diff --git a/gpu/D3Q19.cu b/gpu/D3Q19.cu
index 7106e5b4..bb38059e 100644
--- a/gpu/D3Q19.cu
+++ b/gpu/D3Q19.cu
@@ -89,7 +89,7 @@ __global__ void sum_kernel_block(double *sum, double *input, int n)
 __inline__ __device__
 double warpReduceSum(double val) {
 	for (int offset = warpSize/2; offset > 0; offset /= 2)
-		val += __shfl_down(val, offset);
+		val += __shfl_down_sync(0xFFFFFFFF, val, offset, 32);
 	return val;
 }