NVIDIA Topology-Aware GPU Selection (NVTAGS) is a toolset for high-performance computing (HPC) applications that enables faster solve times with high GPU communication-to-application run-time ratios. NVTAGS intelligently and automatically assigns GPUs to message passing interface (MPI) processes, thereby reducing overall GPU-to-GPU communication time.
Dramatically improves performance by intelligently mapping MPI processes to GPUs for HPC applications with non-uniform GPU-to-GPU communication.
Provides a fully automated process that eliminates the manual, time-consuming mapping of processes to GPUs.
Can be easily integrated into a container, making it portable and easily deployable on any system or architecture.
- Automated topology detection independent of the system and HPC application
- Small memory footprint (few KB) with minimal one-time profiling overhead
- Automatic caching of mapped results for future simulations
- Support for single-node, PCIe, and NVIDIA® NVLink® with NVIDIA Pascal™, Volta™, and Ampere GPUs
- Support for customizing default mapping (e.g., mapping algorithm) and profiling metrics
- Up to 22% performance increase for Chroma
- Up to 41% performance increase for MILC
Ready to get started with NVTAGS?