What does it do?
nvmon is an NVIDIA GPU monitor that runs as a single Python file with no dependencies. It switches between graph, process, and number focused views and lets users choose which GPUs to display. It groups processes by job and shows the account, runtime, memory usage, and command.
Key features
- Switch between GPU graph, process, and number focused views
- Choose which GPUs to display
- Show each process’s account, runtime, memory usage, and command
- Group launcher and worker processes into a single job
- Run on servers without internet access by copying the Python file
Why did it stand out?
It lets you check the status of each GPU and the processes for each job in a terminal on a shared GPU server. It runs as a single Python file with no dependencies, so it can be used on servers without internet access.
Weekly picks · October 12–October 18 ↗