converting-cutile-to-julia

Convert cuTile Python kernels into cuTile.jl implementations with 1-based indexing and broadcasting.

796|82|Updated Nov 13, 2025

Search Tags:#python #gpu #convert #julia #cutile #kernel-port

Ready to run

npx skills add https://github.com/NVIDIA/TileGym --skill converting-cutile-to-julia

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: converting-cutile-to-julia
Source: https://github.com/NVIDIA/TileGym/tree/main/.claude/skills/converting-cutile-to-julia
Command: npx skills add https://github.com/NVIDIA/TileGym --skill converting-cutile-to-julia

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Converts cuTile Python kernels into cuTile.jl implementations, ensuring correct indexing, broadcasting, and memory layout across languages to accelerate cross-language development and performance tuning.

Core Features & Use Cases

Translation patterns: 1-based indexing, broadcasting semantics, and 2D/batched layouts alignment between Python and Julia cuTile.
Validation workflow: static checks, compilation and runtime tests, and a structured validation loop that ensures correctness before deployment.
Porting scenarios: port an existing Python cuTile kernel to Julia cuTile.jl, debug translation discrepancies, and optimize performance with Julia tooling.

Quick Start

Follow the workflow in translations/workflow.md to convert a cuTile Python kernel to cuTile.jl and verify with the Julia test suite.

Frequently Asked Questions about converting-cutile-to-julia

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

FAQPage Schema

How do I convert Python cuTile kernels to Julia?▼

To convert Python cuTile kernels to Julia, translate 1-based indexing, broadcasting semantics, and memory-layout mappings into cuTile.jl implementations while enforcing strict kernel-signature rules and launch parameter conventions.

What is the best way to port GPU kernels from Python to Julia?▼

Porting GPU kernels from Python to Julia involves aligning 2D and batched workloads with cuTile.jl conventions, enforcing strict type mappings, and running a structured validation loop with static checks and runtime tests.

Why does my translated cuTile kernel have indexing or broadcasting errors in Julia?▼

Translation discrepancies in cuTile kernels usually stem from incorrect 1-based indexing conversions or mismatched broadcasting behavior between Python and Julia cuTile, requiring structured debugging and validation workflows.

Do I need to manually adjust memory layouts when porting cuTile kernels to Julia?▼

Yes, converting cuTile kernels requires exact handling of memory-layout mapping for 2D and batched workloads to ensure correct alignment between Python and Julia cuTile implementations.

Can I optimize existing Python cuTile kernels using Julia tooling?▼

You can optimize existing Python cuTile kernels by porting them to cuTile.jl, utilizing Julia's performance tooling while maintaining strict kernel-signature rules and launch parameter conventions.