January 3, 20197 yr 3 hours ago, Output said: This is really cool, is anyone running this 24/7 without issues ? It is cool and no.
January 3, 20197 yr On 10/22/2018 at 5:11 PM, bnevets27 said: What about AMD then? As I've seen reports of users running an AMD gpu under Linux with hardware transcoding working. I don't really think we are waiting on Plex. We need drivers. Even if plex officially supported Nvidia and AMD it won't matter if unraid doesn't have the drivers. It's the same as unraid dvb. Plex supports tuners but plex can't see or do anything with those tuners without the drives being supplied by the unraid dvb plugin. The issue is rather simple, if the right drivers are installed into unraid then the gpu can be passed to the plex docker. The support of that gpu then falls on Plex. The problem is adding drivers for gpus into unraid is not an easy or simple task. I highly doubt Lime-tech wants to spend the time/money/effort into doing so. Even though I think it may attract more customers. Therefore it would fall on the community. For an idea of how much work is involved, listen to CHBMB. He knows much work is involved with compiling drivers into each new build of unraid. I really want this to happen. But from what I've read and been told is not something thats going to come easy. This. I too would love this. I too recognize just how hard it is to compile drivers in a build. Because of the difficulty I was having with Plex, I moved my Plex server to an separate machine that I happened to have to get iGPU transcoding. No more buffering since I moved it.
January 6, 20197 yr With Plex only supporting encoding with Nvidia and not decoding what does this mean for end users? or does the rest of my system just pick up the decoding slack? Also fair play on the work on this to date. I just got my hands on a cheap p4000 and im hoping not to have to create a windows VM just to use it.
January 10, 20197 yr On 1/6/2019 at 9:25 AM, tazire said: With Plex only supporting encoding with Nvidia and not decoding what does this mean for end users? or does the rest of my system just pick up the decoding slack? Also fair play on the work on this to date. I just got my hands on a cheap p4000 and im hoping not to have to create a windows VM just to use it. What can be encoded by hw is being by hw, what can be decoded by hw gets decoded by hw. Either by no capacity left of by drivers/support being absent. So if by whatever reason you can't decode in hardware, cpu will do it. If your GPU gets maxed, cpu will pickup what he can. Still a huge gain as encoding is a harder work than decoding. Decoding HEVC being the big decoding gain that I would love to do.
January 17, 20197 yr Did someone get a proof of concept to work? Holy hell in a handbasket Batman! I'll pay money, literally, I will pay money to have this. If LT were to add it themselves, I would literally pay for it as a separate addon. I don't think I am alone in this either, I hope this is something the smarter people out there can get together and solve. I just ran into stupid transcoding BS with PLEX again so I got to thinking about it again.
January 18, 20197 yr Yeah I've been reading up more on the hardware transcoding. Emby/jellyfin also supports hardware transcoding. They actually support encode and decode. Plex only supports encoding on Linux at this time.If the Nvidia patch can be included with the driver. It will be a whole new world.The holy grail of being able to use a cheap gpu in a low end/power cpu system and be able to transcode without even trying.Heading out to grab a gpu now and I think I might try plex or jellyfin in a Linux VM just to see how much greener the grass really is.
January 26, 20197 yr 54 minutes ago, deaerator said: It would be nice if it was a plug in that you can update. Indeed it would!
January 27, 20197 yr @CHBMB I See there is a commit not long ago, no issues open and no comment on being any issues still, is it Working ? https://github.com/CHBMB/Unraid-NVIDIA-Plugin Just to make sure, what url should we put in the Plugin field ? Edited January 27, 20197 yr by Altheran
January 27, 20197 yr 2 minutes ago, Altheran said: @CHBMB I See there is a commit not long ago, no issues open and no comment on being any issues still, is it Working ? https://github.com/CHBMB/Unraid-NVIDIA-Plugin Just to make sure, what url should we put in the Plugin field ? Oh yeah, we released it and forgot to tell anyone, and if you hadn't reminded me then I would have completely forgot. Can put in whatever URL you want into the plugin field. It won't get you anywhere......
