skills.rest
khrore
khroreCOMMUNITY
Explore all skills by this author →
GitHub

mojo-gpu-fundamentals

Write portable Mojo GPU kernels without CUDA syntax.

Updated Apr 15, 2025
Search Tags:#gpu#kernel#tiling#mojo#memory-management#warp#cuda-mapping
One-click install
npx skills add https://github.com/khrore/nix-config --skill mojo-gpu-fundamentals-khrore
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mojo-gpu-fundamentals
Source: https://github.com/khrore/nix-config/tree/main/dotfiles/common/.codex/skills/mojo-gpu-fundamentals
Command: npx skills add https://github.com/khrore/nix-config --skill mojo-gpu-fundamentals-khrore

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Mojo developers face the challenge of writing GPU-targeted code without CUDA syntax, and this guidance provides standardized Mojo GPU concepts and mappings to bridge the gap.

Core Features & Use Cases

  • CUDA-to-Mojo kernel mappings, memory management primitives, and synchronization patterns.
  • Tiling, vectorization, and shared memory usage with Mojo equivalents for portable GPU kernels.
  • Use Case: port a CUDA kernel to Mojo and run it on NVIDIA, AMD, or Apple Silicon GPUs with minimal changes.

Quick Start

Write a simple Mojo GPU kernel using the mapping guide and run it on a compatible accelerator.

Frequently Asked Questions about mojo-gpu-fundamentals

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

FAQPage Schema
How do I write GPU kernels in Mojo without using CUDA syntax?▼

Mojo GPU porting replaces CUDA syntax with declarative frontmatter-driven structures and explicit imports. It maps kernel definitions, memory management, and synchronization patterns to run natively across NVIDIA, AMD, and Apple Silicon GPUs.

Can I port existing CUDA kernels to Mojo and run them on AMD or Apple GPUs?▼

Yes, you can port existing CUDA kernels to Mojo for AMD or Apple GPUs. The Skill provides CUDA-to-Mojo kernel mappings and memory primitives, enabling minimal code changes for cross-platform accelerator execution.

What is the best way to handle memory management and tiling for Mojo GPU kernels?▼

The best way to handle memory management and tiling for Mojo GPU kernels is by using Mojo's provided equivalents for shared memory and vectorization. These primitives ensure portable and maintainable kernel execution across different hardware.

Does Mojo GPU programming support synchronization patterns across different hardware vendors?▼

Mojo GPU programming supports synchronization patterns across NVIDIA, AMD, and Apple GPUs. It enforces well-defined APIs and standardized concepts to maintain correctness and portability across various hardware vendors.

Do I need explicit imports and declarative frontmatter for Mojo GPU kernel definitions?▼

Yes, you need explicit imports and declarative frontmatter for Mojo GPU kernel definitions. This structure enforces well-defined APIs, ensuring portability, correctness, and maintainability across your Mojo GPU projects.

Related Skills

Explore More Software Engineering Skills→
Harmeet10000Harmeet10000
7

mojo-gpu-fundamentals

Explain CUDA-free Mojo GPU programming with LayoutTensor and shared memory patterns.

Community
Advanced
AscendAscend
31

ascendc-operator-performance-optim

Diagnose and optimize Ascend C operator performance across tiling, data movement, and memory layout.

Official
Advanced
Hundo1018Hundo1018
7

mojo-gpu-fundamentals

Write, launch, and debug Mojo GPU kernels with TileTensor and shared memory.

Community
Advanced
limit5limit5

SKILL_HD_VENDOR_SYNC

Monitor HD camera vendor SDK repositories and sync upstream changes into OmniSight firmware.

Community
Intermediate