ROCm/rocm-systems

What it solves

This repository acts as a "super-repo" that consolidates multiple ROCm systems projects into a single location. It solves the problem of fragmented development by streamlining continuous integration (CI), integration, and development workflows for various ROCm libraries and tools, specifically those required for building PyTorch.

How it works

It organizes various ROCm components—such as hip, rocprofiler, and rocr-runtime—into a structured directory system (projects/ and shared/). By migrating standalone repositories into this unified structure, it enables shared tooling and unified build and test workflows across the different libraries.

Who it’s for

It is designed for developers and contributors working on the ROCm ecosystem, particularly those building or maintaining the low-level systems software that supports AI frameworks like PyTorch.

Highlights

• Unified Workflow: Consolidates numerous standalone projects to improve visibility and collaboration.
• PyTorch Support: Focuses on providing the necessary system requirements for building PyTorch.
• Comprehensive Tooling: Includes a wide array of components including profilers (rocprofiler), runtime environments (rocr-runtime), and hardware interfaces (amdsmi).
• Standardized Nomenclature: Standardizes project naming to match released packages, removing legacy inconsistencies.