mccl-test

Community

Efficiently test and debug GPU communication for distributed training.

Authordongg622
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill streamlines the testing and debugging of multi-GPU communication setups, ensuring reliable data transfer in distributed training environments.

Core Features & Use Cases

  • Performance Testing: Runs benchmark scripts like mccl.sh to evaluate communication bandwidth on multiple GPUs.
  • Debugging Assistance: Provides environment checks, logs analysis, and troubleshooting steps for common GPU communication issues.
  • Use Case: Imagine a developer facing slow or failed multi-GPU communication; this Skill allows them to perform quick tests and gather diagnostic information to identify bottlenecks or misconfigurations.

Quick Start

Use the mccl-test skill to run a performance benchmark on 8 GPUs and automatically review the results.

Dependency Matrix

Required Modules

None required

Components

scriptsreferences

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: mccl-test
Download link: https://github.com/dongg622/china-ai-chip-skill/archive/main.zip#mccl-test

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.