Hyperloom documentation

ROCm Hyperloom is an autonomous agentic system designed to optimize end-to-end inference workloads (targeting both host code and GPU kernels) on AMD GPUs. Using advanced AI agents and profiling tools, Hyperloom analyzes your workload, identifies performance bottlenecks, implements targeted optimizations, and validates the performance and correctness of the optimizations without requiring manual intervention.

The Hyperloom source code is hosted on GitHub at AMD-AGI/Hyperloom.

Install

• Quickstart – Docker and bare metal

• Quickstart – Slurm

Components

• Components

• Hyperloom optimization loop

How to

• Run a Hyperloom optimization

• Quantization with AMD Quark

• Multi-node inference optimization demo

Reference

• API reference

• Environment variables

• Authentication and credentials

• Operations and self-hosting

• Upgrade guide

• Session output schema

• Knowledge base integration

• Operator scripts

• Troubleshooting

To contribute to the documentation, see Contributing to Hyperloom.

Hyperloom is released under the MIT license. For details, see the License page.