Xilinx / Vitis-AI

Vitis AI is Xilinx’s development stack for AI inference on Xilinx hardware platforms, including both edge devices and Alveo cards.
https://www.xilinx.com/ai
Apache License 2.0
1.44k stars 624 forks source link

xdputil status - bus error #1314

Open hudini87 opened 1 year ago

hudini87 commented 1 year ago

Hi all, I'm getting Bus error when executing xdputil status and can't solve it.

Vitis AI version: Vitis AI 2.0 vivado & Petalinux version: 2021.2 SOM: MERCURY XU8 ( ZYNQ MPSOC ZCU7EV)

I'm working in vivado flow (DPU IP and not integrated DPU complied by VITIS) , 1 DPU and connections as appeared in example design.

as mention in https://xilinx.github.io/Vitis-AI/3.0/html/docs/workflow-system-integration.html :

"For Vitis AI releases >= v2.0, Vivado users (Zynq™ Ultrascale+™ and Kria™ applications) must compile VART standalone without XRT, ", i removed XRT from rootfs and put

i see as expected:

root@petalinux:~# dmesg |grep dpu [ 19.312314] xlnx-dpu: Xilinx Deep Learning Processing Unit driver [ 19.318710] xlnx-dpu 81000000.dpuczdx8g: Freq: axilite: 99 MHz, dpu: 249 MHz, dsp: 499 MHz [ 19.345987] xlnx-dpu 81000000.dpuczdx8g: found 1 dpu @250MHz and 0 softmax, dpu registered as /dev/dpu successfully

but when i am trying to probe and communicate with the dpuczdx8g :

root@petalinux:~# xdputil status /usr/bin/xdputil: line 20: 1043 Bus error /usr/bin/python3 -m xdputil $*

root@petalinux:~# show_dpu device_core_id=0 device= 0 core = 0 fingerprint = 0x1000020f6014407 batch = 1 full_cu_name=unknown:dpu0

Can anyone help me please?

quentonh commented 1 year ago

@hudini87 I believe this will help you:

https://xterra2.avnet.com/xilinx/zcu106/zcu106-dpu-example-vivado-flow/-/blob/main/README.dpu_fixed_address_issue.md