[Plugin] Nvidia-Driver

[crocker5731]

7 minutes ago, ich777 said:

Sorry but you came here that the GPU is not working and now you post a screenshot from the template.

 

You have to give me more information what you want, do you want to get the template going with the GPU or without the GPU?

Do you have a GPU in your system?

No I do not have a GPU in my pc. I'm using on bored video. So I do need get it to work with a GPU.

[ich777]

9 minutes ago, crocker5731 said:

No I do not have a GPU in my pc.

Then why do you install the Nvidia driver?

 

9 minutes ago, crocker5731 said:

I'm using on bored video. So I do need get it to work with a GPU.

Then pass through the integrated GPU.

Please don't ask how to do that here since this is the Nvidia Driver support thread not the container support thread or how to get a Intel iGPU working with a Docker container.

 

I don't even know this container so I can't help.

 

As said above you have to remove --runtime=nvidia from the Extra Parameters (with enabled Advanced View in the Docker template).

[crocker5731]

3 minutes ago, ich777 said:

Then why do you install the Nvidia driver?

 

Then pass through the integrated GPU.

Please don't ask how to do that here since this is the Nvidia Driver support thread not the container support thread or how to get a Intel iGPU working with a Docker container.

 

I don't even know this container so I can't help.

 

As said above you have to remove --runtime=nvidia from the Extra Parameters (with enabled Advanced View in the Docker template).

I got it working for now. Thank you for all your help!!!

[tuxflux1]

I'm having trouble updating the driver. Even though I have version 560.31.02 installed, it seems to want to update to the version in the screenshot. And it's not downloading either, just gives the error message. What is happening here?

 

[ich777]

31 minutes ago, tuxflux1 said:

What is happening here?

Please post a screenshot from the entire Nvidia Driver plugin page.

[tuxflux1]

[tuxflux1]

Nevermind. It must have updated itself on its own. It wasn't on the most recent version last time I checked, but it seems to have updated itself in the meantime without me noticing. Sorry for the inconvenience.

[ich777]

1 hour ago, tuxflux1 said:

Nevermind. It must have updated itself on its own. It wasn't on the most recent version last time I checked, but it seems to have updated itself in the meantime without me noticing. Sorry for the inconvenience.

I have unhidden your post again since your plugin is broken, this happens for some users and this is the indication for it:

 

Please uninstall the plugin, reboot, reinstall the plugin and reboot again.

 

EDIT: It should be also possible to just uninstall, install and reboot after thinking again about it.

[clincher]

Got a strange issue and I'm at my wits end.  Bought a Supermicro X11SPI-TF motherboard to upgrade my current rig. (256GB ECC DDR4 / 6132 Xeon Gold CPU / Nvidia RTX2060). I had originally bought a Supermicro X11 dual-proc board but returned it because I was having similar issues, but now I think it's related to hardware compatability, but I want to see if I'm missing something.  Right now, I'm bench testing the motherboard with similar components to my current rig and using Unraid trial USB to test. 

 

Issue is this: I can get UnRAID to boot fine with 2060 installed in any slot (typically Slot 6).  When I install the Nvidia plugin, the machine hangs, then crashes and reboots.  I've tried a number of different Nvidia driver versions including the open source version (with correct modprobe settings).  Everything results in the same behavior.  I have 4G Encoding enabled.  I've toyed with all sorts of PCIe settings in the BIOS and still this is the behavior every time - hang and then reboot.  I get nothing on the terminal or syslogs relating to an error.  It just shows the plugin being installed and then hang/crash.  

 

I've also tried testing with a GTX 1080 and GTX 650 and both exhibit the same issue.   Just seems like there's some anger with Supermicro and GPUs.  Disabling the onboard video of the motherboard has no effect either.  no change in behavior.

 

If I install the plugin without the GPU installed, it will install fine.  I can re-install the GPU and boot to Unraid fine.  Although, when you navigate to the Unraid plugin settings or even run diagnostics, it will hang and then crash.  When I tried to run diag, it crashes here:

 

ls -lA /sys/class/drm/*/device/driver 2>/dev/null|todos >>'/tower-diagnostics-20240817-1519/system/drm.txt'
/usr/bin/nvidia-smi --query 2>/dev/null|todos >>'/tower-diagnostics-20240817-1519/system/nvidia-smi.txt'

 

I'll attach my diags with and without the GPU and plugin installed.  Just wondering if anyone has any experience with Supermicro boards and perhaps there's a setting for the Nvidia driver that I can set that may help.  I may just end up moving on to a different mobo that isn't so tempermental.  

