The TI E2E™ design support forums will undergo maintenance from Sept. 28 to Oct. 2. If you need design support during this time, contact your TI representative or open a new support request with our customer support center.

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: if there any impact on running custom model with multiple subgraphs in TDA4VM?

Part Number: SK-TDA4VM
Other Parts Discussed in Thread: TDA4VM

Hi team,

   My model has 4 outputs, so after model compiled using onnxrt_ep.py i got 4 subgraphs with lots of bin files in 8.02(https://github.com/TexasInstruments/edgeai-tidl-tools) like below:

 I would like to know if there any impact on running this model in TDA4VM instead of one graph? 

By the way, I set the following settings in 8.01:

then i got one graph with 2 bin files as below:

but it doesn't work in 8.02,so i also would like to know what's the requirement to generate one graph, and what's the requirement to generate multiple subgraphs.

Could you help check this case? Thanks.

Best Regards,

yuanfu

  • Hi,

    Sorry for the delay in response. It is expected to work with multiple subgraphs as well. Is the model you are using a TI model?

    Can you set "debug_level" : 1 for model compilation and share the log?

    Also, when you say, model doesn't work, is it on target EVM or PC emulation? And results in hang or poor accuracy?

    Regards,

    Anand