• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 98 subscribers
  • Views 695 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

SK-TDA4VM: J721EXSKG01EVM

Isidora Radovanovic
Prodigy 135 points

Part Number: SK-TDA4VM

Hi,

I'm trying to run a python script that compiles an ONNX model on the TDA4 starter kit, but I get the following error even though the tidl_model_import_onnx.so file is in the tidl_tools folder:


Error - /opt/edgeai-tidl-tools/tidl_tools/tidl_model_import_onnx.so: cannot open shared object file: No such file or directory
Segmentation fault (core dumped)

[docker] root@j7-evm:/opt/edgeai-tidl-tools/tidl_tools# ls
PC_dsp_test_dl_algo.out itidl_rt.h libtidl_tfl_delegate.so libvx_tidl_rt.so.1.0 tidl_graphVisualiser.out tidl_model_import_onnx.so tidl_model_import_tflite.so
device_config.cfg libtidl_onnxrt_EP.so libvx_tidl_rt.so ti_cnnperfsim.out tidl_graphVisualiser_runtimes.out tidl_model_import_relay.so

Please let me know how I can resolve this error.

Thanks,

Isidora

  • 0
    Prodigy 135 points

    When I try to run only model inference (with the model artifacts that were compiled on the EVM), the graph verification fails:

    /opt/edgeai-tidl-tools/tidl_tools
    libtidl_onnxrt_EP loaded 0x3bc73c30
    Final number of subgraphs created are : 1, - Offloaded Nodes - 128, Total Nodes - 128
    APP: Init ... !!!
    MEM: Init ... !!!
    MEM: Initialized DMA HEAP (fd=4) !!!
    MEM: Init ... Done !!!
    IPC: Init ... !!!
    IPC: Init ... Done !!!
    REMOTE_SERVICE: Init ... !!!
    REMOTE_SERVICE: Init ... Done !!!
    4393.973004 s: GTC Frequency = 200 MHz
    APP: Init ... Done !!!
    4393.973078 s: VX_ZONE_INIT:Enabled
    4393.973098 s: VX_ZONE_ERROR:Enabled
    4393.973116 s: VX_ZONE_WARNING:Enabled
    4393.973637 s: VX_ZONE_INIT:[tivxInitLocal:130] Initialization Done !!!
    4393.973820 s: VX_ZONE_INIT:[tivxHostInitLocal:86] Initialization Done for HOST !!!
    4393.977756 s: VX_ZONE_ERROR:[ownContextSendCmd:815] Command ack message returned failure cmd_status: -1
    4393.977782 s: VX_ZONE_ERROR:[ownContextSendCmd:851] tivxEventWait() failed.
    4393.977795 s: VX_ZONE_ERROR:[ownNodeKernelInit:538] Target kernel, TIVX_CMD_NODE_CREATE failed for node TIDLNode
    4393.977804 s: VX_ZONE_ERROR:[ownNodeKernelInit:539] Please be sure the target callbacks have been registered for this core
    4393.977812 s: VX_ZONE_ERROR:[ownNodeKernelInit:540] If the target callbacks have been registered, please ensure no errors are occurring within the create callback of this kernel
    4393.977823 s: VX_ZONE_ERROR:[ownGraphNodeKernelInit:583] kernel init for node 0, kernel com.ti.tidl ... failed !!!
    4393.977835 s: VX_ZONE_ERROR:[vxVerifyGraph:2055] Node kernel init failed
    4393.977843 s: VX_ZONE_ERROR:[vxVerifyGraph:2109] Graph verify failed
    TIDL_RT_OVX: ERROR: Verifying TIDL graph ... Failed !!!

  • 0
    TI__Prodigy 565 points

    Hi,

    For running python example on EVM the models need to be compiled first on PC(X86).

    If you do file of the tidl_model_import_onnx.so you will get to know it is X86 version.

    If you go through the readme on tidl_tools you will come across following (copying here for your easy reference : github.com/.../edgeai-tidl-tools

    • Run below script to validate all the python examples available in the repository. This script would run both model-compilation and inference when executed on PC
    ./scripts/run_python_examples.sh
    • This script would run only inference of example models when executed on target device like J7ES EVM or SK. So this script must be first executed on PC to generate the artifacts needed for inference and then copy below folders from PC to target device before running this script on device
    ./model-artifacts
./models


    Regards,
    Fadil