PYNQ: PYTHON PRODUCTIVITY

DPU PYNQ ZCU104

Hi,

I’m trying to run DPU PYNQ on ZCU104. I follow the guide to download

  • git clone --recursive --shallow-submodules GitHub - Xilinx/DPU-PYNQ: DPU on PYNQ
  • cd DPU-PYNQ/upgrade
  • sudo make
  • sudo pip3 install pynq-dpu
  • cd $PYNQ_JUPYTER_NOTEBOOKS
  • sudo pynq get-notebooks pynq-dpu -p .

It works well with the dpu_resnet50.ipynb example. When I try to change the .xmodel which I compiled from Vitia AI by myself, but after long time SSH connection get disconnected while executing the jupyter notebook.

Can someone help me? Thanks you very much.

What version of VitisAI / PYNQ SDCard are you using? Are you recompiling the resnet xmodel?

I’m using PYNQ v2.6 and Vitis-AI 1.3.2(latest). Yes, I try to recompile resnet xmodel from Vitis-AI Model Zoo. Also, there is a resnet xmodel which is already compiled, it didn’t work either.

Can you make sure you’re using the correct Vitis-AI tag / docker to rebuild and that you target the ZCU104 dpu config in the DPU-PYNQ repo? If they don’t line up correctly, we’ve seen xmodel’s hang.

I’m following the Vitis-AI flow which is from download from Vitis-AI Model Zoo.
This is my compile result.