Small optimization for BoxReciprocalSumsGPU

GOMC-WSU · Sep 7, 2024 · c8efcd0 · c8efcd0
1 parent c396711
commit c8efcd0
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/src/GPU/CalculateEwaldCUDAKernel.cu b/src/GPU/CalculateEwaldCUDAKernel.cu
@@ -140,7 +140,7 @@ __global__ void BoxReciprocalSumsGPU(double *gpu_x, double *gpu_y,
                                      double *gpu_sumRnew, double *gpu_sumInew) {
   int image = blockIdx.x;
   double sumR = 0.0, sumI = 0.0;
-#pragma unroll 16
+#pragma unroll 8
   for (int particleID = threadIdx.x; particleID < atomNumber; particleID += THREADS_PER_BLOCK_SM) {
     double dot = DotProductGPU(gpu_kx[image], gpu_ky[image],
                                gpu_kz[image], gpu_x[particleID],