Current State of Graphics Virtualization Upstream
summary
Three upstream virtualization models exist and trade performance for portability: direct hardware virtualization gives near-native speed but is vendor‑specific and SKU‑restricted; API virtualization (virtio‑gpu/Virgl) doubles CPU work by duplicating OpenGL state tracking and caching on guest and host yet supports tens of millions of users; the mediated approach balances both. Vulkan‑based Venus and Mesa/kernel implementations already land or are available, Zinc layers GL over Vulkan, and a kernel Excel framework is approaching merge for NPU/ML virtualisation.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:40Context
- 1:53Where
- 6:54How
- 11:55API Virtualization
- 20:17Vulcan Virtualization
- 26:18OpenGL Virtualization
related talks
33mGPU Para-Virtualization on Xen
29mXenGT: A software based Intel Graphics Virtualization Solution
39mnouveau and kernel GPU VMA management
38mStateless Video Decoder Virtualization through Virtio Video and VA-API
03h 01mHands-on Introduction to Virtualization
08mNova GPU Workshop: vGPU + VFIO: status and 2026 plans
46mGraphic Testing Without Hardware: Discovering the Power of VKMS!
23mHyper-V's Virtual Secure Mode in KVM project update