cuda-optimizer

Name: cuda-optimizer
Availability: InStock
Author: jarmak-personal

Analyze CUDA kernel code and produce performance improvements.

Updated Mar 10, 2026

Search Tags:#performance #optimization #gpu #kernel #code-review #cuda

Ready to run

npx skills add https://github.com/jarmak-personal/vibeSpatial --skill cuda-optimizer

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: cuda-optimizer
Source: https://github.com/jarmak-personal/vibeSpatial/tree/main/.agents/skills/cuda-optimizer
Command: npx skills add https://github.com/jarmak-personal/vibeSpatial --skill cuda-optimizer

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill enables users to optimize existing CUDA and GPU kernel code, improving runtime efficiency and performance.

Core Features & Use Cases

Code Analysis: Reads CUDA kernel source files and associated dispatch logic to identify optimization opportunities.
Automated Rewrites: Produces concrete, applicable code modifications such as memory access improvements, synchronization elimination, and kernel source enhancements.
Use Case: Developers working on vibeSpatial GPU kernels can leverage this Skill to systematically analyze code and implement performance best practices, reducing manual trial-and-error efforts.

Quick Start

Use the cuda-optimizer Skill to review and optimize your existing CUDA kernel files in vibeSpatial.

Frequently Asked Questions about cuda-optimizer

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

FAQPage Schema

How do I optimize existing CUDA kernel code for better GPU performance?▼

To optimize existing CUDA kernel code, you can analyze kernel source files and dispatch logic to identify memory access improvements, synchronization elimination, and targeted rewrites that align with NVIDIA best practices.

What are common CUDA synchronization patterns that slow down GPU execution?▼

Unnecessary synchronization patterns in GPU modules often slow down execution; analyzing and eliminating these host-device boundary interactions helps produce concrete performance improvements and enhanced execution speed.

Can I automate code review for GPU kernel performance improvements?▼

Yes, you can automate GPU code review by reading existing CUDA kernel source files to identify optimization opportunities and produce concrete, applicable code modifications for enhanced runtime efficiency.

Does this CUDA optimization approach work with host-device boundary interactions?▼

This CUDA optimization approach specifically targets host-device boundary interactions and kernel source files to ensure code adheres to NVIDIA best practices through automated rewrites and suggestions.

What's the best way to reduce manual trial-and-error when improving CUDA kernel performance?▼

The best way to reduce manual trial-and-error is to systematically analyze your CUDA code to implement performance best practices through targeted rewrites and concrete suggestions for memory access and synchronization.