GindaChen

GindaChen / nsys-ai

Public

Terminal UI for NVIDIA Nsight Systems profiles — timeline viewer, kernel navigator, NVTX hierarchy

gindachen.github.ionsys-ai
18
6
100% credibility
Found Mar 07, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

nsys-ai is an AI-powered tool for analyzing NVIDIA Nsight Systems profiles, offering web UI, terminal TUI, and HTML export features to visualize GPU kernel timelines and NVTX hierarchies.

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is nsys-ai?

nsys-ai is a Python tool that visualizes and analyzes NVIDIA Nsight Systems profiles from your terminal or browser, turning raw .nsys-rep files into interactive timelines, kernel hierarchies, and NVTX trees. Run `pip install nsys-ai` then `nsys-ai my_profile.nsys-rep` to open a multi-GPU timeline viewer with zoom, search, and AI chat—no CUDA or extra deps needed. It solves the pain of sifting through massive GPU traces without firing up the full Nsight GUI, especially on remote Linux servers where you want quick kernel and hierarchy insights via nvidia terminal linux commands.

Why is it gaining traction?

Unlike Nsight's GUI or raw SQL queries, nsys-ai delivers instant web timelines with progressive rendering, Perfetto-style terminal TUIs for timeline and NVTX hierarchy navigation, and one-shot diffs like `nsys-ai diff before.sqlite after.sqlite`. The AI agent (`nsys-ai agent ask profile.sqlite "why bubbles?"`) spots pipeline issues, idle gaps, and NCCL overlap automatically. Developers love the zero-config install and exports to Perfetto JSON or HTML for sharing.

Who should use this?

GPU performance engineers profiling PyTorch training loops on multi-node clusters, ML ops folks debugging kernel launch overhead or memory transfers in terminal nvidia driver sessions on Ubuntu, and teams iterating on distributed models who need fast before/after comparisons without desktop tools. Ideal for linux terminal nvidia driver workflows where nvidia smi terminal monitoring meets deep trace analysis.

Verdict

Grab it if you're knee-deep in Nsight traces—solid for quick terminal kernel hierarchy dives despite 18 stars and 1.0% credibility score signaling early alpha maturity. Docs shine with CLI refs and schema explorers, but expect occasional rough edges; pair with nsys export sqlite for best results.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.