatchEarlyCounts
                     // atomic is a bit faster
                     warp_offset = atomicAdd(&warp_offsets[bin], popc);
                 }
-                warp_offset = SHFL_IDX_SYNC(warp_offset, leader, bin_mask);
+                warp_offset = SHFL_IDX_SYNC(warp_offset, leader, WARP_MASK);
                 if (lane == leader) *p_match_mask = 0;
                 WARP_SYNC(WARP_MASK);
                 ranks[u] = warp_offset + popc - 1;
@@ -1105,7 +1105,7 @@ struct BlockRadixRankMatchEarlyCounts
                     // atomic is a bit faster
                     warp_offset = atomicAdd(&warp_offsets[bin], popc);
                 }
-                warp_offset = SHFL_IDX_SYNC(warp_offset, leader, bin_mask);
+                warp_offset = SHFL_IDX_SYNC(warp_offset, leader, WARP_MASK);
                 ranks[u] = warp_offset + popc - 1;
             }
         }