$ The best sources for Linux GPU, CUDA and ROCm news

Where to follow GPU drivers, Mesa, CUDA and ROCm on Linux: vendor blogs, release notes and news sites, and how to combine them so updates do not break your setup.

Why GPU news matters on Linux

A working GPU setup on Linux is a stack: the kernel driver, the userspace driver (Mesa for the open drivers, or NVIDIA's proprietary driver), the compute layer (CUDA or ROCm) and the framework or inference engine on top. Most breakage happens when one layer moves and another does not.

Following the right sources means you see version changes before you upgrade, not after something stops working.

Open drivers, Mesa and the kernel

Phoronix closely tracks Mesa, the kernel graphics drivers and GPU benchmarks, and is often the first place a driver change is reported. Pair it with your distro's own news channel, which tells you when those changes actually reach your packages.

NVIDIA and CUDA

The NVIDIA Technical Blog explains new CUDA features, libraries and inference work. For upgrade decisions, the CUDA Toolkit release notes and the driver release notes are the primary sources: they list required driver versions, deprecated GPUs and known issues.

AMD and ROCm

The AMD ROCm blogs cover performance work and how-tos for running models on AMD GPUs. The ROCm releases page on GitHub is the most reliable place to see what changed in each version; check the ROCm compatibility documentation before upgrading, because supported GPUs and distros can differ between releases.

Inference engines

GPU support often arrives through the inference engine rather than the driver. vLLM's release notes show when a new GPU generation, kernel or quantization path is supported, and are worth following if you serve models locally.

How to combine them

A practical setup is one news site for drivers (Phoronix), the release notes for the one compute stack you use (CUDA or ROCm), one vendor blog, and the release notes of your inference engine. Read the release notes before upgrading the driver or compute stack, and pin versions that work.

If you would rather not check these pages every day, Linux Doctor sends a free daily email with the new items from these sources and publishes the same list on the latest checkup page.

All sources mentioned

Phoronix

Linux hardware and open-source news: GPU drivers, Mesa, kernel features and benchmarks.

NVIDIA Technical Blog

CUDA, inference optimisation and GPU-accelerated AI deep dives.

AMD ROCm Blogs

AMD's ROCm engineering blog: running AI on Radeon and Instinct GPUs.

ROCm Releases

Release notes for the ROCm GPU compute stack.

vLLM Releases

Release notes for the vLLM high-throughput inference server.

Hacker News: Linux

Hacker News stories about Linux, kernels, distros and GPU stacks with 100+ points.

Related

GPU and driver newslatest checkupall sources