ci.rocm.debian.net: a Debian CI with AMD GPUs, for AI/ML packages
summary
An overview of ci.rocm.debian.net, a Debian continuous-integration system forked from Debian’s official CI to run on AMD GPUs using the ROCm stack (open-source alternative to NVIDIA CUDA). Covers how the team bootstrapped a GPU-enabled CI with support for a dozen+ AMD GPU architectures, the policy and technical challenges encountered, and concrete steps for maintainers to adapt packages so they are exercised by this ROCm-focused CI.
AI-written summary · about
related talks
05mBuild AI Anywhere with ROCm™ Software: AMD and PyTorch Bring...
03mKeynote: AMD & PyTorch: A Powerful Combination for Generative AI
32mPowering Your Generative AI Workloads with AMD and Open-Source ROCm
20mBringing PyTorch Monarch to AMD GPUs: Single-Controller Distributed Tra...
15mDeploy Hugging Face Models on Azure powered by AMD ROCm software
13mFrom Silicon to Scale: Training at Scale
41mPanel: AI Infra Best Practices: Enterprise Do’s and Don’ts
12mLightning Talk: PyTorch 2.0 on the ROCm Platform