Forum Discussion

Montagner__Guilherme's avatar
Montagner__Guilherme
Icon for New Contributor rankNew Contributor
5 years ago

Deep learning on NI FPGA

Hello,
I am new to Intel development environments.
In my project, I aim to download a deep learning inference solution to a Kintex-7 FPGA on NI CompactRIO-9040
I wanted to know whether it is possible to do this within the OpenCL framework, possibly using some Intel SDK or other software.
I can also switch to a Zynq-7000 SoC which is present on some NI devices.
Is it possible?

Thank you,
Guilherme

3 Replies

  • GRN's avatar
    GRN
    Icon for Occasional Contributor rankOccasional Contributor

    Hi,

    Kintex-7 FPGA is Xilinx board.

    I think, you should use Xilinx OpenCL software like SDAccel

  • Hi GRN, thank you for your response.
    Apparently SDAccel doesn't support kintex-7, only newer accelerator devices.
    I saw that I can design a custom platform with Intel FPGA SDK Pro edition.
    My question is whether it would work, given the FPGA is embedded on a SoC, inside the compactRIO controller, which is accessible from a PC with a USB cable (not PCIe). I could also try to install Intel FPGA SDK and RTE on the compactRIO controller, which in turn communicates to the FPGA in a more direct manner

    Thanks in advance

  • My question is: Is it possible to design a custom platform for the NI FPGA using Intel FPGA SDK Custom platform toolkit, and run it on the SoC processor? Does it need to have PCIe communication to the FPGA?