[Plugin] Nvidia-Driver

[stormshaker]

Thanks… will try there.

[Agent531C]

On 8/17/2021 at 6:28 AM, ich777 said:

Please read the comment above yours from @alturismo too.

This would also make sense to me.

 

Now that its a few days later, Ive tried out the suggestions. I set the iommu to soft, and it didn't help. I can't disable AMD-V, as I run a vm that I work in, but I did disable the power option, and yet the gpu still drops off.

I wouldnt think the hardware is causing the problem, since it isnt that old, but it is a mini board. I don't know if they have limits on power draw or anything of that sort set up.

[ich777]

20 minutes ago, Agent531C said:

I wouldnt think the hardware is causing the problem, since it isnt that old, but it is a mini board. I don't know if they have limits on power draw or anything of that sort set up.

I don't think so but as always I think that can be a hardware compatibility issue of some kind...

Or anything related to the power management for PCIe devices on some AMD Motherboards.

 

Please also read this post from @SiRMarlon:

 

 

[SiRMarlon]

On 8/26/2021 at 8:57 AM, Agent531C said:

 

Now that its a few days later, Ive tried out the suggestions. I set the iommu to soft, and it didn't help. I can't disable AMD-V, as I run a vm that I work in, but I did disable the power option, and yet the gpu still drops off.

I wouldnt think the hardware is causing the problem, since it isnt that old, but it is a mini board. I don't know if they have limits on power draw or anything of that sort set up.

 

Disable your C-State I was having the same issues with my GPU dropping off and after I did this the issue went away. 

[frodr]

Hi,

 

The GPU will not do hardware transcoding in Plex anymore. Or Folding at Home. The GPU Dashboard shows switching 0% to 7% continuously with 0 processes running. And about 30-35 Watt usage. 

 

The mainboard is changed  lately, and I have seen processes run on the GPU (Plex, FaH). 

 

Any ideas?

 

// Frode 

 

 

 

 

 

 

maxx-diagnostics-20210902-1713.zip

[ich777]

50 minutes ago, frodr said:

Or Folding at Home. The GPU Dashboard shows switching 0% to 7% continuously with 0 processes running.

Are you sure that it isn't working, have you tried to issue the command 'nvidia-smi' from the command line yet and see what it tells you, please post a screenshot when transcoding something or using it in fold at home.

 

If it shows some utilization it definitely is doing something.

 

Have you tried to restart the server yet? From what motherboard did you switched from? Try to disable the C-States.

[frodr]

1 hour ago, ich777 said:

Are you sure that it isn't working, have you tried to issue the command 'nvidia-smi' from the command line yet and see what it tells you, please post a screenshot when transcoding something or using it in fold at home.

 

If it shows some utilization it definitely is doing something.

 

Have you tried to restart the server yet? From what motherboard did you switched from? Try to disable the C-States.

I am sure it isn't working. Disabling C-states did not change anything. 

 

Isn't it strange with Trotling at "Yes HW" when not in use?

 

Mainboard before was Asus WS C621E Sage.

 

// Frode 

 

Edited September 2, 2021 by frodr

[ich777]

10 minutes ago, frodr said:

I am sure it isn't working. Disabling C-states did not change anything. 

Can you open a container console window and try to issue the same command just to be sure everything is working inside the container?

 

What you can do too is that you open up a terminal from unRAID and type in 'watch nvidia-smi' this will call nvidia-smi every 2 seconds and then try to force transcode a video file and see if anything shows up in the line where it says "No running processes found"

[frodr]

23 minutes ago, ich777 said:

Can you open a container console window and try to issue the same command just to be sure everything is working inside the container?

 

What you can do too is that you open up a terminal from unRAID and type in 'watch nvidia-smi' this will call nvidia-smi every 2 seconds and then try to force transcode a video file and see if anything shows up in the line where it says "No running processes found"

Plex is working. Embarrassing. "Enable HW transcoding" was disabled in Plex server setup. I had an error with the cache, needed to reformat, did app restore. But did not check the Plex server setting.

 

Sorry for the trouble. 

 

It is a FaH problem I have, not a driver issue.

 

Thank you very much for the help. 

 

// Frode

[darkrom]

I can't figure out what I'm doing wrong but it seems like everything should work, but no luck.

 

Please see the album below this is driving me nuts. This should show every relevant setting. I have the same config in plex and neither plex or jellyfin is showing any GPU usage when it should be transcoding, but the CPU usage is still climbing as high as ever. I must be missing something and could really appreciate some help!

 

https://imgur.com/a/5xEfXdv

[ich777]

3 hours ago, darkrom said:

I can't figure out what I'm doing wrong but it seems like everything should work, but no luck.

Do you force a transcrode from within the web player or the application where you play the file?

What kind of media do you want to transcode?

For Plex you need a Plex pass to get transcoding to work.

