portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-09 23:45:36 +00:00
Code Releases Activity

bug fix for Intel GPUs

Browse Source
This commit is contained in:
Meifeng Lin 2023-12-19 08:03:28 -06:00
parent 1381dbc8ef
commit 37d1d87c3c
2 changed files with 5 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
Grid/threads/Accelerator.cc
View File

@ -14,15 +14,10 @@ void acceleratorThreads(uint32_t t) {accelerator_threads = t;};
#define ENV_LOCAL_RANK_MVAPICH "MV2_COMM_WORLD_LOCAL_RANK"
#define ENV_RANK_MVAPICH "MV2_COMM_WORLD_RANK"
#ifdef __CUDA_ARCH__
#warning "ifdef cuda arch"
#endif
// fold omptarget into device specific acceleratorInit()
#if defined(GRID_CUDA) || (defined(GRID_OMPTARGET) && defined(__CUDA_ARCH__))
#include <cuda_runtime_api.h>
//#if defined(GRID_CUDA) || (defined(GRID_OMPTARGET) && defined(__CUDA_ARCH__))
#if defined(GRID_OMPTARGET)
#warning "using cuda from opmtarget"
cudaDeviceProp *gpu_props;
cudaStream_t copyStream;
cudaStream_t computeStream;
@ -206,7 +201,7 @@ void acceleratorInit(void)
#endif
#if defined(GRID_SYCL) || (defined(GRID_OMPTARGET) && defined(__SYCL_DEVICE_ONLY__))
#if defined(GRID_SYCL) //|| (defined(GRID_OMPTARGET) && defined(__SYCL_DEVICE_ONLY__))
cl::sycl::queue *theGridAccelerator;
cl::sycl::queue *theCopyAccelerator;
@ -278,7 +273,7 @@ void acceleratorInit(void)
}
#endif
#if (!defined(GRID_CUDA)) && (!defined(GRID_SYCL))&& (!defined(GRID_HIP)) && (!defined(GRID_OMPTARGET))
#if (!defined(GRID_CUDA)) && (!defined(GRID_SYCL))&& (!defined(GRID_HIP))// && (!defined(GRID_OMPTARGET))
void acceleratorInit(void){}
#endif

4
benchmarks/Benchmark_su3.cc
View File

@ -36,8 +36,8 @@ int main (int argc, char ** argv)
{
Grid_init(&argc,&argv);
#define LMAX (31)
#define LMIN (31)
#define LMAX (8)
#define LMIN (8)
#define LADD (8)
int64_t Nwarm=500;