Is there a way on a Linux server to tune the number of threads/cores/processors dedicated to transcoding? I have an older MacPro that is bogging down on CPU load but it is only hitting 150-225% core usage before playback stalls to catch up… I’m hoping to use more than 2.5 of the 8 cores available.
1 Like
The codecs are the limiting factor.
Audio = 1
Subtitles = 1
VC1 = 1 (it’s single threaded)
H.264 = 2-3
HW acceleration usage will not show in top.
Thanks. Am I able to use the old N460GTX sitting on the shelf here to accelerate that?
You need to verify what the card is capable of.
“Any old card” isn’t likely to have HEVC or 4K capability.
Check with Nvidia and their driver support.
There are so many graphics card variants available that it’s impossible for us to keep track of what works.
I think I might be missing something. If I’m using an NVIDIA card to help with transcode does plex-transcode just use it by default if the NVIDIA drivers are installed? I’m mostly just going h.264/5 is there a doc link that might explain this in detail so I can stop asking. 
If the nvidia card is installed, and the drivers are ver 418.30 or higher,
and PMS can successfully process one empty frame (a blank test frame of the target encoding) of video using the card , then it will use it first.
If not, then it looks to the iGPU (ASIC).
That failing, it falls back to the CPU.
If it works, it would be of limited help.
It appears the GTX460 supports decoding of H.264 video (I could not find an explicit listing of its decode capabilities). It does not support encoding video.
If you want to use hardware acceleration for H.265 video, you’ll want a GTX 1050, based on the Pascal architecture, or better.
See Nvidia’s Video Encode and Decode GPU Support Matrix for details on NVDEC capabilities of Nvidia video cards.
The Plex Media Server Hardware Transcoding Cheat Sheet at elpamsoft.com has additional useful information regarding the performance of various Nvidia cards.
The GTX 460 is based on Nvidia’s Fermi architecture (Wikipedia).
According to NVIDIA FFmpeg Transcoding Guide, it supports decoding (NVDEC), but not encoding (NVENC).
NVENC started with Kepler generation cards, the successor to Fermi.
Fermi cards are not listed in Nvidia’s Video Encode and Decode GPU Support Matrix. Kepler cards are listed. They support decoding of H.264. They do not support decoding of H.265. Given Fermi is a generation older, it would not support decoding of H.265.
NVIDIA FFmpeg Transcoding Guide
All NVIDIA GPUs starting with the Kepler generation support fully-accelerated hardware video encoding, and all GPUs starting with Fermi generation support fully-accelerated hardware video decoding. As of July 2019 Kepler, Maxwell, Pascal, Volta and Turing generation GPUs support hardware encoding, and Fermi, Kepler, Maxwell, Pascal, Volta and Turing generation GPUs support hardware decoding.
1 Like
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
