diff --git a/ggml/src/ggml-cuda/fattn-mma-f16.cuh b/ggml/src/ggml-cuda/fattn-mma-f16.cuh
index 7f4cfd5511ff..e172393c4c35 100644
--- a/ggml/src/ggml-cuda/fattn-mma-f16.cuh
+++ b/ggml/src/ggml-cuda/fattn-mma-f16.cuh
@@ -1535,8 +1535,6 @@ static __device__ __forceinline__ void flash_attn_ext_f16_process_tile(
             }
         }
 
-        __syncthreads();
-
         // Write back combined meta data:
 #pragma unroll
         for (int imeta = 0; imeta < nmeta; ++imeta) {
@@ -1556,10 +1554,9 @@ static __device__ __forceinline__ void flash_attn_ext_f16_process_tile(
             float2 * dstk_fixup_meta = dstk_fixup + (gridDim.x + blockIdx.x)*ncols;
             dstk_fixup_meta[(threadIdx.y/np)*cols_per_warp + threadIdx.x] = make_float2(KQ_cmn, KQ_crs);
         }
-    } else if (np > 1) {
-        // Warps with threadIdx.y % np == 0 execute a __syncthreads() in the if branch.
-        // Therefore, all other warps also need to execute a __syncthreads().
-        // Otherwise the points at which warps synchronize with each other would become misaligned.
+    }
+
+    if constexpr (np > 1) {
         __syncthreads();
     }