Blackwell

Linux 7.2 Seeds "Blackwell-Next": A Deep Dive into the nvgrace-gpu VFIO CXL DVSEC Change

Linux 7.2’s VFIO pull request dropped a commit with a codename I hadn’t seen before: Blackwell-Next. A Phoronix post brought this to my attention - Linux 7.2 Begins Making Preparations For NVIDIA “Blackwell-Next” - which, on the face of it looks like a minor prep patch. It is — but it’s also a clean window into where NVIDIA is taking its CPU-coherent GPU stack, how CXL is quietly becoming the standard signaling interface for next-generation accelerators, and what that means if you’re building infrastructure or tooling on top of these platforms.

vLLM Recipe: RedHatAI/Qwen3.6-35B-A3B-NVFP4 on DGX Spark

This is a vLLM Recipe - a production-ready Docker Compose configuration for running open-weight models on local hardware. It documents the exact setup, configuration rationale, and benchmark results so you can get a model running quickly. You are welcome to change the parameters to suit your workloads. This worked for me, so I hope you find it helpful.

This recipe covers Qwen3.6-35B-A3B-NVFP4 - a Mixture-of-Experts model with 35B total parameters but only ~3B active at inference - quantized to NVFP4 by Red Hat AI and running on the NVIDIA DGX Spark (my GigaByte AI Top Atom) with a GB10 Blackwell GPU and 128 GB of unified CPU/GPU memory.