self-diagnosing-resource-use

Name: self-diagnosing-resource-use
Availability: InStock
Author: yale-som-hpc

Diagnose Slurm job CPU, memory, and GPU utilization using sacct, seff, and nvidia-smi.

5|1|Updated Apr 29, 2026

Search Tags:#gpu #hpc #resource-management #slurm #job-accounting

Ready to run

npx skills add https://github.com/yale-som-hpc/claude-code-marketplace --skill self-diagnosing-resource-use

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: self-diagnosing-resource-use
Source: https://github.com/yale-som-hpc/claude-code-marketplace/tree/main/plugins/hpc/skills/self-diagnosing-resource-use
Command: npx skills add https://github.com/yale-som-hpc/claude-code-marketplace --skill self-diagnosing-resource-use

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps researchers evaluate whether a Yale SOM HPC Slurm job used its allocated CPUs, memory, GPUs, and wall time efficiently, and guides adjustments for the next submission.

Core Features & Use Cases

Post-run accounting with sacct and seff to quantify CPU, memory, GPU, and elapsed time usage.
CPU/memory/GPU diagnostics with clear interpretation rules to identify wasteful over-requesting or under-utilization.
Guidance and checks for adjusting future resource requests, job scripts, and data placement to optimize performance and fairness.

Quick Start

After a Slurm job completes, run the checkup to assess actual resource usage and plan the next job accordingly.

Frequently Asked Questions about self-diagnosing-resource-use

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I check Slurm job resource usage after completion?▼

To check Slurm job resource usage after completion, analyze accounting data from sacct and seff to quantify CPU efficiency, memory consumption, GPU utilization, and elapsed wall time for your finished jobs.

How do I diagnose low GPU utilization in my HPC jobs?▼

Diagnose low GPU utilization in HPC jobs by reviewing nvidia-smi data alongside sacct records to identify under-utilized GPU allocations and determine whether your job script requests more GPUs than it actively uses.

How do I right-size CPU and memory requests for Slurm job scripts?▼

Right-size CPU and memory requests for Slurm job scripts by comparing allocated resources against actual usage metrics from seff, then adjusting future submissions using practical thresholds to prevent wasteful over-requesting.

Can I use sacct and seff to evaluate wall-time usage on an HPC cluster?▼

Yes, you can use sacct and seff to evaluate wall-time usage on an HPC cluster by reviewing elapsed time metrics to determine if your job submission overestimated the required runtime.

What thresholds should I use to adjust HPC resource allocations?▼

Use practical efficiency thresholds derived from CPU, memory, GPU, and wall-time usage data to guide adjustments, ensuring future HPC resource allocations match actual workload demands and promote cluster fairness.

Why does my Slurm job show high memory allocation but low usage?▼

A Slurm job shows high memory allocation but low usage when the requested memory limit exceeds actual consumption, a pattern identified through seff diagnostics that indicates you should reduce memory requests in future job scripts.