·hccl-test

HCCL (Huawei Collective Communication Library) performance testing for Ascend NPU clusters. Use for testing distributed communication bandwidth, verifying HCCL functionality, and benchmarking collective operations like AllReduce, AllGather, AlltoAll. Covers MPI installation, tool compilation, execution parameters, and result analysis.

34Installs·2Trend·@ascend-ai-coding

Installation

$npx skills add https://github.com/ascend-ai-coding/awesome-ascend-skills --skill hccl-test

How to Install hccl-test

Quickly install hccl-test AI skill to your development environment via command line

  1. Open Terminal: Open your terminal or command line tool (Terminal, iTerm, Windows Terminal, etc.)
  2. Run Installation Command: Copy and run this command: npx skills add https://github.com/ascend-ai-coding/awesome-ascend-skills --skill hccl-test
  3. Verify Installation: Once installed, the skill will be automatically configured in your AI coding environment and ready to use in Claude Code, Cursor, or OpenClaw

Source: ascend-ai-coding/awesome-ascend-skills.

SKILL.md

View raw

HCCL性能测试工具用于测试HCCL(Huawei Collective Communication Library)集合通信的功能正确性以及性能。

| Atlas 训练系列产品 | 4096 | - | | Atlas A2 训练系列产品 | 32K | - | | Atlas A3 训练系列产品/Atlas A3 推理系列产品 | 32K | AlltoAll/AlltoAllV 最大 8K | | Atlas 300I Duo 推理卡 | - | - |

| AllReduce | allreducetest | 多对多 | 梯度聚合、参数同步 | | AllGather | allgathertest | 多对多 | 数据聚合、参数收集 | | AllGatherV | allgathervtest | 多对多 | 变长数据聚合 | | AlltoAll | alltoalltest | 多对多 | 数据重排、负载均衡 | | AlltoAllV | alltoallvtest | 多对多 | 变长数据全对全通信 | | Broadcast | broadcasttest | 一对多 | 配置分发、初始化 | | Reduce | reducetest | 多对一 | 结果收集到根节点 |

HCCL (Huawei Collective Communication Library) performance testing for Ascend NPU clusters. Use for testing distributed communication bandwidth, verifying HCCL functionality, and benchmarking collective operations like AllReduce, AllGather, AlltoAll. Covers MPI installation, tool compilation, execution parameters, and result analysis. Source: ascend-ai-coding/awesome-ascend-skills.

Facts (cite-ready)

Stable fields and commands for AI/search citations.

Install command
npx skills add https://github.com/ascend-ai-coding/awesome-ascend-skills --skill hccl-test
Category
</>Dev Tools
Verified
First Seen
2026-02-25
Updated
2026-03-10

Browse more skills from ascend-ai-coding/awesome-ascend-skills

Quick answers

What is hccl-test?

HCCL (Huawei Collective Communication Library) performance testing for Ascend NPU clusters. Use for testing distributed communication bandwidth, verifying HCCL functionality, and benchmarking collective operations like AllReduce, AllGather, AlltoAll. Covers MPI installation, tool compilation, execution parameters, and result analysis. Source: ascend-ai-coding/awesome-ascend-skills.

How do I install hccl-test?

Open your terminal or command line tool (Terminal, iTerm, Windows Terminal, etc.) Copy and run this command: npx skills add https://github.com/ascend-ai-coding/awesome-ascend-skills --skill hccl-test Once installed, the skill will be automatically configured in your AI coding environment and ready to use in Claude Code, Cursor, or OpenClaw

Where is the source repository?

https://github.com/ascend-ai-coding/awesome-ascend-skills