Is there a way to achieve cross-node point-to-point communication without consuming SM (Streaming Multiprocessor) resources?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Improving Network Performance of HPC Systems Using NVIDIA Magnum IO NVSHMEM and GPUDirect Async | 3 | 653 | March 10, 2025 | |
| What is the setting to enable NVSHMEM on multi-node multi-GPU platform with IB | 7 | 200 | January 2, 2025 | |
| NVSHMEM applicability in setting of two PCIe connected GPUs | 6 | 1034 | October 12, 2021 | |
| NVSHMEM working and Unified Virtual Addressing (UVA) | 2 | 35 | April 8, 2025 | |
| Can I achieve multi-GPUs communication in one node through MPI? | 1 | 360 | January 20, 2021 | |
| Maximizing OpenMM Molecular Dynamics Throughput with NVIDIA Multi-Process Service | 1 | 24 | June 4, 2025 | |
| How to get SM utilization metrics for processes using MPS? | 1 | 138 | November 5, 2024 | |
| Measure SM utilization per process | 1 | 1326 | January 11, 2024 | |
| What is the best way to partition the SM of a GPU? | 2 | 1147 | August 17, 2023 | |
| How NVLink Will Enable Faster, Easier Multi-GPU Computing | 10 | 762 | June 15, 2016 |