Jump to content
linuxserver.io

[Plugin] Linuxserver.io - Unraid Nvidia

1954 posts in this topic Last Reply

Recommended Posts

5 minutes ago, rmeaux said:

Use one like this or whatever flavor of port you have. 

Thanks for the reply - but I don't think that's what he was referring to.  There's a "stub" command you can set for specific devices which seems to make the host OS ignore them? I've seen it mentioned a few times in the forum but no clear process on how to do it. Like what file is edited, etc...

 

Share this post


Link to post
34 minutes ago, mbezzo said:

Thanks for the reply - but I don't think that's what he was referring to.  There's a "stub" command you can set for specific devices which seems to make the host OS ignore them? I've seen it mentioned a few times in the forum but no clear process on how to do it. Like what file is edited, etc...

 

Haha, I'll learn something too. I may be doing this next week. I found this: 

 

Step 1. Go to tools -> system devices and look for the ID of what you want to stub

 

Step 2. Copy the number that looks like this:

 

8086t:8d31

 

Step 3. Go back to "Main" and click on "Flash"

 

Step 4. Add this to your syslinux config 

pci-stub.ids=8086:8d31

(replace "8086:8d31" with the number from Step 2.)

 

 

Let me know if this is helps. I already have a P2000. Getting a free 1050ti next week for gaming. May be in your shoes. 

Share this post


Link to post
16 hours ago, rmeaux said:

/tmp is using the Ram as a temporary transcoding location.

What do you have in your plex settings --> transcoder --> "Transcoder temporary directory" blank?

Yes. this is correct. In plex, the transcoder directory is blank, but in the docker setting it is set to /tmp (for the mapping)

Share this post


Link to post
4 hours ago, drpete said:

Yes. this is correct. In plex, the transcoder directory is blank, but in the docker setting it is set to /tmp (for the mapping)

Docker setting is just a suggestion. The gui is where you set it now. If it is unset, it will default to a location under /config

Share this post


Link to post
Haha, I'll learn something too. I may be doing this next week. I found this: 
 
Step 1. Go to tools -> system devices and look for the ID of what you want to stub
 
Step 2. Copy the number that looks like this:
 
8086t:8d31
 
Step 3. Go back to "Main" and click on "Flash"
 
Step 4. Add this to your syslinux config 
pci-stub.ids=8086:8d31

(replace "8086:8d31" with the number from Step 2.)
 
 
Let me know if this is helps. I already have a P2000. Getting a free 1050ti next week for gaming. May be in your shoes. 



Now this is what I was looking for! I’ll be trying it this weekend and will report back!

Thanks!


Sent from my iPhone using Tapatalk

Share this post


Link to post
Posted (edited)

Just an update on the memory issue with plex. I uninstalled the nvidia plug in and removed the parameters that were added to run the gpu (quadro p2200). I have now gone 27 hours with no crash or plex shutdown due to memory being filled. 

 

I am gonna run a few more days and make sure no more issues. The weird part is if it is stable for a few more days then that rules out plex or unraid issues. But leaves it down to drivers and/or gpu card. When I installed the card software, it was working and I saw the transcode happen when i "watch nvidia-smi". Is it possible I did something wrong typing the parameters but still allowed it to work but not properly? Could this be because I bought the P2200 and not the P2000? Any thoughts would be appreciated. As always thanks for the help.

 

ps. just wondered. My motherboard is supermicro x10sri-f. Do I need to do any bios changes (or maybe update)? Just brainstorming now for anything.

Edited by drpete
thougt of something else

Share this post


Link to post
51 minutes ago, drpete said:

Just an update on the memory issue with plex. I uninstalled the nvidia plug in and removed the parameters that were added to run the gpu (quadro p2200). I have now gone 27 hours with no crash or plex shutdown due to memory being filled. 

 

I am gonna run a few more days and make sure no more issues. The weird part is if it is stable for a few more days then that rules out plex or unraid issues. But leaves it down to drivers and/or gpu card. When I installed the card software, it was working and I saw the transcode happen when i "watch nvidia-smi". Is it possible I did something wrong typing the parameters but still allowed it to work but not properly? Could this be because I bought the P2200 and not the P2000? Any thoughts would be appreciated. As always thanks for the help.

 

ps. just wondered. My motherboard is supermicro x10sri-f. Do I need to do any bios changes (or maybe update)? Just brainstorming now for anything.

FWIW I'm running a P2200 for Plex without issue...

Share this post


Link to post
5 minutes ago, mbezzo said:

FWIW I'm running a P2200 for Plex without issue...

 

thanks..that makes me feel better...Gonna wait another day and retry the install. I read in the early posts here that copy and past of the parameters sometimes puts a blank space. Will check for this. i would imagine if I entered something wrong it wouldnt work, but may it would but cause my errors.

