Run Smaller Models Locally on AMD Radeon™ GPUs
Experiment with using AMD ROCm™ for AMD Radeon GPU programming. Find sample playbooks, local AI development tools, tutorials, open-source projects, documentation, and more.
Install ROCm:
Run Sample Local AI Workloads
Try out the ROCm software stack for edge AI use cases.
Featured Workloads
Getting Started with Ollama
Run large language models locally with Ollama and start chatting via terminal, desktop app, or REST API in minutes.
Get Access to AMD Radeon GPUs
Want to Test Drive AMD Radeon GPU programming with ROCm?
Experiment with powerful AMD Radeon GPUs for developing local AI applications based on your geography.
Learning Resources
Explore more topics to find relevant learning resources to help you stay updated with the latest technological advancements from AMD.
Technical Blogs
Find example implementations, tips, and updates for AI developers by AI developers.
Documentation
Learn about ROCm tools, compilers, runtimes, and debugging or deployment and orchestration on AMD Instinct GPUs.
Community and Support
Join the AMD AI Developer Program
Access cloud credits, connect with AMD engineers, receive early invitations to events, gain recognition opportunities, enter member only sweepstakes, publish technical content, and unlock exclusive training and workshops
Join AMD AI Academy
AMD AI Academy offers expert-led, self-learning courses for AI developers, enterprise AI professionals, and kernel developers.
Frequently Asked Questions
Developing local AI applications on GPUs is ideal when you need low latency, real-time responsiveness, cost control, or data privacy without relying on the cloud. By programming AMD Radeon GPUs with ROCm, developers can run AI workloads locally for use cases such as:
- Local LLMs and copilots for retrieval-augmented generation (RAG), coding assistance, and private AI applications
- Creative and media workflows like image generation, video and audio enhancement, and 3D rendering
- AI-powered gaming and real-time graphics
Successive ROCm releases have seen significant leaps in performance. See the performance results page to get a reference point for evaluating how AMD Instinct GPUs perform with ROCm software running popular AI models for inference (e.g., vLLM, xDiT) and training (e.g., PyTorch, Megatron-LM, and JAX MaxText).
AMD Radeon RX, Radeon PRO, and Radeon AI Pro GPUs with ROCm support offer powerful local AI fine-tuning and inference. PyTorch can be installed for ROCm on these GPUs using PIP (for Linux and Windows) or Docker (for Linux) installation methods. The AMD ROCm team actively contributes to open-source development and collaborates closely with framework organizations, including PyTorch.
