Part Number:TDA2
Hi,
I saw in the data sheet (eve_sw_datasheet.pdf) that the performances for TI provided kernels and apps is given in cycles per pixel. For kernel, this is based on the block size and for apps it is based on the image size.
Can you please share with us how to calculate these cycles per pixel for both apps and kernels?
Do we use EVE core utilization to calculate this?
What is the operating frequency of EVE?
Will there be a difference in the performance metric (cycles per pixel, local link latency etc), if the application is run through VisionSDK and when run standalone?
Regards,
Surbhi