搜索

PythonReactSEOPDF
OpenClaw

musify

Convert CUDA code to MUSA (Moore Threads GPU) using the musify tool. Use when migrating CUDA codebases to MUSA platform, converting CUDA kernels/APIs to MUSA equivalents, or porting NVIDIA GPU code to Moore Threads GPUs. Supports batch conversion of .cu, .cuh, .cpp, .h files with text-based API mapping.

dongyang-mt0 来源安装数☆ 0 仓库 Stars

安装与使用

请根据来源仓库和你的 AI 工具说明安装此技能。先阅读技能文件和依赖要求,再添加到工作流。

SkillHub CLI

skillhub install dongyang-mt/musify

打开来源仓库,查看说明和兼容环境后,按来源提供的命令或步骤安装。

将来源链接提供给你的 Agent,让它先读取说明、核对依赖,并向你说明将执行的安装步骤。

查看来源 ↗使用指南

技能说明

Musify - CUDA to MUSA Code Conversion

Musify converts CUDA code to MUSA (Moore Threads GPU architecture) using text-based API mapping.

Installation

# Install dependencies
pip install ahocorapy

# musify-text should be available in MUSA toolkit

Basic Usage

# Convert files in-place (CUDA -> MUSA)
musify-text --inplace -- file1.cu file2.cpp

# Convert and create new files (default)
musify-text --create -- a.cu b.cpp

# Print to stdout
musify-text -t -- file.cu

Batch Conversion

# Find and convert all CUDA/C++ files in directory
musify-text --inplace -- $(find /path/to/project -name '*.cu' -o -name '*.cuh' -o -name '*.cpp' -o -name '*.h')

# Using ripgrep (recommended)
musify-text --inplace -- $(rg --files -g '*.cu' -g '*.cuh' -g '*.cpp' -g '*.h' /path/to/project)

Options

Option Description
-t, --terminal Print output to stdout
-c, --create Create new files with converted code (default)
-i, --inplace Modify files in-place
-d {c2m,m2c} Conversion direction: c2m (CUDA→MUSA, default), m2c (MUSA→CUDA)
-m <file.json> Custom API mapping file (can specify multiple)
--clear-mapping Clear default mapping, use only custom mappings
-l {DEBUG,INFO,WARNING} Log level

Exclusion Markers

Prevent specific code sections from being converted:

// MUSIFY_EXCL_LINE - Exclude this line
char *str = "cudaMalloc"; // MUSIFY_EXCL_LINE

// MUSIFY_EXCL_START - Start exclusion block
char *apis[] = {
    "cudaInit",
    "cudaFree"
};
// MUSIFY_EXCL_STOP - End exclusion block

Common CUDA → MUSA Mappings

CUDA MUSA
cuda prefix musa prefix
CUDA MUSA
cu prefix (driver API) mu prefix
__cuda __musa
cudaMalloc musaMalloc
cudaFree musaFree
cudaMemcpy musaMemcpy
cudaLaunchKernel musaLaunchKernel
__global__ __global__ (unchanged)
`__devi...

来源信息

Source
github
Repository
openclaw/skills
Path
skills/dongyang-mt/musify
Version
—
Commit
040e893c69bc3bf65ee0cea619b84a293c63a3e6
分类来源
rules
状态
active
可见性
public

这些信息来自收录时的来源分析,可能滞后;使用前请核对最新来源文件。

来源内容

SKILL.md
# Musify - CUDA to MUSA Code Conversion

Musify converts CUDA code to MUSA (Moore Threads GPU architecture) using text-based API mapping.

## Installation

```bash
# Install dependencies
pip install ahocorapy

# musify-text should be available in MUSA toolkit
```

## Basic Usage

```bash
# Convert files in-place (CUDA -> MUSA)
musify-text --inplace -- file1.cu file2.cpp

# Convert and create new files (default)
musify-text --create -- a.cu b.cpp

# Print to stdout
musify-text -t -- file.cu
```

## Batch Conversion

```bash
# Find and convert all CUDA/C++ files in directory
musify-text --inplace -- $(find /path/to/project -name '*.cu' -o -name '*.cuh' -o -name '*.cpp' -o -name '*.h')

# Using ripgrep (recommended)
musify-text --inplace -- $(rg --files -g '*.cu' -g '*.cuh' -g '*.cpp' -g '*.h' /path/to/project)
```

## Options

| Option | Description |
|--------|-------------|
| `-t, --terminal` | Print output to stdout |
| `-c, --create` | Create new files with converted code (default) |
| `-i, --inplace` | Modify files in-place |
| `-d {c2m,m2c}` | Conversion direction: c2m (CUDA→MUSA, default), m2c (MUSA→CUDA) |
| `-m <file.json>` | Custom API mapping file (can specify multiple) |
| `--clear-mapping` | Clear default mapping, use only custom mappings |
| `-l {DEBUG,INFO,WARNING}` | Log level |

## Exclusion Markers

Prevent specific code sections from being converted:

```c
// MUSIFY_EXCL_LINE - Exclude this line
char *str = "cudaMalloc"; // MUSIFY_EXCL_LINE

// MUSIFY_EXCL_START - Start exclusion block
char *apis[] = {
    "cudaInit",
    "cudaFree"
};
// MUSIFY_EXCL_STOP - End exclusion block
```

## Common CUDA → MUSA Mappings

| CUDA | MUSA |
|------|------|
| `cuda` prefix | `musa` prefix |
| `CUDA` | `MUSA` |
| `cu` prefix (driver API) | `mu` prefix |
| `__cuda` | `__musa` |
| `cudaMalloc` | `musaMalloc` |
| `cudaFree` | `musaFree` |
| `cudaMemcpy` | `musaMemcpy` |
| `cudaLaunchKernel` | `musaLaunchKernel` |
| `__global__` | `__global__` (unchanged) |
| `__devi...

相关技能

OpenClaw

bambu-studio-ai

heyixuan2

Bambu Lab 3D printer control and automation. Activate when user mentions: printer status, 3D printing, slice, analyze model, generate 3D, AMS filament, print monitor, Bambu Lab, or any 3D printing task. Full pi...

OpenClaw

geo-optimization

capt-marbles

Generative Engine Optimization (GEO) for AI search visibility. Optimize content to appear in ChatGPT, Perplexity, Claude, and Google AI Overviews. Use when optimizing websites, pages, or content for LLM discove...

OpenClaw

parakeet-stt

carlulsoe

Local speech-to-text with NVIDIA Parakeet TDT 0.6B v3 (ONNX on CPU). 30x faster than Whisper, 25 languages, auto-detection, OpenAI-compatible API. Use when transcribing audio files, converting speech to text, o...

OpenClaw

feishu-process-feedback

carlzhao007

飞书消息自动处理与进度反馈技能。安装后后台运行,监听飞书任务消息并自动创建独立进程处理。 在处理前后发送实时进度反馈(任务确认、进度百分比、完成通知)。 支持任务类型识别、智能解析、错误重试、并发控制、状态持久化...