Identify CUDA performance bottlenecks from Nsight Systems profiling data.
npx skills add https://github.com/VCERS/MatClaw --skill nsys-optimizer
Please help me install this Agent Skill. Skill: nsys-optimizer Source: https://github.com/VCERS/MatClaw/tree/main/skills/nsys-optimizer Command: npx skills add https://github.com/VCERS/MatClaw --skill nsys-optimizer
Profiling and optimizing CUDA-based simulations suffer from slow runtimes and opaque bottlenecks; this skill provides a structured method to baseline, diagnose, optimize, and verify improvements using Nsight Systems.
Run an initial Nsight Systems profiling session on your CUDA application to collect baseline performance data.
High-intent search queries and answers about installing and using this skill.
To identify CUDA performance bottlenecks, run an initial Nsight Systems profiling session to collect baseline data. This skill analyzes the profiling results to pinpoint hotspot kernels and memory patterns, guiding a structured workflow to diagnose and target improvements for your GPU-accelerated application.
The best way to optimize GPU-accelerated simulations is an iterative loop: profile with Nsight Systems, diagnose bottlenecks, apply targeted improvements to hotspot kernels, and re-profile to verify performance gains. This structured method ensures repeatable measurements and verified acceleration.
Yes, you need Nsight Systems data and CUDA kernels. The optimization workflow requires baseline profiling data to drive repeatable measurements, analyze hotspot kernels, diagnose memory bottlenecks, and verify iterative improvements for your application.
Yes, scipy is supported as a dependency. You can use scipy alongside Nsight Systems data to assist with performance benchmarking and profiling analysis, driving the diagnosis of bottlenecks and verification of improvements in your CUDA applications.
CUDA applications suffer from slow runtimes and opaque bottlenecks due to inefficient kernel execution and memory patterns. By applying a systematic Nsight Systems profiling workflow, you can baseline performance, diagnose the specific memory bottlenecks, and optimize hotspot kernels.
igamenovoerConfigure CUDA build environments for Pixi projects with CMake and nvcc.
m0atAnalyze CUDA kernel performance bottlenecks with Nsight Systems and Nsight Compute.
xjtulycRun end-to-end psychometric validation on questionnaire item response datasets.
fsx950223Profile GPU kernels with rocprofv3 and generate optimization plans.
