skills.rest
dongg622
dongg622COMMUNITY
Explore all skills by this author →
GitHub

profiling-computing

Analyze Ascend NPU profiling logs and CSV files to identify compute bottlenecks.

Updated Aug 27, 2026
Search Tags:#performance#optimization#profiling#NPU#ascend#operator analysis
One-click install
npx skills add https://github.com/dongg622/china-ai-chip-skill --skill profiling-computing
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: profiling-computing
Source: https://github.com/dongg622/china-ai-chip-skill/tree/main/Ascend/profiling-analysis/profiling-computing
Command: npx skills add https://github.com/dongg622/china-ai-chip-skill --skill profiling-computing

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill helps identify slow and inefficient computation processes within Ascend NPU systems, improving performance tuning.

Core Features & Use Cases

  • Performance Bottleneck Analysis: Detects high-耗时算子 and provides insights to improve system efficiency.
  • Operator Shape Extraction: Extracts detailed shape information of key operators for deeper analysis.
  • Use Case: A developer notices slow inference times and uses this Skill to pinpoint the operators causing delays, then modifies the model accordingly.

Quick Start

Use the profiling-computing skill to analyze a profiling directory and generate a performance bottleneck report.

Frequently Asked Questions about profiling-computing

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

FAQPage Schema
How do I identify performance bottlenecks in Ascend NPU systems?▼

To identify performance bottlenecks in Ascend NPU systems, you analyze profiling data to detect high-overhead operators. This process pinpoints inefficient computations causing slow inference times, enabling targeted model modifications.

Can I extract operator shape information from Ascend profiling logs?▼

Yes, you can extract detailed operator shape information from Ascend profiling logs. This extracts key operator shapes for deeper performance analysis, facilitating the diagnosis of compute bottlenecks within your model.

How do I analyze Ascend profiling CSV files for slow inference times?▼

You analyze Ascend profiling CSV files by running performance diagnostics to detect compute bottlenecks. This compatibility with profiling logs allows you to pinpoint specific operators causing inference delays.

What is the best way to optimize Ascend hardware performance?▼

The best way to optimize Ascend hardware performance is performing detailed analysis of NPU system profiling data. This identifies compute bottlenecks and operator inefficiencies, directly facilitating performance tuning.

Do I need specific dependencies to perform Ascend NPU operator analysis?▼

No specific dependencies are required to perform Ascend NPU operator analysis. The Skill operates independently using provided scripts and references to analyze profiling directories and generate performance bottleneck reports.

Related Skills

Explore More Software Engineering Skills→
xiechyxiechy
3

get-available-resources

Detect host CPU, GPU, memory, and disk resources into a .claude_resources.json file.

Community
Intermediate
zechenzhangAGIzechenzhangAGI
11.5k

model-pruning

Prune large language models using one-shot methods like Wanda and SparseGPT.

Community
Intermediate
mindspore-aimindspore-ai
258

triton-ascend-fused-operator-optimization

Generate deterministic fusion plans for Ascend NPU operators.

Official
Advanced
timurgaleevtimurgaleev
5

canary

Monitor post-deploy web apps for defects and performance regressions.

Community
Advanced