Opencl/Cuda support about linfa HOT 5 CLOSED

Probably not, at least not anytime soon .. this is at least blocked on rust-ndarray/ndarray#840 and then we have to re-implement many of the LAPACK routines in GPU accelerated versions. This is just not feasible with the current size/interest in the ecosystem. We are similar to scikit-learn in scope and application.

If you want a neural network+AD library, à la tensorflow, then take a look at tch-rs which interfaces with Torch.

from linfa.

but we should add a section to the README explaining the situation! Will do this later this day

from linfa.

I will close this because it is really out of scope

from linfa.

We now have a new repo that implements CUDA in rust here. I can help implementing this if issue is reopened.

from linfa.

The Rust-CUDA repo is still in the experimental stage and the ndarray GPU integration ticket is still open, so this is still a no-go. We should really add this explanation to the README though.

from linfa.