 

I've run memtest+ on the memory - no issues.  

 

Grasping at straws.  Thanks!

 

 

tower-diagnostics-20240817-1537 with gpu no plugin.zip tower-diagnostics-20240817-1511 sans GPU with plugin.zip

Edited August 17 by clincher

[alturismo]

5 hours ago, clincher said:

Grasping at straws.  Thanks!

 

tower-diagnostics-20240817-1537 with gpu no plugin

>> GPU is visible, no plugin, no errors

 

64:0d.3 System peripheral [0880]: Intel Corporation Sky Lake-E LMDP Channel 2 [8086:204b] (rev 04)
	Subsystem: Intel Corporation Sky Lake-E LMDP Channel 2 [8086:0000]
65:00.0 VGA compatible controller [0300]: NVIDIA Corporation TU106 [GeForce RTX 2060 Rev. A] [10de:1f08] (rev a1)
	Subsystem: Micro-Star International Co., Ltd. [MSI] TU106 [GeForce RTX 2060 Rev. A] [1462:3755]
65:00.1 Audio device [0403]: NVIDIA Corporation TU106 High Definition Audio Controller [10de:10f9] (rev a1)
	Subsystem: Micro-Star International Co., Ltd. [MSI] TU106 High Definition Audio Controller [1462:3755]
65:00.2 USB controller [0c03]: NVIDIA Corporation TU106 USB 3.1 Host Controller [10de:1ada] (rev a1)
	Subsystem: Micro-Star International Co., Ltd. [MSI] TU106 USB 3.1 Host Controller [1462:3755]
	Kernel driver in use: xhci_hcd
65:00.3 Serial bus controller [0c80]: NVIDIA Corporation TU106 USB Type-C UCSI Controller [10de:1adb] (rev a1)
	Subsystem: Micro-Star International Co., Ltd. [MSI] TU106 USB Type-C UCSI Controller [1462:3755]
b2:05.0 System peripheral [0880]: Intel Corporation Sky Lake-E VT-d [8086:2034] (rev 04)
	Subsystem: Intel Corporation Sky Lake-E VT-d [8086:0000]

 

tower-diagnostics-20240817-1511 sans GPU with plugin

>> no card mounted / visible ? plugin installed successfully ... error, yes as there is no card ...

>> device's 65:00.N >> missing ...

 

64:0d.3 System peripheral [0880]: Intel Corporation Sky Lake-E LMDP Channel 2 [8086:204b] (rev 04)
	Subsystem: Intel Corporation Sky Lake-E LMDP Channel 2 [8086:0000]
b2:05.0 System peripheral [0880]: Intel Corporation Sky Lake-E VT-d [8086:2034] (rev 04)
	Subsystem: Intel Corporation Sky Lake-E VT-d [8086:0000]

 

may look if the card sits properly

may also NOT use the open source one, just regular latest driver, also remove the modprobe 

 

 

so, to really see a error there is a boot required with GPU installed and visible (lspci), then install the driver.

 

may activate for debugging now syslog to flash when you say your system crashes while the nvidia-smi is triggered

(deactivate it when debug is done)

 

general spoken

check seat of the card

check PSU is good enough and connected properly

try another slot on the Board

BIOS - latest update done

BIOS - UEFI boot, rBAR and above 4g is activated

[shabash]

On 8/11/2024 at 10:08 PM, ich777 said:

 

 

No, why?

As you can see here your GPU is still supported and my GPU (Nvidia T400) is still working just fine with Jellyfin and Unraid 7.0

 

What container are you using? From which maintainer?

 

I just checked and I see that there are about 750 downloads on the latest Nvidia Driver so I assume the driver is working for other users too.

It looks like 550.107.02 isn't an option in the plugin. What should I choose?

