TDA4VM: Issues with GPU RTI watchdog in SDK8.6

Hello,

I'm not aware of the GPU watchdog. Where did you get this information?

The GPU driver handles all of the communication with the GPU, so there is no need to worry about handling GPU internal modules. If you could elaborate on what your goal is I can guide you to the correct instructions.

Thanks,

Erick

HI,

Other core RTI watchdog on their own core and feed the dog, but GPU did not find the relevant information, how to open and feed the dog on the GPU?

Thanks

Hello,

I'm checking with our team, but I believe these are dedicated per core for safety diagnostic purposes. I'll need more information from them on how this is supposed to be used.

Thanks,

Erick

Hello,

Deeply expect your reply!

Thanks.

Hi,

I really appreciate your help! 

but there is still a question, if I want to deploy SDL_RTI_service on the GPU, where should I integrate it? In SDK8.6 I can't find the code running on the GPU.

Thanks.

Hi,

The watchdog needs to be serviced, in relation to the GPU functionality.  This can be done by S/W running on the A72 core, where the A72 S/W would "pet"/service the watchdog as result of an indication (callback, interrupt) from GPU.   

For example if you have an application that is expecting 30 frames per second from the GPU, then upon receipt of each 'n' frames you could service the watchdog, which would prevent it from expiring.   The DWWD window would need to be configured according to how many frame drops, or how many 'ms' the system is willing to wait before watchdog expires.

Regards,

kb

Hi,

Thank you very much for your answer. I have roughly understood the whole deployment process.
But some of the details are still unknown,
How to get frames from GPU, is there an API?
Or is there an example of deploying RTI15?

Thanks。