croq-tune

Automate GPU kernel tuning through iterative profiling, ideation, and implementation.

Updated Apr 10, 2026

Search Tags:#automation #deep learning #performance optimization #profiling #compiler #GPU #kernel tuning

Ready to run

npx skills add https://github.com/LancerLab/croqtile-tuner --skill croq-tune

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: croq-tune
Source: https://github.com/LancerLab/croqtile-tuner/tree/main/.claude/skills/croq-tune
Command: npx skills add https://github.com/LancerLab/croqtile-tuner --skill croq-tune

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill automates the process of tuning GPU kernels by managing iterative profiling, idea generation, and implementation, thus optimizing GPU performance.

Core Features & Use Cases

Automated Tuning Loop: Launches ongoing experiments to discover optimal kernel configurations using AI-driven strategies.
Workflow Management: Coordinates profiling, idea proposals, implementation, verification, measurement, and decision-making processes.
Use Case: A developer wants to maximize GPU TFLOPS for a custom kernel without manual trial-and-error, by letting the AI continually refine and optimize the kernel parameters.

Quick Start

Launch the croq-tune skill to start an indefinite tuning session targeting your GPU and specific kernel parameters, with minimal user intervention.

Frequently Asked Questions about croq-tune

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

FAQPage Schema

How do I automate GPU kernel tuning for deep learning workflows?▼

Automated GPU kernel tuning is handled by launching an indefinite AI-driven session that manages iterative profiling, idea generation, and implementation to maximize performance. The continuous cycle coordinates measurement and decision-making to refine kernel parameters with minimal manual intervention.

What is the best way to optimize GPU TFLOPS for custom kernels without manual trial-and-error?▼

Optimizing GPU TFLOPS for custom kernels requires continuous profiling and structural changes managed by AI-driven strategies. The automated tuning loop proposes, implements, verifies, and measures configurations to discover optimal performance parameters without manual trial-and-error.

Can I use automated profiling to improve kernel efficiency in high-performance computing tasks?▼

Automated profiling is integrated directly into the tuning cycle to improve kernel efficiency for high-performance computing tasks. The workflow continuously profiles, proposes ideas, implements structural changes, and measures results to maximize performance within complex workflows.

Does automated kernel tuning require manual intervention during the optimization cycle?▼

Automated kernel tuning requires minimal user intervention once the session is launched. The workflow enforces strict automation and safety protocols throughout the cycle, independently managing profiling, implementation, verification, and decision-making processes.

When do I need AI-driven kernel optimization for my GPU workflows?▼

AI-driven kernel optimization is needed when developers want to maximize GPU performance for deep learning or high-performance computing tasks without manual trial-and-error. It continuously refines kernel parameters by integrating multiple profiles and structural changes within complex workflows.