skills.rest
alexsun07
alexsun07COMMUNITY
Explore all skills by this author →
GitHub

sglang-torch-profiler-analysis

Analyze SGLang torch-profiler output to generate kernel, overlap, and fuse-pattern reports.

1|Updated Feb 7, 2026
Search Tags:#deep learning#performance analysis#model optimization#SGLang#torch-profiler
One-click install
npx skills add https://github.com/alexsun07/agent_skills --skill sglang-torch-profiler-analysis-alexsun07
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sglang-torch-profiler-analysis
Source: https://github.com/alexsun07/agent_skills/tree/main/skills/sglang-torch-profiler-analysis
Command: npx skills add https://github.com/alexsun07/agent_skills --skill sglang-torch-profiler-analysis-alexsun07

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires torch, torch-profiler, sglang, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides an automated analysis of SGLang torch-profiler output, helping users quickly identify performance bottlenecks and optimization opportunities in their models.

Core Features & Use Cases

  • Performance Analysis: Inspects existing trace.json(.gz) files or profiles directories to generate a compact report.
  • Kernel Mapping: Maps kernels back to Python code paths, providing insights into model behavior.
  • Overlap and Fusion Analysis: Determines overlap headroom and potential fusion opportunities, suggesting optimizations.
  • Use Case: When profiling a live SGLang server, this Skill can automatically analyze the output and return a report highlighting areas for improvement.

Quick Start

Analyze the profile output from a live server using the sglang-torch-profiler-analysis skill.

Frequently Asked Questions about sglang-torch-profiler-analysis

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

FAQPage Schema
How do I analyze SGLang torch-profiler output for performance bottlenecks?▼

To analyze SGLang torch-profiler output, you can use an automated skill that inspects trace.json(.gz) files or profiling directories to identify performance bottlenecks and optimization opportunities in deep learning models.

What is kernel mapping in deep learning model profiling?▼

Kernel mapping in deep learning model profiling is the process of tracing GPU kernels back to their original Python code paths to provide insights into model behavior and locate performance bottlenecks.

How do I find overlap headroom and fusion opportunities in torch-profiler traces?▼

Finding overlap headroom and fusion opportunities involves analyzing torch-profiler traces to determine potential parallel execution gaps and suggesting kernel fusion optimizations to improve model performance.

Do I need torch and sglang libraries to run a performance analysis on trace files?▼

Yes, you need the torch, torch-profiler, and sglang Python libraries installed in your environment to profile a live SGLang server and analyze its output trace files.

Can I profile a live SGLang server and automatically generate a performance report?▼

Yes, you can profile a live SGLang server and automatically generate a compact performance report that includes kernel, overlap-opportunity, and fuse-pattern tables for optimization.

Related Skills

Explore More Software Engineering Skills→
FutureMLS-LabFutureMLS-Lab
556

sglang-torch-profiler-analysis

Analyze SGLang torch-profiler traces to identify dominant GPU kernels and fusion patterns.

Official
Advanced
dyyounggdyyoungg

llm-torch-profiler-analysis

Analyze LLM torch-profiler traces to identify kernel, overlap, and fusion patterns.

Community
Advanced
idanmoradarthasidanmoradarthas
37

ds-utils-metrics

Analyze and visualize sklearn-compatible model performance with metrics and plots.

Community
Intermediate
LancerLabLancerLab

perf-nsight-compute-analysis

Analyze NVIDIA Nsight Compute reports to classify GPU kernel bottlenecks.

Official
Intermediate