topic Re: Training on Intel GPU in Intel® Optimized AI Frameworks

Training on Intel GPU

Eltablawy__Alaa — Thu, 25 Feb 2021 00:03:26 GMT

Hi,

Is it supported to train TensorFlow in python on Intel GPU or FPGA? I understand that OpenVINO can be used by python code to target specific hw for inference only, so what's the case for training?

Re:Training on Intel GPU

PrasanthD_intel — Thu, 25 Feb 2021 07:39:40 GMT

Hi,

We have a dedicated thread for AI-related issues(Intel® Optimized AI Frameworks - Intel Community). I'm moving this thread to that forum for better support.

Regards

Prasanth

Re:Training on Intel GPU

AdrianM_Intel — Thu, 25 Feb 2021 18:26:30 GMT

Hello Eltablawy__Alaa,

If you have further questions or need support on your workload optimization, Please submit your queries at the TensorFlow GitHub issues with the label "comp:mkl" or the https://community.intel.com/t5/Intel-Optimized-AI-Frameworks/bd-p/optimized-ai-frameworks.

Regards,

Adrian M.

Intel Customer Support Technician

Re:Training on Intel GPU

AthiraM_Intel — Fri, 26 Feb 2021 13:04:22 GMT

Hi,

Thanks for reaching out to us.

OpenVINO is a toolkit provided by Intel to facilitate faster inference of deep learning models. OpenVINO is not used for training.

Regarding the support to train TensorFlow in python on Intel GPU or FPGA, we will check with the concerned team and get back to you soon.

Thanks.

Re: Training on Intel GPU

AthiraM_Intel — Wed, 03 Mar 2021 12:02:33 GMT

Hi,

Regarding the GPU or FPGA support for training tensorflow in python, we will have Intel dGPU support in the future release, but FPGA support is not in current roadmap.

If you have any further issue, please let us know.

Thanks.

Re:Training on Intel GPU

AthiraM_Intel — Mon, 08 Mar 2021 11:16:00 GMT

Hi,

Could you please give us an update? Has the solution provided helped?

Thanks.

Re:Training on Intel GPU

AthiraM_Intel — Tue, 16 Mar 2021 04:13:06 GMT

Hi,

We have not heard back from you. We would discontinue monitoring this issue . If you need further assistance, please post a new thread.

Thanks