jon_ 22 Posted December 30, 2018 Share Posted December 30, 2018 Hi All I've seen quite a few threads where adding a GPU has been discussed for supporting multiple streams, but how fast are the GPU based solutions for encoding for offline sync - so one file at a time? Back story is that I've got an old Dell server that's got a Xeon X3440 @ 2.53GHz in it - passmark is 4540. It copes well for day to day stuff as there are only a few remote clients for my system, and nearly everything local is direct played, but when I'm off on a business trip and I queue up a whole bunch of stuff to get synced to my tablet it takes a long while to transcode everything. Most files convert at around 4-5x, some a bit slower. At the moment I've got the majority of the stuff I'll take with me pre-converting, but it chews up a lot of space and it's not practical to do this for the whole library. My current thinking is for a lower powered / energy efficient system + GPU (probably intel) - but it would obviously be dependant on if the GPU will do conversion of one file at a time spectacularly faster than 4-5x, or if I'm better off building a system with a bigger CPU(s).. Thanks! Link to comment Share on other sites More sharing options...
Luke 36879 Posted December 30, 2018 Share Posted December 30, 2018 Hi, there is a setting under sync settings related to converting at full speed. Can you check how this is set? It will affect how long sync transcoding will take. With the upcoming 3.6 release, sync transcoding will support all of the same GPU features are streaming transcoding. Link to comment Share on other sites More sharing options...
jon_ 22 Posted December 30, 2018 Author Share Posted December 30, 2018 Hi, there is a setting under sync settings related to converting at full speed. Can you check how this is set? It will affect how long sync transcoding will take. With the upcoming 3.6 release, sync transcoding will support all of the same GPU features are streaming transcoding. Thanks - it's set to allow full speed and max threads. It's slow because my CPU is old and slow The question is more if a modern GPU will cope with supporting (for example) 10x parallel transcodes, will it do a single sync transcode at 10x the speed or thereabouts - do the gains in parallelism translate to faster single conversions? Link to comment Share on other sites More sharing options...
Luke 36879 Posted December 30, 2018 Share Posted December 30, 2018 It depends. Many GPU's will limit the number of simultaneous encodes, like Nvidia for instance limits to 2. When this happens with Emby it will detect the failure and automatically switch to CPU, which will allow everything to continue to work. @@softworkz may have some insight about other vendors. Link to comment Share on other sites More sharing options...
bennymac 47 Posted December 30, 2018 Share Posted December 30, 2018 (edited) Cool i'd be interested in knowing more too. Does that mean with say a Quadro P2000 card in a server with an "unrestricted" amount of encodes/decodes as per https://developer.nvidia.com/video-encode-decode-gpu-support-matrix we can let the server go nuts? - based on the limit of the card of course. or would that only apply to a server that is currently not headless.. Edited December 30, 2018 by bennymac Link to comment Share on other sites More sharing options...
softworkz 3301 Posted December 31, 2018 Share Posted December 31, 2018 Cool i'd be interested in knowing more too. Does that mean with say a Quadro P2000 card in a server with an "unrestricted" amount of encodes/decodes as per https://developer.nvidia.com/video-encode-decode-gpu-support-matrix we can let the server go nuts? - based on the limit of the card of course. or would that only apply to a server that is currently not headless.. For nvidia it doesn't matter whether a monitor is connected or not. And for the previous question: yes, nuts. You can read some more here: https://github.com/MediaBrowser/wiki/wiki/Hardware-Acceleration-Overview Link to comment Share on other sites More sharing options...
softworkz 3301 Posted December 31, 2018 Share Posted December 31, 2018 It depends. Many GPU's will limit the number of simultaneous encodes, like Nvidia for instance limits to 2. 2 is the limit for consumer grade boards. Professional boards don't have this restriction. Detailed limits here: https://developer.nvidia.com/video-encode-decode-gpu-support-matrix Link to comment Share on other sites More sharing options...
jon_ 22 Posted December 31, 2018 Author Share Posted December 31, 2018 OK - to rephrase the question - on the latest beta, using hardware acceleration to convert a typical file for background sync (1080p mkv TV show) to something like the built in 4 meg built in TV or Mobile profile, what sort of conversion speeds to you see? (If you don't have that to hand, just a ballpark for conversion speeds would be useful ) Link to comment Share on other sites More sharing options...
bennymac 47 Posted December 31, 2018 Share Posted December 31, 2018 Thanks heaps mate. Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now