Test cases to validate the correctness of hpctoolkit for GPU-accelerated applications.
Clone
git clone --recursive https://github.com/Jokeren/hpctoolkit-gpu-samples
Setup
export OMP_NUM_THREADS=<#threads>
export HPCTOOLKIT_GPU_TEST_REP=<#repeat times>
Run
CUDA Programs
cd <sample path>
make ARCH=<GPU arch>
./<application name> <device id (default 0)>
OpenMP Programs
cd <sample path>
make SHOWFLAGS="-L<OpenMP path> -lomp"
LD_LIBRARY_PATH=<OpenMP path>:$LD_LIBRARY_PATH ./<application name> <device id (default 0)>
| Case | Purpose |
|---|---|
| cu_multi_contexts_multi_streams | Within each context, multiple CPU threads launch kernels to streams concurrently |
| Case | Purpose |
|---|---|
| cuda_vec_add | cudaLaunchKernel |
| cuda_cooperative_group | cudaLaunchCooperativeKernel |
| cu_vec_add | cuLaunchKernel |
| cu_multi_entries | cuLaunchKernel for difference kernels with the same calling context |
| cu_cooperative_group | cuLaunchCooperativeKernel (ERROR) |
| target_vec_add | omp target |
| Case | Purpose |
|---|---|
| cu_call_path | acyclic call graph |
| cu_call_path_recursive | recursive device function calls |
| cu_call_path_recursive_mutual | mutual recursive device function calls |
| cu_call_path_long | long call path with missing samples in the middle |
| cu_call_path_thread_aware | different CPU threads pass different parameters |
| cuda_call_path_dynamic | dynamic parallelism |
| cuda_call_path_dynamic_recursive | recursive call with dynamic parallelism |
| Case | Purpose |
|---|---|
| nvdisasm | nvdisasm correctness check samples |
| cuobjdump | cuobjdump correctness check samples |
| cupti_test | cupti_test correctness check samples |
| Case | Purpose |
|---|---|
| cuda_pc_sampling_tuning | pc sampling is performed on all SMs independently |
| cuda_shared_memory_stall | no stall reason indicates shared memory latency |
| Case | Purpose | URL |
|---|---|---|
| Laghos | large-scale application; RAJA and CUDA programming model comparison | https://github.com/CEED/Laghos |
| target_lulesh | OMP Target performance | https://computation.llnl.gov/projects/co-design/lulesh |
| RAJAPerf | CUDA and RAJA performance test suite | https://github.com/LLNL/RAJAPerf |
| sw4 | 3-D seismic modeling | https://github.com/geodynamics/sw4 |
| cuda_tensor_contraction | nekbone | https://nek5000.mcs.anl.gov/ |
| cuda_tensor_transpose | ExaTENSOR | https://iadac.github.io/projects/ |
| target_tensor_transpose | OMP Target version of ExaTENSOR | https://iadac.github.io/projects/ |