 

I'm using LSIO Plex

Edited August 18 by shabash

[ich777]

52 minutes ago, shabash said:

What should I choose?

I don't know since you still didn't provided Diagnostics.

Troubleshooting without more information is really hard, sorry but I can't help until you provide your Diagnostics.

[shabash]

4 minutes ago, ich777 said:

I don't know since you still didn't provided Diagnostics.

Troubleshooting without more information is really hard, sorry but I can't help until you provide your Diagnostics.

Sorry about that

https://mega.nz/file/AVtmRCzL#6Vl5JDbwqzwB5IIWu4vhfMiGf2XG_j05YdOsAvDFii0

[ich777]

43 minutes ago, shabash said:

Sorry about that

Just for next time, you can upload that directly here by just dropping the zip into the text field here.

 

Please upgrade to Unraid 6.12.11

 

However, I think this issue is related to the container not to the driver since everything seems to be working fine from the driver side.

[clincher]

11 hours ago, alturismo said:

 

tower-diagnostics-20240817-1511 sans GPU with plugin

>> no card mounted / visible ? plugin installed successfully ... error, yes as there is no card ...

>> device's 65:00.N >> missing ...

Yes, there were two diagnostics.  One with the GPU installed and no plugin and one diagnostic without (sans) the GPU installed and plugin installed.  The card is seated fine when installed.

 

If I try and boot the system WITH the GPU installed and plugin installed, the system will hang during initialization and crash.

 

may also NOT use the open source one, just regular latest driver, also remove the modprobe 

 - Yes, I've tried the standard driver and the open source. Both exhibit the same behavior.  Hang and then crash reboot.

11 hours ago, alturismo said:

so, to really see a error there is a boot required with GPU installed and visible (lspci), then install the driver.

 

may activate for debugging now syslog to flash when you say your system crashes while the nvidia-smi is triggered

(deactivate it when debug is done)

 

general spoken

check seat of the card

check PSU is good enough and connected properly

try another slot on the Board

BIOS - latest update done

BIOS - UEFI boot, rBAR and above 4g is activated

 

I've tried all of the above.  My BIOS doesn't have an RBAR setting anywhere that I can find.  4G is enabled.  Newest BIOS for Supermicro is installed.

 

 I'm guessing this is just a Supermicro incompatability.  Unfortunately, I can't get any useful logs or diagnostics because the system freezes up and doesn't not record anything to syslog or console.

 

When the GPU is installed and plugin is installed, the system hangs at the following console statement during bootup:

 

Starting atd: /usr/sbin/atd -b 15 -l 1
Starting Samba: /usr/sbin/smbd -D
				/usr/sbin/wsdd2 -d -4
				/usr/sbin/winbindd -D
Starting mcelog daemon: /usr/sbin/mcelog --daemon

Then the system hangs and reboots.  I can ping the system IP and see that network begins to come up for about 3-4 pings, but stops when the system hangs.   This is essentially how I am testing if the system hangs.

 

Think I'm going to start searching for an alternative motherboard.

 

 

 

[alturismo]

1 hour ago, clincher said:

Think I'm going to start searching for an alternative motherboard.

 

 

13 hours ago, alturismo said:

try another slot on the Board

 

also tested ?

[clincher]

2 minutes ago, alturismo said:

 

also tested ?

Yes, there are two x16 slots.  I've tried both.

[ich777]

19 hours ago, clincher said:

Grasping at straws.  Thanks!

Does the card work in another computer?

 

Please be aware that some users with Supermicro/Dell/HP Servers/Motherboards have issues because of Firmware incompatibly issues, some users even got a new Firmware from support that eventually fixed the issues.

It would be really interesting what is showing up when you install the driver, I bet it is related to memory allocation but I really can't tell for sure.

[clincher]

1 hour ago, ich777 said:

Does the card work in another computer?

 

Please be aware that some users with Supermicro/Dell/HP Servers/Motherboards have issues because of Firmware incompatibly issues, some users even got a new Firmware from support that eventually fixed the issues.

It would be really interesting what is showing up when you install the driver, I bet it is related to memory allocation but I really can't tell for sure.

Yes, works fine in another computer.  As stated above, I tried two other Nvidia GPUs with the same behavior (GTX 1080 and GTX 650).  Both hang and crash when installed on the board.  Maybe I'll hit up Supermicro and see if there's something known.

