ncu-profiling
Provides tools for deep kernel performance profiling.
Install
mkdir -p .claude/skills/ncu-profiling && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/18671" && unzip -o skill.zip -d .claude/skills/ncu-profiling && rm skill.zipInstalls to .claude/skills/ncu-profiling
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Use for kernel-level analysis with Nsight Compute after a benchmark is reproducible.Key capabilities
- →Confirm benchmark reproducibility
- →Run `profile-ncu.ps1` script
- →Capture output path in `artifacts/profiles/`
- →Interpret GPU occupancy
- →Interpret memory throughput
How it works
The skill executes a profiling script with Nsight Compute on a reproducible benchmark. It then interprets the captured kernel-level metrics to analyze GPU performance.
Inputs & outputs
When to use ncu-profiling
- →Profiling kernel performance
- →Analyzing GPU metrics
- →Debugging performance bottlenecks
About this skill
Nsight Compute Profiling
Use this skill for kernel-level analysis with Nsight Compute.
Workflow
- Confirm benchmark reproducibility first.
- Run
scripts/perf/profile-ncu.ps1. - Capture the output path in
artifacts/profiles/. - Interpret occupancy, memory throughput, shared memory pressure, and instruction mix.
Guardrail
Do not make optimization claims from one metric alone.
When not to use it
- →When making optimization claims from one metric alone
- →When the benchmark is not reproducible
Limitations
- →Requires benchmark reproducibility
- →Does not make optimization claims from one metric alone
- →Focuses on kernel-level analysis with Nsight Compute
How it compares
This skill standardizes the Nsight Compute profiling workflow and guides the interpretation of specific GPU metrics, unlike manually running the profiler and interpreting raw data.
Compared to similar skills
ncu-profiling side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| ncu-profiling (this skill) | 0 | 4mo | No flags | Intermediate |
| code-coverage-with-gcov | 15 | 3mo | Review | Intermediate |
| 3d-games | 16 | 6mo | No flags | Advanced |
| cpp-pro | 18 | 3mo | No flags | Advanced |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by loveSunning
View all by loveSunning →You might also like
code-coverage-with-gcov
gadievron
Add gcov code coverage instrumentation to C/C++ projects
3d-games
davila7
3D game development principles. Rendering, shaders, physics, cameras.
cpp-pro
sickn33
Write idiomatic C++ code with modern features, RAII, smart pointers, and STL algorithms. Handles templates, move semantics, and performance optimization. Use PROACTIVELY for C++ refactoring, memory safety, or complex C++ patterns.
debug-lldb
regenrek
Capture and analyze thread backtraces with LLDB/GDB to debug hangs, deadlocks, UI freezes, IPC stalls, or high-CPU loops across any language or project. Use when an app becomes unresponsive, switching contexts stalls, or you need thread stacks to locate lock inversion or blocking calls.
pc-games
davila7
PC and console game development principles. Engine selection, platform features, optimization strategies.
function-call-tracing
gadievron
Instrument C/C++ with -finstrument-functions for execution tracing and Perfetto visualization