[Apache TVM Discuss] Vitis-AI Integration : Multi-thread c++ application hang

[venkataraju koppada via Apache TVM Discuss <no...@discuss.tvm.ai>]

Hi @jtuyls / @mak ,

We are trying to port Qt based ML application on zcu104 platform using and while running on board the application got hang with just empty freeze screen.


We have debugged it further and noticed that there is not issue QT libs and display part.

Without  DPU inference calls we were able to running QT multi-threaded application.

But when we add call run (inference) function then only we see application hang.

 

Following is the code snippet 
 

thread1(){

While(read cam feed){

Push frames into a shared queue;

}

}

thread2(){

in_img=read_input_queue()

Do the preprocess;

lock()

RUN() // FPGA call

Write output to another shared buff; 

unlock();

}

thread3(){

out_2thd =  shared_queue();

Do some preprocess;

lock()

RUN() // FPGA call

unlock()

Post process

}

 
Note: Same lock I have used for both threads .

Please have a look at and do you have any suggestion on this?
Like what might be the issue? Do you see any memory  or power limitations?

Thanks and Regards,

Raju





---
[Visit Topic](https://discuss.tvm.apache.org/t/vitis-ai-integration-multi-thread-c-application-hang/9295/1) to respond.

You are receiving this because you enabled mailing list mode.

To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/c16b968f3db39bfc973c89b1a7e0bfe11cd16a239847c3f9f77ca1fbd7409494).