Please make sure to check off these prerequisites before submitting a bug report.
[x] Test that the bug appears on the current version of the master branch. Make sure to include the commit hash of the commit you checked out.
[x] Check that the issue hasn't already been reported, by checking the currently open issues.
[x] If there are steps to reproduce the problem, make sure to write them down below.
[x] If relevant, please include the hls4ml project files, which were created directly before and/or after the bug.
Quick summary
CNN implementation on Pynq-Z2 doesn't seem to pass the data through the board correctly. Running NeuralNetworkOverlay.predict() on the board gets stuck after debug logs "Transfer OK". The model is "Resource" optimized with hls4ml due to lack of resources on the Pynq-board.
Details
CNN model created with Tensorflow/Keras and somewhat follows the hls4ml-tutorials CNN-implementation notebook.
Pynq version 2.6., hls4ml version 0.7.0, Vivado 2020.1.
Expected behavior
Neural network Overlay runs prediction normally.
Actual behavior
Prediction gets stuck after debug message "Transfer OK".
Additional context
I had problems even fitting a CNN on the PYNQ-Z2 board with hls4ml. It might just be an resource issue and the implemented neural network isn't working because of that. I'm new to FPGA-development so I don't quite grasp the transformation process and what's done behind the APIs when I build and generate the bitfiles for the board.
Prerequisites
Please make sure to check off these prerequisites before submitting a bug report.
Quick summary
CNN implementation on Pynq-Z2 doesn't seem to pass the data through the board correctly. Running
NeuralNetworkOverlay.predict()
on the board gets stuck after debug logs "Transfer OK". The model is "Resource" optimized with hls4ml due to lack of resources on the Pynq-board.Details
CNN model created with Tensorflow/Keras and somewhat follows the hls4ml-tutorials CNN-implementation notebook. Pynq version 2.6., hls4ml version 0.7.0, Vivado 2020.1.
Expected behavior
Neural network Overlay runs prediction normally.
Actual behavior
Prediction gets stuck after debug message "Transfer OK".
Additional context
I had problems even fitting a CNN on the PYNQ-Z2 board with hls4ml. It might just be an resource issue and the implemented neural network isn't working because of that. I'm new to FPGA-development so I don't quite grasp the transformation process and what's done behind the APIs when I build and generate the bitfiles for the board.