What container are you using?

What does the overwiew section in Jellyfin tell you?

Don't forget that it is possible that the file that you transcode maybe also needs to be transcoded in terms of audio and that is producing also load on the CPU.

[alturismo]

3 hours ago, darkrom said:

but the CPU usage is still climbing as high as ever.

what does the dashboard say while playback, here from emby which should be pretty similar

 

 

here you see how the decode/encode is running, in my case vaapi (intel hw acc)

[darkrom]

8 hours ago, ich777 said:

Do you force a transcrode from within the web player or the application where you play the file?

What kind of media do you want to transcode?

For Plex you need a Plex pass to get transcoding to work.

What container are you using?

What does the overwiew section in Jellyfin tell you?

Don't forget that it is possible that the file that you transcode maybe also needs to be transcoded in terms of audio and that is producing also load on the CPU.

 

 

I am using binhex plex and binhex jellyfin.  With plex I do have plex pass, but at this point I've been mostly just trying to get it working in either.

 

I see this in jellyfin so I know its transcoding, purposely picking a 4k to 720p in this case.

 

 

 

 

@alturismo  this seems to be the jellyfin version of that screen in emby, but it does confirm it is transcoding.

[ich777]

1 minute ago, darkrom said:

I see this in jellyfin so I know its transcoding, purposely picking a 4k to 720p in this case.

Can you try only just for testing purposes with my or the linuxserver Jellyfin container?

 

I have no problem utilizing NVENC (Nvidia) or VAAPI (Intel) in my container.

[darkrom]

1 minute ago, ich777 said:

Can you try only just for testing purposes with my or the linuxserver Jellyfin container?

 

I have no problem utilizing NVENC (Nvidia) or VAAPI (Intel) in my container.

 

 

Absolutely I'll give yours a rip.  Thanks for the help lets see how it goes!

[ich777]

14 minutes ago, darkrom said:

Absolutely I'll give yours a rip.  Thanks for the help lets see how it goes!

Maybe I got time later today to send you a screenshot from my Jellyfin instance with NVENC running if you don't get it to work.

[darkrom]

25 minutes ago, ich777 said:

Maybe I got time later today to send you a screenshot from my Jellyfin instance with NVENC running if you don't get it to work.

Its working on your container!  Wow so simple and I never would have thought to try that thank you.  I guess I should swap from binhex to yours on both!

[ich777]

1 minute ago, darkrom said:

Its working on your container!  Wow so simple and I never would have thought to try that thank you.  I guess I should swap from binhex to yours on both!

Try to force an update from the binhex container, it should theoretically work on binhex's container without a problem from what I know.

[darkrom]

14 minutes ago, ich777 said:

Try to force an update from the binhex container, it should theoretically work on binhex's container without a problem from what I know.

No luck.  Maybe I'll reach out to binhex, I must have done something wrong but I don't see what.

[ich777]

27 minutes ago, darkrom said:

No luck.  Maybe I'll reach out to binhex, I must have done something wrong but I don't see what.

Don't know if he also uses Jellyfin but I can tell for sure that @binhex is using Plex woth a Quadro P2000 I think.

 

Maybe a reinstallation from the container does the trick. 

Go through the first few pages please from this thread, there where also people with the same issue and I think a reinstallation from the container did the job.

[binhex]

Just now, ich777 said:

Don't know if he also uses Jellyfin but I can tell for sure that @binhex is using Plex woth a Quadro P2000 I think.

i am indeed.

[squirrellydw]

What Branch is the most stable version?

[ich777]

5 hours ago, squirrellydw said:

What Branch is the most stable version?

That's a difficult question...

 

If you want the stable driver then stick with the Latest Production Branch.

 

If you want the latest and greatest stick with the New Feature Branch.

 

They should be both functional but the Production Branch is basically the stable branch.

[Rolucious]

Is there a way to download an older driver, that isn't listed in the options?

I want to use the 460.73 driver.

 

I'm using nsf miner for ethereum mining, while running a plex docker in the background.

The older driver is the only way I know how to get the overclocking of the gpu to work.

Since the nsfminer docker isn't supported anymore, I was hoping to stick to the older driver,while figuring out a more future proof solution.

[ich777]

3 hours ago, Rolucious said:

I want to use the 460.73 driver.

I think you are talking about 460.73.01

 

3 hours ago, Rolucious said:

Is there a way to download an older driver, that isn't listed in the options?

If you are talking about 460.73.01, you can but you have to be on unRAID version 6.9.2

 

The reason for this is that I only compile the latest current available Production and the New Feature Branch from the driver at the time a new unRAID version is released and all following drivers for this unRAID version.

After a new unRAID version is released this process starts from the beginning.

 

Also don't forget that older versions from the Nvidia driver may fail to compiles on newer Kernel versions.