Diagnose remote GPU usage via SSH with per-GPU memory and process reports.
npx skills add https://github.com/majiayu000/spellbook --skill gpu-use
Please help me install this Agent Skill. Skill: gpu-use Source: https://github.com/majiayu000/spellbook/tree/main/skills/gpu-use Command: npx skills add https://github.com/majiayu000/spellbook --skill gpu-use
This Skill enables quick visibility into remote GPU usage by SSH-ing into a server and presenting per-GPU memory usage, running processes, and the containers those processes belong to.
SSH into the target server and run the GPU usage diagnostic to display per-GPU memory, running processes, and container mappings.
High-intent search queries and answers about installing and using this skill.
Monitor remote GPU usage by SSH-ing into the target server to collect per-GPU memory, utilization, and running processes. It coordinates nvidia-smi and docker commands to generate a readable report of GPU activity.
Yes, you can map running GPU processes to Docker containers on a remote server. It correlates PIDs from nvidia-smi with container names via SSH, providing clear attribution for active processes across hosts.
To check remote GPU usage, you need Bash as an allowed tool and SSH readiness to the target server. This allows you to execute nvidia-smi and docker commands for data collection and process-to-container mapping.
Remote GPU diagnostics works by collecting per-GPU memory, utilization, and process lists via SSH. It maps processes to container names to identify memory bottlenecks and active processes across server clusters.
Yes, you can use a custom SSH target to diagnose GPU utilization on different hosts. It allows specifying custom SSH targets or using a default server alias for flexible remote monitoring.
The best way to get a structured report of GPU resources is by running a GPU usage diagnostic via SSH. It fetches memory, utilization, and process-to-container mappings to generate a concise view of cluster activity.
caw111Orchestrate machine learning experiments on SSH remote GPU servers with OOM retries.
mf-dynatraceMonitor and analyze host and process metrics using DQL.
sayalinvidiaInstall Holoscan SDK in a Docker container and validate GPU-enabled execution.
zerwizProvision and manage Lambda Labs GPU cloud instances for ML training.
