hello-rocmROCm tutorials and practice cases for AMD GPUs

A structured learning path for AMD ROCm in LLM workflows, from environment setup and deployment to fine-tuning and operator optimization.

Start with environment setup

View practice cases

简体中文

hello-rocm Skill

Give this project to your AI assistant in one copy

Copy the sentence below and paste it into any tool that supports Skills, Rules, or Agent configuration. Let the tool decide how to load src/hello-rocm-skill.

Use it for GPU architecture lookup, ROCm quick installs, vLLM / Ollama / llama.cpp deployment, troubleshooting, and learning paths.

Unified environment baseline

Set up ROCm, PyTorch, and uv first so deployment, fine-tuning, and operator optimization tutorials share the same foundation.

LLM deployment

Deploy models on AMD GPUs with LM Studio, vLLM, Ollama, llama.cpp, and related ROCm workflows.

LLM fine-tuning

Follow ROCm fine-tuning examples for models such as Qwen3 and Gemma4.

Operator optimization

Learn AMD AI hardware, the ROCm software stack, HIP operators, PyTorch custom operators, and performance-oriented practices.

References

Curated ROCm, AMD GPU, and AI development resources for deeper reading.

AMD practice cases

Community examples covering local agents, model training, vision applications, and developer tools on AMD platforms.

Recommended Path

Read Environment first and finish the ROCm, PyTorch, and Python toolchain setup.
Move to Deploy and validate your environment with LM Studio, vLLM, or another inference workflow.
Continue with Fine-tune to understand LoRA fine-tuning on ROCm.
Explore Operator Optimization if you want to go deeper into low-level performance and operator development.
Check AMD Practice when you need complete project examples.

Content Map

Section	Content
Environment	ROCm installation, configuration, validation, and GPU architecture references
Deploy	Multi-framework local deployment for models such as Qwen3 and Gemma4
Fine-tune	LLM fine-tuning notes and examples on ROCm
Operator Optimization	AMD AI hardware, ROCm software stack, HIP operators, and PyTorch custom operators
References	ROCm and AMD AI ecosystem resources
AMD Practice	Application cases and community projects for AMD platforms

Who This Is For

Learners who have an AMD GPU and want to run or fine-tune LLMs locally.
Developers who want a structured ROCm workflow instead of scattered setup notes.
Practitioners who want to move from inference deployment into operator development and performance analysis.