GPU Compute is now going to be available on the WSL 2 as announced by Microsoft. Native Linux command –line tools are allowed by WSL to run directly on Windows 10. WSL stands for Windows Subsystem for Linux which is running on about 3.5M monthly active devices. It is said to be one of the most requested features by Microsoft.
Artificial Intelligence (AI) and Machine Learning (ML) workflows are the ones the preview of GPU Compute for WSL2 seems to be compatible with also allowing WSL users in the operation of Machine Learning training workloads directly on Windows. There has been revelation of plans of Microsoft about it getting the preview of GPU Compute available within WSL for Windows Insiders ready.
Clarke Rahrig, the Program Manager of Windows AI Platform Team, in his own words in a recent blog post of his said “We are excited to let you know that the preview of GPU compute is now available within WSL 2 to Windows Insiders (Build 20150 or higher)! This preview will initially support artificial intelligence (AI) and machine learning (ML) workflows, enabling professionals and students alike to run ML training workloads across the breadth of GPUs in the Windows ecosystem.”
On the other hand, NVIDIA GPU users with a native Linux environment are setting up NVIDIA CUDA preview in WSL 2 on advice. Workflows on AMD and Intel GPUs are currently accelerated by this package.
If you are familiar with the Linux environment, then you can run the TensorFlow with DirectML package inside WSL 2. But if you’re familiar with Windows then you can do the same only on native Windows.
Windows 10 Insider Preview Build 20150 also brings in wsl-install a new command that allows the user to install WSL with single command along with GPU Compute support. The wsl-update is also added which is basically a command allowing users to manage the Linux kernel used by WSL 2 distros, by Microsoft.