OpenCL & CUDA GPU support
It would be nice to have (relatively) easy access to the dedicated GPU's on our Win10 PC's, and OpenCL support for the AMD Drivers & CUDA for NVidia drivers etc.
What I am thinking on specifically (although there are many many uses for it and CUDA support) is hashing functions that are lackluster within windows itself, but benefit immensely within Linux and Unix.
Cryptocurrency mining for example, is known to be much faster on Linux systems than Windows. Do you see any support for OpenCL and/or CUDA in the future?
It’s on the backlog, but not currently planned.
We know many of you would like to be able to accelerate your numerical analysis code via CUDA/OpenCL, etc.
One thing to explore in the meantime is to try running your code in Windows. Windows actually has very powerful GPU-acceleration capabilities and you may well find that much of your existing code may well work already or require trivial effort to port to Windows itself.
in the meantime, bear with us, and UPVOTE THIS ISSUE if you care (please don’t create another duplicate ask – that doesn’t help us prioritize asks).
Nate Gardner commented
There are many large, Linux-only libraries that depend CUDA or OpenCL that have no comparable options on Windows. However, the outputs would be incredibly useful in a Windows environment. For instance, I have CV libraries that output visual odometry data. There's nothing that compares in terms of quality and performance available for Windows (most robotics developers use Linux), but the consumer of this data would preferably run in a Windows environment. With CUDA support in WSL, this would be possible.
Consider that this is far better than a GPU passthrough virtualization option or vGPU, because there's no hypervisor and the GPU can be actively used by both systems. I'm sure there are many other use-cases like mine!
Pate Blomqvist commented
Just starting on ML and was hoping to use WLS to easily manage packages and use the extensive resources that has been written for Ubuntu on Keras/Tensorflow/Theano setup.
Sadly it’s a no for now. Upvote!
Nikhil Verma commented
What's the status as of January 2018?
Qisong Wang commented
We are waitting for the 'SOMEDAY'.
its 2018 now what are u guys waiting for
Oleg Jakushkin commented
This is a top upvoted feature for a long time. A real deal-breaker. When will we see it in development (not ON THE BACKLOG)?
when are you going to support this feature ?
Hey I'm really willing to switch back to windows as soon as you guys support this.
You can try https://github.com/plaidml/plaidml
Upvoting because this is is a necessary thing.
The entire reason I was excited about WSL was to be able to use my GPU-accelerated code on Windows without having to port to Visual Studio, which is a huge effort...
This feature is still a necessity for a lot of people, please consider taking it off the backlog.
Shawn Hind commented
Would love to see this happen.
First few corrections from Microsoft admin:
- "One thing to explore in the meantime is to try running your code in Windows."
most commercial software companies dont have time to "try running in windows" : most need something that works under minimum effort: no recompilation, no source code change, just run: trying to recompile thousand of lines of code written for Linux is a total no way for the majority.
- "Windows actually has very powerful GPU-acceleration capabilities" : not better than Linux. Example: nvidia visual profiler
- "you may well find that much of your existing code may well work already or require trivial effort to port to Windows itself" : wrong assertion: only trivial programs like "Hello world" would easily pass.
I UPVOTE THIS ISSUE.
Pier Lim commented
It would be a killler feature - I can see more windows laptops sold if this is a reality, especially for developers.
Pawel Kunstman commented
Aaron Styles commented
I'm just diving into the world of ML, but have and AMD gpu. Community support for OpenCL is lacking, but there are some solutions out there that help, getting them to work on Windows is crazy hard.
Bill Gates commented
Come on, It's been a whole year, and this is still on the Back log... We need CUDA in our research and we don't want to bother physically installing Linux on every machines and frequently switch between operating systems.
Vignesh S commented
Guys, 2017 and beyond is the Era of Deep learning and Machine Learning, the Augumented Era. Not everyone are expert programmers and write from scratch using CUDA library. We use other high level open source libraries like Pytorch, Tensorflow, Keras, etc. So we need to be able to access the GPU. When this was announced "Bash on Windows" the very first thing that came to my mind was the amazing endless possibilities. But now it's just shattered and this has proven to be no good use. I still cant believe it's on a Backlog. At least start to work on it. If its going to come late its OK, but this has to come.