Downloading deep learning to NI FPGAs

[Montagner__Guilherme]

Hi, good morning to everyone.I'm new to Intel development environment. But I need to deploy a custom deep learning solution to a custom FPGA board.

My FPGA model is Xilinx Kintex-7, which is embedded on a National Instruments Compact-RIO

I've been oriented to use Intel FPGA RTE for OpenCL but I still have some questions.

how can I run the command with the "-d FPGA" flag?

Thanks,Guilherme

[SuryaPSC_Intel]

Hi Montagner,

Please confirm you have followed all the steps of Install Intel® Distribution of OpenVINO™ toolkit for Linux* with FPGA Support. You may also refer to FPGA Plugin for more information. Also, please verify you are using one of the OpenVINO Supported Devices.

Best Regards,

Surya

[Montagner__Guilherme]

Hi.
It is not on the list of supported devices, but another user, Iffa, told me I could try installing Intel RTE for FPGA and trying to run a command with the -d flag on, to check if it worked.

Thanks for your reply

[Iffa_Intel]

Hi there,

Yes, what i meant with that -d is the aocl command lol,

Somewhat like the one I attached.

Note that the previous steps that I gave is just a workaround,  Openvino only fully supports Intels boards.

It might luckily works or it might not works at all.

Thanks,

Sincerely,

Iffa

[Montagner__Guilherme]

Hi. Thanks for your response.
I tried running the classification_sample_async.exe with the -d FPGA flag, and got an error on dliaPlugin.dll.
I suppose that happens because I don't have a supported FPGA board attached.
When I run aocl diagnose I get "Cannot find board_env.xml in C:\...\a10_ref, which I suppose is the Arria 10 board which is specified somewhere as the one being used.
What do yo uthink? Should I test it again with another command?

[Max_L_Intel]

Hi Guilherme.

OpenVINO toolkit currently supports FPGA devices only based on Intel Arria 10 and Arria 10 GX boards. So we don't expect that being workable for non-Arria FPGAs.

Best regards, Max.