I’m trying to figure out why my transcoding is chewing up my CPU, even with one 1080p stream.
Movie is stored on a NAS, I have a dedicated mobo,10500 processor & nvme drive just to run plex server and transcode.
I have plex pass, hardware acceleration & accelerated encoding is enabled.
Using gpuz I can see that the integrated gpu is idling and my cpu is at 100%. As far as I can tell this CPU should do well on hardware encoding settings, although I can’t figure out what I’m doing wrong.
Any help is appreciated, thanks!
Server Version#: 1.32.5.7349
Player Version#: chrome browser
Something very weird happened, the problem went away briefly - I even saw the gpu doing the work. After this single time working, it went back to the initial problem. From working to not working this all happened around the same time in the logs: 10pm - 10:05pm
Hardware transcoding for both Nvidia and Intel have been broken since the update in July when they added the option to choose the hardware and added AMD. If you go back and install 1.32.4.7195 it will work fine.
It would be nice if this was fixed but it has been crickets from the developers.
I specced this server with a cpu that had intel quick sync specifically because it was recommended here for transcoding. Will a dedicated GPU work better?
You’ll want a card that can decode HEVC 10-bit 4:2:0 video.
Most GTX cards are limited to five simultaneous encodes by Nvidia (Some have no encode capability).
Most RTX/Quadro cards have no pre-set limit (some are limited to 0 or 3 encodes).
The limit is shown in the ENCODE section, in the “Max # of concurrent sessions” column (the limit is on encodes, decodes have no pre-set limit).
Elpamsoft has transcoding performance info for Nvidia GPUs.
A 1050Ti is the bare minimum. Many people use P2000s. See the “Streams for VMRAM” section on the Elpamsoft page. Basically, the more VRAM the better and avoid 2GB cards.
@FordGuy61 much appreciated. You pointed me in the right direction and after looking into all this I ended up first transcoding with a Nvidia gpu, and then moving to linux so I could use the intel gpu. Everything is working much better now, thanks.