January 28, 20197 yr We haven't released anything, so stop installing things that aren't finished.Sent from my Mi A1 using Tapatalk
January 28, 20197 yr 5 hours ago, huntastikus said: My apologies, I misread your comment, I didn't detect the sarcasm. Uninstalling now. Impatiently waiting for this to be ready though! You guys 🤘!!! If you had actually skimmed through this thread before attempting to install, you might have saved yourself some trouble. Start reading here, click on the little arrow in the upper right of this image. It's only one page back.
January 28, 20197 yr If you had actually skimmed through this thread before attempting to install, you might have saved yourself some trouble. Start reading here, click on the little arrow in the upper right of this image. It's only one page back.Actually that won't work. We tried that and we just saw a lot of segfaults on our machines. This has never been as simple as people think it will be about getting drivers installed on the host. I managed that 5 months ago.....Sent from my Mi A1 using Tapatalk
January 28, 20197 yr 29 minutes ago, CHBMB said: Actually that won't work. We tried that and we just saw a lot of segfaults on our machines. This has never been as simple as people think it will be about getting drivers installed on the host. I managed that 5 months ago..... Sent from my Mi A1 using Tapatalk Shoot, I meant to point to the post that started the CF the LAST time somebody started using the unreleased project and asking for support. I was trying to give some background. If this fresh batch of folks had actually read back a page or two, all their questions would have been answered instead of posting the same stuff over again.
January 28, 20197 yr Shoot, I meant to point to the post that started the CF the LAST time somebody started using the unreleased project and asking for support. I was trying to give some background. If this fresh batch of folks had actually read back a page or two, all their questions would have been answered instead of posting the same stuff over again.YupSent from my Mi A1 using Tapatalk
January 28, 20197 yr 4 minutes ago, jonathanm said: Shoot, I meant to point to the post that started the CF the LAST time somebody started using the unreleased project and asking for support. I was trying to give some background. If this fresh batch of folks had actually read back a page or two, all their questions would have been answered instead of posting the same stuff over again. I did install the project but never asked for support, as you saw while you did your skimming. The project never mentioned, released, unreleased, not working, beta, alpha, not working until I brought it up here. CHBMB did clear it up later.
January 30, 20197 yr I do have NVENC accelerated transcoding running reliably (from what I can tell) on my own system using an edited Jellyfin docker, a custom FFMPEG for it and the unraid scripts in my github. The only thing I need really is a less buggy ffmpeg build - that said if you supply your own FFMPEG compiled with NVENC support the way I've done it seems stable and I didn't see any crashes. I edited the dockerfile for jellyfin to add the appropriate drivers (to the docker) and used the kernel scripts I edited from yours to add the drivers to Unraid itself (the host). @CHBMB not sure if this will help you at all with your development? Edited January 30, 20197 yr by aterfax
January 30, 20197 yr This is coming from pure ignorance, but could you get by with just the docker container having the drivers? Then just pass the hardware straight in? That way you don't have to fight trying to get drivers working in Slackware.
January 30, 20197 yr This is coming from pure ignorance, but could you get by with just the docker container having the drivers? Then just pass the hardware straight in? That way you don't have to fight trying to get drivers working in Slackware.Nope, that doesn't work. Plus it means you would have to bake Nvidia drivers into each container. We're testing a method, we're getting it ready to be released as a beta. Patience. Sent from my Mi A1 using Tapatalk
January 30, 20197 yr I haven't done much with sharing hardware in docker, so I'll take your word for it. I'm guessing it's because it needs it working on the host for it to work in the container? Would you need drivers in both then?
January 30, 20197 yr 22 minutes ago, m0ngr31 said: I haven't done much with sharing hardware in docker, so I'll take your word for it. I'm guessing it's because it needs it working on the host for it to work in the container? Would you need drivers in both then? Some devices don't need a driver, others do. So their is not a clear path. Our method doesn't need the driver installed in the container.
January 30, 20197 yr Okay, that's interesting. So is the solution is to inject the nvidia drivers into the bzImage or something?
January 31, 20197 yr 12 hours ago, m0ngr31 said: Okay, that's interesting. So is the solution is to inject the nvidia drivers into the bzImage or something? Yes, the drivers are installed in the bzroot file.
Archived
This topic is now archived and is closed to further replies.