I am working on a requirement from a Customer who wants to setup a ADAS Platform.

[RJauh]

Need help in identifying, which Solution of Intel can fulfill this requirement?

Processor can be Skylake, Disc Size can be around 1 TB SSD In mirror, Low Graphics Support - unsure if NVIDIA GPU or Intel has a product ?

Local AI Accelerator - I am unsure whether it should be ARIA or Intel Movidus ?

[Aswathy_C_Intel]

Thanks for reaching out to us. This forum mainly handles technical questions related to Intel AI(frameworks and others). Since this is a question related to requirement analysis, we will check with the concerned team and let you know.

[Aswathy_C_Intel]

Presently we don't have a dedicated forum for handling this issue. It is recommended to refer to the below links for getting some ideas on what you are looking for. https://www.intel.com/content/www/us/en/automotive/automotive-overview.html https://blogs.intel.com/iot/tag/adas/#gs.i3zgtf Hope this helps. Please confirm.

[Aswathy_C_Intel]

Could you please confirm if the issue is resolved. Please be informed that the thread will get closed within 2 business days assuming that the solution provided was helpful.

[RJauh]

Hi I am still struggling to find a firm answer. I am working with Ritesh Patel ( Singapore) from Intel, if you know someone who knows the answer, please forward these Questions to him/her. Deep Learning Training using FPGA 1. Do the Intel FPGA support Caffe<https://www.zdnet.com/article/caffe2-deep-learning-wide-ambitions-flexibility-scalability-and-advocacy/>, CNTK<https://www.microsoft.com/en-us/cognitive-toolkit/>, DeepLearning4j<https://www.zdnet.com/article/machine-learning-in-the-cloud-is-the-new-battlefield/>, H2O<https://www.h2o.ai/>, MXnet<https://mxnet.apache.org/>, PyTorch<https://pytorch.org/>, SciKit<http://scikit-learn.org/stable/>, and TensorFlow ? Which model of FPGA? 2. Are FPGA suggested for use in Deep Learning Training? Which model ? 3. How do they compare to NVIDIA Tesla V 100? 4. Can FPGA be stacked, to get the required throughput? 5. Which one to use and where ? Intel Nervana –T or Intel Arria 10 or Intel Stratix 10? Thanks and regards Rahul Jauhari +919810136438

[William_J_Intel]

“1. Do the Intel FPGA support Caffe<https://www.zdnet.com/article/caffe2-deep-learning-wide-ambitions-flexibility-scalability-and-advocacy/>, CNTK<https://www.microsoft.com/en-us/cognitive-toolkit/>, DeepLearning4j<https://www.zdnet.com/article/machine-learning-in-the-cloud-is-the-new-battlefield/>, H2O<https://www.h2o.ai/>, MXnet<https://mxnet.apache.org/>, PyTorch<https://pytorch.org/>, SciKit<http://scikit-learn.org/stable/>, and TensorFlow ? Which model of FPGA?

Our OpenVINO tool (Platform agnostic) supports Caffe, TensorFLow, MxNet…for sure. Refer to OpenVINO webpage for more info

2. Are FPGA suggested for use in Deep Learning Training? Which model ?

FPGAs can do training and we have used them for the forward (Inferencing) path, but we encourage use of Nirvana Crest family of devices for training acceleration. FPGAs focus on inference really.

3. How do they compare to NVIDIA Tesla V 100?

Depends on model, batch size, power, TCO of the system. IN general the GPU can be more images/sec, but the question is about deployment in a system. WE keep the data in the FPGA and between FPGAs without needing to go back and forth to the host, so we often have higher system performance and almost always significantly lower latency.

4. Can FPGA be stacked, to get the required throughput?

Yes. This is especially beneficial when persistence is achieved. FPGAs tend to get significant performance gains here.

5. Which one to use and where ? Intel Nervana –T or Intel Arria 10 or Intel Stratix 10?”

Nervana is for training. Arria 10 is the only FPGA inference solution we have right now. WE will be supporting S10 shortly when D5005 card is officially released and OpenVINO supports it

[Aswathy_C_Intel]

We have limited answers for your questions. Hope all your questions are about Intel FPGA. Intel FPGA supports caffe and Tensorflow for inferencing through Openvino. For more details, Please refer : https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/solution-sheets/intel-fpga-dl-acceleration-suite-solution-brief%E2%80%93en.pdf. As of now, DL training is not supported in Intel FPGA. Multiple FPGAs could be used for increasing inference throughput. For further details, we will contact the core team and let you know if they can help.

[Aswathy_C_Intel]

Hi, we are moving this question to a more appropriate forum.

[Aswathy_C_Intel]

Hi, Could you please confirm if the answer provided by Williiam Jenkins was helpful. Please be informed that the thread will get closed within 2 business days assuming that the solution provided was helpful.

[Aswathy_C_Intel]

Since we have not received a response, we are closing this thread with the assumption that the issue got resolved. Feel free to open a new thread if you face further issues. After case closure, you will receive a survey email. We appreciate it if you can complete this survey regarding the support you received.