diff --git a/cpp/include/raft/sparse/convert/detail/bitmap_to_csr.cuh b/cpp/include/raft/sparse/convert/detail/bitmap_to_csr.cuh
index 0f1b542246..866923d647 100644
--- a/cpp/include/raft/sparse/convert/detail/bitmap_to_csr.cuh
+++ b/cpp/include/raft/sparse/convert/detail/bitmap_to_csr.cuh
@@ -230,8 +230,8 @@ RAFT_KERNEL __launch_bounds__(bitmap_to_csr_tpb)
     int l_bits    = raft::detail::popc(l_bitmap);
     int l_sum_32b = 0;
     BlockScan(scan_storage).InclusiveSum(l_bits, l_sum_32b);
-    __syncthreads();
     l_sum = l_sum_32b + g_sum - l_bits;
+    __syncthreads();
 
 #pragma unroll
     for (int i = 0; i < BITS_PER_BITMAP; i++) {