Can Deep learning HDL toolbox be used for a customized FPGA board without DDR memory?

Hello. I am a MATLAB user. Now I have a project where needs to deploy a bi-directional LSTM neural network into an FPGA board. The network has already been trained and tested in MATLAB R2021B.

I’ve tried to use the Deep Learning HDL Toolbox product in MATLAB for generating the required Verilog code but have run into a problem. The FPGA board I’m using is a customized board, and the FPGA model is Xilinx Artix-7 family XC7A75T FGG484ABX1917, which is not one of the supported devices listed in the toolbox user guide. Besides, the board does not have any DDR memory chips embedded (using on-chip RAM to implement the data storage).

I would like to know if this board is compatible with the toolbox? Does the use of that toolbox necessarily require the FPGA board to have DDR memory with it?

Many thanks for your reply.

Answer 1

[Bharathi Yogaraj]

Below example shows how to create custom board and generate a deep learning processor IP core for the custom board.

https://www.mathworks.com/help/deep-learning-hdl/ug/define-custom-board-and-reference-design-for-dl-ip-core-workflow.html

DDR memory is compulsary for deployment of DL IP as of R2023a release.