[TDA4EVM] How to debug TIDL model in TDA4

Hi,

We have import a caffe model by tidl_j7_01_00_00_00.

The DL model output result, which is processed by import tool, looks correct.

When I integrate the model to TDA4 demo app: app_tidl_avp2, the PC build also output correct results.

However, when I build j7 code and execute in TDA4EVM, it take couple minutes to process three images.

This DL model has been verified in TDA2 and can process less than 100 ms.

What I want to know is: "How can I debug this issue ?".

Thanks,

Charles