
WSL 3, unveiled at Build 2026, replaces the Hyper-V VM with a lighter paravirtualized machine for near-native GPU and NPU throughput via DirectML 2.0. Tools like Ollama, llama.cpp, and PyTorch can use system accelerators directly from Linux without the performance tax of WSL 2.
Tap to vote and see what everyone thinks.
Summary by ByteBrief
AMD openSIL and Coreboot now run on EPYC