[ich777]

1 minute ago, clincher said:

Yes, works fine in another computer.  As stated above, I tried two other Nvidia GPUs with the same behavior (GTX 1080 and GTX 650).  Both hang and crash when installed on the board.  Maybe I'll hit up Supermicro and see if there's something known.

Most likely like I said above a Firmware or hardware incompatibility issue.

[clincher]

5 hours ago, ich777 said:

Most likely like I said above a Firmware or hardware incompatibility issue.

You were correct.  This ended up being a CPU issue.  I have three 6132 Xeons I got from work.  I was finally able to get video on my 2060 AND get the plugin working after trying the 3rd Xeon!  

Thanks for the advice!

[heavyghost]

Heyo gang, unraid os newbie here, hope my question isn't too stupid.

 

I have gotten my hands on an old HP ProLiant Microserver (N36L) and want to beef it up with a gpu to help with Jellyfin transcoding. Its got a 16x PCI low rise slot and I was thinking of getting a GT1030 card in it (Pascal) for a cheap upgrade. It looks like the card is currently supported by Nvidia here: Appendix A. Supported NVIDIA GPU Product,.

 

However, I'm trying to figure out which driver it would actually use with the Nvidia Driver plugin when I actually hook it up. I know there are some changes coming up with the next version of UnRaid OS and the Linux kernel it will use possibly breaking compatibility for GPUs using the v470.xx driver, which is used by Maxwell cards and older (I think?).

 

Could anyone advise how long Pascal cards will supported by the Nvidia-Driver plugin?

TIA

[ich777]

51 minutes ago, heavyghost said:

and I was thinking of getting a GT1030 card in it (Pascal) for a cheap upgrade.

This is the worst choice for transcoding because according to the Nvidia Matrix here, this card isn't even capable of Encoding h264 so that you can use it for transcoding.

 

54 minutes ago, heavyghost said:

It looks like the card is currently supported by Nvidia here: Appendix A. Supported NVIDIA GPU Product,.

You can also take a look here.

Pascal cards should be supported for a few more years now but you never know since Nvidia moved on to the Open Source Kernel module by default for the 560 driver series and up, so to speak only Turing and newer cards will be supported there.

However there is always the option to use the closed source Kernel module (which I do in my packages because there are a lot of Pascal cards out there) with 560 drivers, but I really can't tell if it's developed further.

 

57 minutes ago, heavyghost said:

However, I'm trying to figure out which driver it would actually use with the Nvidia Driver plugin when I actually hook it up.

All will work except for the Open Source option in the driver plugin.

 

58 minutes ago, heavyghost said:

there are some changes coming up with the next version of UnRaid OS and the Linux kernel it will use possibly breaking compatibility for GPUs using the v470.xx driver

Not for now.

 

58 minutes ago, heavyghost said:

Could anyone advise how long Pascal cards will supported by the Nvidia-Driver plugin?

I would rather go that route and buy something like a Nvidia T400, it is low power, you can usually get it for cheap, is Turing based, low profile (depends if you get the bracket with the card) and is enough for 4x 4k transcodes (depending on the source and output settings).

[heavyghost]

Awesome, thanks for the detailed reply @ich777, especially that nvidia support matrix, super handy!

 

The Quadro T400 looks like a decent option but it's a bit pricey for my budget. Any suggestions for a supported (used) low rise cards around the $75 USD mark?

[ich777]

8 minutes ago, heavyghost said:

The Quadro T400 looks like a decent option but it's a bit pricey for my budget. Any suggestions for a supported (used) low rise cards around the $75 USD mark?

Please don't call it Quadro...

Nvidia is very particular about their products and AFAIK they dropped the QUADRO "thing" for this line of cards, its a NVIDIA T400.

 

You should be able to get such a card used for much cheaper than they cost new. Keep in mind that there are two T400's out there one with 2GB and one with 4GB VRAM <- I have the 2GB variant and it is well enough for transcoding.

 

Other than that you can just step down to a Pascal card eg: Nvidia Quadro P400 however I would recommend that you invest a bit more and spend the extra on the T400