1 minute ago, forumi0721 said:

Hi, I’m currently using an RTX 5060 Ti.

When I try to use Latest / Production Branch / New Feature Branch, I get a message recommending that I use the Open Source Driver instead.
However, I noticed that the Open Source Driver version is lower than the other branches:

• Latest: v580.76.05

• Production Branch: v580.76.05

• New Feature Branch: v575.64.05

• Open Source Driver: v575.57.08

On GitHub, I can see that the Open Source Driver goes up to v580.76.05, but in Unraid it only shows v575.57.08.
Does this mean the Open Source Driver is no longer being updated for Unraid?

Thanks in advance for your clarification.

You need to use open source. Otherwise your gpu will not work. At least my 5060 ti didnt want without open source drivers.

3 minutes ago, xSniper said:

You need to use open source. Otherwise your gpu will not work. At least my 5060 ti didnt want without open source drivers.

Thanks for your comment. Yes, I’m already using the Open Source driver — I understand that’s required for the 5060 Ti to work.

My question was specifically about the version difference: in the NVIDIA plugin, the officially supported Open Source Driver is still at v575.57.08, while the other branches (Latest / Production) are already at v580.76.05.

That’s why I was asking whether the Open Source Driver is going to be updated further, or if it will remain behind the other branches in Unraid.

It will get updated, dont worry ;)

10 hours ago, hobz said:

I have updated the BIOS for my Z270E, last updated in 2018.
Enabled Above 4G Decoding and it is definitely booting in UEFI.

There's no Resizable BAR Support, but I was wondering whether its possible to install an older graphics driver via the plugin.

Only the ones which are available through the plugin.

3 hours ago, forumi0721 said:

My question was specifically about the version difference: in the NVIDIA plugin, the officially supported Open Source Driver is still at v575.57.08, while the other branches (Latest / Production) are already at v580.76.05.

Really important information missing, your Diagnostics.

I only compile the Open Source driver when a new Unraid version is released, most people use their cards only for transcoding and the performance impact is negligible with an older driver.

However that all changes with the new drivers because the next driver from 580 series should be all be based on the Open Source Driver and the option Option Source will be removed a bit after that takes effect.

10 hours ago, ich777 said:

Really important information missing, your Diagnostics.

I only compile the Open Source driver when a new Unraid version is released, most people use their cards only for transcoding and the performance impact is negligible with an older driver.

However that all changes with the new drivers because the next driver from 580 series should be all be based on the Open Source Driver and the option Option Source will be removed a bit after that takes effect.

Thank you for the clarification.

I tried installing the Production Branch (v580.76.05), but I’m getting the following messages in the logs:

[   28.008621] nvidia: module license 'NVIDIA' taints kernel.
[   28.088155] nvidia-nvlink: Nvlink Core is being initialized, major device number 238
[   28.112649] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  580.76.05  Thu Aug  7 20:32:41 UTC 2025
[   28.117677] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms  580.76.05  Thu Aug  7 19:56:57 UTC 2025
[   28.128666] NVRM: The NVIDIA GPU 0000:01:00.0 (PCI ID: 10de:2d04)
               NVRM: installed in this system requires use of the NVIDIA open kernel modules.
[   28.129486] [drm:nv_drm_dev_load [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Failed to allocate NvKmsKapiDevice
[   28.129496] [drm:nv_drm_register_drm_device [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Failed to load device
...
(repeated messages: requires use of the NVIDIA open kernel modules)

It seems the card refuses to load unless the Open Source Driver is used.
I have attached my diagnostics for reference.

stonecoldsvr-diagnostics-20250821-1210.zip

14 hours ago, forumi0721 said:

My question was specifically about the version difference: in the NVIDIA plugin, the officially supported Open Source Driver is still at v575.57.08, while the other branches (Latest / Production) are already at v580.76.05.

as sample today

so, keep your Unraid updated and your NV driver also will be ...

2 hours ago, forumi0721 said:

I tried installing the Production Branch (v580.76.05), but I’m getting the following messages in the logs:

Please read my response again:

13 hours ago, ich777 said:

However that all changes with the new drivers because the next driver from 580 series should be all be based on the Open Source Driver and the option Option Source will be removed a bit after that takes effect.

Nvidia already said that they will deprecate support for older cards (10xx series and lower) in the 580 driver series but it seems like in the first release from this driver series the cards are still supported.

You have to stick to the Open Source driver version for now.

7 hours ago, ich777 said:

Please read my response again:

Nvidia already said that they will deprecate support for older cards (10xx series and lower) in the 580 driver series but it seems like in the first release from this driver series the cards are still supported.