 

Curious if you are using encoding and decoding or just decoding? Also are you on 6.8? Thanks again

Share this post


Link to post
5 minutes ago, drpete said:

 

thanks..that makes me feel better...Gonna wait another day and retry the install. I read in the early posts here that copy and past of the parameters sometimes puts a blank space. Will check for this. i would imagine if I entered something wrong it wouldnt work, but may it would but cause my errors.

 

Curious if you are using encoding and decoding or just decoding? Also are you on 6.8? Thanks again

This kinda stuff usually works or it doesn't - so I think yer clearly on the works side of things :)

 

I'm doing both enc/dec and on 6.8!

Share this post


Link to post

Well i made it 37 hours and had the memory crash :( The hunt for the fix continues.

 

Share this post


Link to post
45 minutes ago, drpete said:

The hunt for the fix continues.

are you tried long memtest for your RAM?

Share this post


Link to post

 

On 1/1/2020 at 8:49 PM, rmeaux said:

Haha, I'll learn something too. I may be doing this next week. I found this: 

 

Step 1. Go to tools -> system devices and look for the ID of what you want to stub

 

Step 2. Copy the number that looks like this:

 

8086t:8d31

 

Step 3. Go back to "Main" and click on "Flash"

 

Step 4. Add this to your syslinux config 


pci-stub.ids=8086:8d31

(replace "8086:8d31" with the number from Step 2.)

 

 

Let me know if this is helps. I already have a P2000. Getting a free 1050ti next week for gaming. May be in your shoes. 

Alright, found the device IDs, added them (both the video and audio portions) to syslinux file (added it to the bottom of the global config part) rebooted the server and still can't get a VM to boot with the card passed through... Just not sure where to go from here!

Share this post


Link to post
14 hours ago, uldise said:

are you tried long memtest for your RAM?

Gonna try this soon. I doubt its this becuase I see no errors with any other dockers or vm's. But i thought this also. Re-installed plex completely fresh install and chose a different version (used lnio but chose public instead of latest) and did clean install of nvidia plugin and re entered all gpu info checking for no spaces. Will run this for a week or so and see. If I still get error then Im gonna try a memtest. Thanks for suggestion though, i think it is a great idea. Will keep everyone informed.

Share this post


Link to post
6 hours ago, Tylercollins590 said:

Thanks for this great plugin.

 

Is there anyway to downgrade CUDA to 10.0?

No, much like the drivers (as explained in the first post.) We compile this with whatever the latest versions are available.

Share this post


Link to post

drpete I have a Quadro P2200 installed and plex has not crashed.  That being said I'm not sure if the encoding/decoding is working correctly for me.  Perhaps someone can enlighten me on my setup.

 

I followed spaceinvaader's video,  no issues there.  Using the latest unraid/nvidia build 6.8 and lsio plex docker.  When I run a sample video transcode test, my 4 core E3-1245 v5 xeon cpu maxes out on load. 

 

This is what's shown on the plex dashboard:

Video- 4K (HEVC Main 10 HDR)
4K (H264)—Transcode

AudioEnglish (TRUEHD 7.1)
EAC3—Transcode

 

- when running the command "watch nvidia-smi" the empty processes box gets populated

Unless I understand it wrong, since it says both video and audio are being transcoded, I should see the hw beside them in plex to show it's being done by the hardware?

 

- when running the command "nvidia-smi dmon -s u" all 5 columns are zero's

Shouldn't these columns have numbers in them? (gpu, sm, mem, enc, dec)

 

So I decided to force plex to decode by using this information as the Plex Transcode

#!/bin/sh
/usr/lib/plexmediaserver/Plex\ Transcoder2 -hwaccel nvdec "$@"

 

Now when I rerun the sample, the process gets populated (same as before) but now for the 5 columns, these 3 headings have numbers in them (sm, mem, dec)

 

Only thing I can think of is that my cpu is not good enough, was hoping that the video card would have helped since I heard of people running like 15 streams with a dedicated video card and I can't even do 1 stream.

 

 

Share this post


Link to post
16 minutes ago, Mogo said:

drpete I have a Quadro P2200 installed and plex has not crashed.  That being said I'm not sure if the encoding/decoding is working correctly for me.  Perhaps someone can enlighten me on my setup.

 

I followed spaceinvaader's video,  no issues there.  Using the latest unraid/nvidia build 6.8 and lsio plex docker.  When I run a sample video transcode test, my 4 core E3-1245 v5 xeon cpu maxes out on load. 

 

This is what's shown on the plex dashboard:

Video- 4K (HEVC Main 10 HDR)
4K (H264)—Transcode

AudioEnglish (TRUEHD 7.1)
EAC3—Transcode

 

- when running the command "watch nvidia-smi" the empty processes box gets populated

Unless I understand it wrong, since it says both video and audio are being transcoded, I should see the hw beside them in plex to show it's being done by the hardware?

