Hi Team,
For a product we are using the imx6 quad to capture the video frames using the CSI, convert it to RGBA format using the gpu and also render some overlay frames on top of the camera video by using the GPU again and then convert it to 420 format by IPU and feed it to VPU ( h264) for getting encoded to stream over network.
In this scenario, if we use 1280x1024@30fps, all seems working well, but if we increased the fps to 50, then vpu takes much more time to encode( it was taking 25 ms for 30fps case, now it takes around 50ms).
But if we reduce the GPU load by avoiding overlay rendering , then at the same 50fps, VPU takes only around 25ms itself.
That means, VPU peroformance is affected by the GPU load?
Is there any way to avoid this dependency between VPU and GPU performance?
Hello abeeschmp,
The only solution is to try to reduce the memory with VPUmem or GPUmem to see if let you get the fps that you need.
Regards
Hello Bio_TICFSL,
Thanks for the reply. Could you please explain more on that ? You mean to reduce the resolution to get higher fps or something other?