- How to Implement Performance Metrics in CUDA C/C++
- Improving Network Performance of HPC Systems Using NVIDIA Magnum IO NVSHMEM and GPUDirect Async
- NCCL vs NVSHMEM
- INTRODUCTION TO CUDA’s MULTI-PROCESS SERVICE (MPS)
Query About GPU Settings: --help-query-gpu to get all the available settings
sudo nvidia-smi --query-gpu compute_mode --format=csv
Enabling NVIDIA MPS
nvidia-smi -c EXCLUSIVE_PROCESS
nvidia-cuda-mps-control –d
Disabling NVIDIA MPS
echo quit | sudo nvidia-cuda-mps-control
sudo nvidia-smi -i 0 -c DEFAULT
GPU Topology
sudo nvidia-smi topo -m