Clojurians Log v2
uncomplicate
Yes. Neanderthal supports GPU through CUDA (nvidia) and OpenCL (Intel, Nvidia, AMD, etc.).
For NN, you actually want Deep Diamond. On linux it supports onednn, which works on both cpu, and gpu (intel develops it, so I'd be surprised if it doesn't support ARC, although I don't have that hardware, so obviously I haven't had opportunity to try it out).
New Diamond ML supports many execution providers for NN inference through onnxruntime, so amond these million options, I doubt that you won't be to find one for your hardware, and If you can't you'll probably have trouble finding it anywhere else 🙂
thanks. I wasn't worried about code not working at all but rather about which specific Intel behaviors could be leveraged. If that sounds vague its because wow! the Intel marketing department makes up so many terms that don't match the way I think about software. This page was very helpful https://github.com/uxlfoundation/oneDNN
Hi neanderthal people 😉 I'm looking to setup a new linux developer workstation and am a bit confused on neanderthal's ability to take advantage of CPU/GPU capabilities...especially base CPU and iGPU. I want to get a recent model Intel Core Ultra 7 which contains an iGPU. From the Intel product naming side it has "Intel ARC graphics with 8 Xe cores". Does Neanderthal target drivers that take advantage of this? Any specific advantage of the CPU? I'm avoiding adding a separate NVIDIA card for now. I just want a linux dev machine for small-ish neural net dev and some mid size inference. I think it comes down to naming...both hardware and the GPU programming models have shifting terms which makes all this a bit confusing. thanks, Jon