 

- when running the command "nvidia-smi dmon -s u" all 5 columns are zero's

Shouldn't these columns have numbers in them? (gpu, sm, mem, enc, dec)

 

So I decided to force plex to decode by using this information as the Plex Transcode

#!/bin/sh
/usr/lib/plexmediaserver/Plex\ Transcoder2 -hwaccel nvdec "$@"

 

Now when I rerun the sample, the process gets populated (same as before) but now for the 5 columns, these 3 headings have numbers in them (sm, mem, dec)

 

Only thing I can think of is that my cpu is not good enough, was hoping that the video card would have helped since I heard of people running like 15 streams with a dedicated video card and I can't even do 1 stream.

 

 

 

Hi, did you made sure to "check" the settings inside PLEX settings itself?  Especially, in the "transcoder" section :

 

 

"Use hardware acceleration if available"

"Use hardware-accelerated video encoding"

 

If you already check them, it is suppose to use it.

Share this post


Link to post
13 minutes ago, Pducharme said:

 

Hi, did you made sure to "check" the settings inside PLEX settings itself?  Especially, in the "transcoder" section :

 

 

"Use hardware acceleration if available"

"Use hardware-accelerated video encoding"

 

If you already check them, it is suppose to use it.

In Plex I have the following checked "Use hardware acceleration when available"

Share this post


Link to post
13 hours ago, Mogo said:

In Plex I have the following checked "Use hardware acceleration when available"

I suspect it'll be the audio that's getting transcoded via the CPU.

Share this post


Link to post

with "nvidia-smi dmon -s u" you can see more details what the gpu is doing (decode/encode)

Share this post


Link to post
Posted (edited)

Occasionally my GPU seemingly 'disappears' from Unraid/Docker. Running nvidia-smi produces:

 

image.png.1220b01b2cef5e8beabb117ec223e7fb.png

 

but if I check Settings/Unraid Nvidia the device is still present:

image.png.5523f096a512592b9dd67138a7e1ef83.png

 

The server log is filled with some nvidia driver errors:

image.png.98a07ed0359fa1e79c1601fd2e78c119.png

A server reboot usually fixes the issue but it's rather annoying. How would I go about debugging this?

 

Running 6.8.0

 

EDIT: Just after I posted this nvidia-smi showed the card correctly and the docker containers that use the GPU were working correctly, I did not reboot. It basically just showed back up again.

Edited by PTRFRLL
Add additional info

Share this post


Link to post

After installing the plugin, I was unable to locate it under settings. It is showing as installed under plugins. 

 

Running 6.8 stable

 

Edit: After a reboot, the plugin is now showing under settings. 

Share this post


Link to post
Occasionally my GPU seemingly 'disappears' from Unraid/Docker. Running nvidia-smi produces:
 
image.png.1220b01b2cef5e8beabb117ec223e7fb.png
 
but if I check Settings/Unraid Nvidia the device is still present:
image.png.5523f096a512592b9dd67138a7e1ef83.png
 
The server log is filled with some nvidia driver errors:
image.png.98a07ed0359fa1e79c1601fd2e78c119.png
A server reboot usually fixes the issue but it's rather annoying. How would I go about debugging this?
 
Running 6.8.0
 
EDIT: Just after I posted this nvidia-smi showed the card correctly and the docker containers that use the GPU were working correctly, I did not reboot. It basically just showed back up again.
The settings page just reads a record of what cards were present at boot, it isn't dynamically loaded at each page load.

Your card is dropping out on occasion. The cause could be anything.

Try a different slot, BIOS up to date.

Different Unraid version and/or associated Nvidia version, but ultimately there's not much I can do to help unless you conclusively identify a driver issue and even then you're probably going to have to wait until a new upstream driver is released.

Sent from my Mi A1 using Tapatalk

Share this post


Link to post
21 minutes ago, CHBMB said:

The settings page just reads a record of what cards were present at boot, it isn't dynamically loaded at each page load.

Your card is dropping out on occasion. The cause could be anything.

Try a different slot, BIOS up to date.

Different Unraid version and/or associated Nvidia version, but ultimately there's not much I can do to help unless you conclusively identify a driver issue and even then you're probably going to have to wait until a new upstream driver is released.

Sent from my Mi A1 using Tapatalk
 

Ok, I'll assume it's hardware related and not Unraid or this plugin until I rule that out. Thanks for the suggestions.

Share this post


Link to post

Hello,

 

I recently set this up with a Plex pass and my CPU breathed a sigh of relief as my GPU took over transcoding (thanks for making this!)

 

My question is, previous to do this, I had transcode going to /tmp (ie RAM).  Should I turn this off or leave it be?

 

Thanks!

Share this post


Link to post

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.