TDA2: SRV using GPU

Hi,

Please find your answer below

(1) Transfer 4 test images from PC to EVM...I do not know how I can do this yet

Please use network_rx_tx link to receive or send data over network

(2) Run OpenGL-based SRV algorithm on the 4 images...I am assuming OpenGL code can only execute on the TDA2x GPU

yes

(3) Transfer output SRV image from EVM to PC...again, not sure how this is to be done

same as que (1)

Question 1 - Can you point me to some relevant and useful references to implement this scenario?

Please go through the below usecase

vision_sdk\apps\src\hlos\adas\src\usecases\lvds_vip_multi_cam_3d_srv

vision_sdk\apps\src\rtos\usecases\network_rx_tx

Question 2 - In your opinion, how many CPU cores from the EVM will be required at a minimum to implement this scenario?

Please check the usecase jpg file to have idea which all cores are required

Question 3 - What is "host CPU" in the context of TDA2x EVM? Does it refer to A15 core 0?

Yes, as you have to run vision sdk linux.

Question 4 - How is the GPU programmed on TDA2x? Is it possible to access it through SYS/BIOS?

Please go through the below module to have some idea about sgxmodule which uses GPU.

vision_sdk\apps\src\hlos\modules\sgx3Dsrv

Question 5 - What should be my development environment for this implementation? Windows or Linux?

On your pc side you can have either of windows or linux as VSDK supports both OS.

Regards,

Anuj