Consider following scenario, I'm decoding on GPU using HW implementation and I've exhausted all GPU resources. Now I would like to switch to CPU decoding. How do I achieve it, what kind of device I have to create? Do I have to use system memory in case of SW implementation?
EDIT001: Ok, looks like I know how to force decoding to work with CPU only, however I cant get the VPP working with SW implementation. Tried the sample_vpp and sample_decode examples, both fail with -3 when the MFXVideoVPP's Init is called. Then I found this https://software.intel.com/en-us/forums/intel-media-sdk/topic/639188 , where the following stated "Currently, VPP composition is not supported or possible in SW mode.". The question is what else does not work in SW mode?
Sorry for the long delay.
The support of SW & HW modes change frequently according to MSDK version.
The best way to find out is to read the documentation with the SDK- release notes will update you on new feature available, sample notes will let you know about restrictions.