Template X buffer ops on setFillerCoords

[alexxy/gromacs.git] / src / gromacs / nbnxm / cuda / nbnxm_cuda.cu

diff --git a/src/gromacs/nbnxm/cuda/nbnxm_cuda.cu b/src/gromacs/nbnxm/cuda/nbnxm_cuda.cu
index 57504d06bb15e130d71eaf81a953d5b87af2996c..f5135d7bfbf6a1a1de7639a54c05ef2eb6609f88 100644 (file)
--- a/src/gromacs/nbnxm/cuda/nbnxm_cuda.cu
+++ b/src/gromacs/nbnxm/cuda/nbnxm_cuda.cu
@@ -859,14 +859,15 @@ void nbnxn_gpu_x_to_nbat_x(const Nbnxm::Grid&        grid,
         config.sharedMemorySize = 0;
         config.stream           = stream;
 
-        auto       kernelFn      = nbnxn_gpu_x_to_nbat_x_kernel;
+        auto kernelFn = setFillerCoords ? nbnxn_gpu_x_to_nbat_x_kernel<true>
+                                        : nbnxn_gpu_x_to_nbat_x_kernel<false>;
         float4*    d_xq          = adat->xq;
         const int* d_atomIndices = nb->atomIndices;
         const int* d_cxy_na      = &nb->cxy_na[numColumnsMax * gridId];
         const int* d_cxy_ind     = &nb->cxy_ind[numColumnsMax * gridId];
-        const auto kernelArgs    = prepareGpuKernelArguments(
-                kernelFn, config, &numColumns, &d_xq, &setFillerCoords, &d_x, &d_atomIndices,
-                &d_cxy_na, &d_cxy_ind, &cellOffset, &numAtomsPerCell);
+        const auto kernelArgs =
+                prepareGpuKernelArguments(kernelFn, config, &numColumns, &d_xq, &d_x, &d_atomIndices,
+                                          &d_cxy_na, &d_cxy_ind, &cellOffset, &numAtomsPerCell);
         launchGpuKernel(kernelFn, config, nullptr, "XbufferOps", kernelArgs);
     }