What's Changed
Added
- Add
nvidia-ml-py13.610.43 to support list by @XuehaiPan. - Add
nvidia-ml-py13.615.71 to support list by @XuehaiPan. - Add a responsive Rich GPU dashboard example with host metrics, MIG support, and GPU process snapshots by @XuehaiPan in #231.
Fixed
- Guard NVML field queries against a buffer overrun in NVIDIA driver 595.91.07 that can crash
nvitop-exporterwhile collecting NVLink throughput by @XuehaiPan in #232. - Fix shared-memory accounting on unified-memory devices (e.g., NVIDIA DGX Spark and RTX Spark N1X) to include reclaimable cache in free memory and calculate used memory from available capacity by @johnnynunez in #234.
Full Changelog: v1.7.1...v1.7.2