I have a server that I use for various background tasks I don’t want to bog my main rig with, one of which being plex transcoding. All my media is in H.265, so it uses a lot of horsepower to transcode down for streaming to anything except directly to my main pc. The system consists of dual Xeon 2670’s and 12 dedicated gigs of ram. Through my hypervisor I have set for the windows 10 vm running plex, up to 28 of the 32 cores are shared with the vm. However, there are some issues. If I completely stop and video transcoding that is running on a different vm, to convert all my remux blu rays to H.265, I am BARELY able to get 2 4k 10bit to 1080p 20mbps transcodes. I read plex wants a passmark of 17,000 for a single 4k transcode and I get 17,834 without anything else and about 13700 if I’m transcoding in the back, giving priority to plex.
I really don’t like buffering, especially since the whole purpose of my server was to spend dozens of hours troubleshooting, then have it work perfectly in the end. So having it buffer so easily isn’t fine for me. I looked around and saw the Quadro p2000 is a popular choice to boost the amount of streams many people can run considerably.I was wondering once and for all how much truth would there be to this, along with whether or not there is any quality degradation using hardware acceleration. I heard if you have a chip older than haswell you will be a quality drop, but I’m not sure how that works with Xeons, especially with my first gen 2670’s.
So, would it be worth grabbing a gpu to passthrough for hardware acceleration and would I see a quilty drop by adding one?