You have to stick to the Open Source driver version for now.

Thank you for the explanation.
I understand now that starting with the 580 series everything will move to the Open Source driver.

In this case, I’ll stick with the released 6.12.42-unraid kernel (or the beta) and wait for the next version.
Thanks again!

Are there known issues with the open source driver? I upgraded to 7.1.4 today and swapped out my 1080TI for a 5080 and my system is completely unusable. I keep having my CPU (5900x) pegged at 100% with the culprit being Xorg, which usually points to a display/driver issue. The webui, ssh, docker containers, etc. everything halts. I was previously running 6.X and everything was working flawlessly. At this point it seems to either be the opensource driver or 7.1.4, or both, but it's completely hosed my system.

Looking at the output of dmesg I see:

dmesg logs

[  627.215967] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[  627.216010] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[  627.216784] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 1021.247207] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 1021.247287] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 1021.253483] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 1490.317300] br-2c59c634a98f: port 1(veth9886930) entered disabled state

[ 1490.317767] veth9886930 (unregistering): left allmulticast mode

[ 1490.317772] veth9886930 (unregistering): left promiscuous mode

[ 1490.317777] br-2c59c634a98f: port 1(veth9886930) entered disabled state

[ 1770.432845] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 1770.432891] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 1770.460024] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 2248.983618] br-ef5b47530577: port 1(veth926cb67) entered disabled state

[ 2248.983721] veth8d97af8: renamed from eth0

[ 2249.084737] br-ef5b47530577: port 1(veth926cb67) entered disabled state

[ 2249.085149] veth926cb67 (unregistering): left allmulticast mode

[ 2249.085154] veth926cb67 (unregistering): left promiscuous mode

[ 2249.085158] br-ef5b47530577: port 1(veth926cb67) entered disabled state

[ 2344.157509] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 2344.157540] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 2344.158139] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 2944.742794] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 2944.742830] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 2944.744048] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 3544.206283] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 3544.206321] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 3544.207614] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 4144.654801] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 4144.654834] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 4144.656173] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 4744.145701] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 4744.145739] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 4744.146981] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 5344.639279] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 5344.639313] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 5344.640567] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 5943.992658] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 5943.992696] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 5943.993961] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

[ 6460.826757] br0: port 2(vnet0) entered disabled state

[ 6460.827439] vnet0 (unregistering): left allmulticast mode

[ 6460.827443] vnet0 (unregistering): left promiscuous mode

[ 6460.827448] br0: port 2(vnet0) entered disabled state

[ 6464.808303] br0: port 2(vnet3) entered blocking state

[ 6464.808309] br0: port 2(vnet3) entered disabled state

[ 6464.808317] vnet3: entered allmulticast mode

[ 6464.808441] vnet3: entered promiscuous mode

[ 6464.808822] br0: port 2(vnet3) entered blocking state

[ 6464.808825] br0: port 2(vnet3) entered forwarding state

[ 6469.487830] br0: port 4(vnet2) entered disabled state

[ 6469.488622] vnet2 (unregistering): left allmulticast mode

[ 6469.488626] vnet2 (unregistering): left promiscuous mode

[ 6469.488631] br0: port 4(vnet2) entered disabled state

[ 6470.810524] br0: port 4(vnet4) entered blocking state

[ 6470.810530] br0: port 4(vnet4) entered disabled state

[ 6470.810539] vnet4: entered allmulticast mode

[ 6470.810628] vnet4: entered promiscuous mode

[ 6470.811169] br0: port 4(vnet4) entered blocking state

[ 6470.811175] br0: port 4(vnet4) entered forwarding state

[ 6544.504784] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[ 6544.504842] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[ 6544.506123] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

I've tried uninstalling nvidia drivers, rebooting, then reinstalling the drivers but it's the same result. The box just hangs every few minutes for about 5 minutes pegged at 100% CPU usage and then it goes back to normal for a few minutes.

I had a gl.inet comet KVM plugged into the GPU before. I have unplugged the HDMI cable and that may have fixed it? Will need to continue to do troubleshooting, but it would appear with these drivers that you can only operate everything completely headless otherwise it has these lockups?

Edited August 22, 2025Aug 22 by fnhanks
adding additional question/logs

5 hours ago, fnhanks said:

Will need to continue to do troubleshooting

Diagnostics would be really be helpful.

Try to lower the PCIe Generation in your BIOS to Gen 3 or Gen 4, make sure that resizable bar support and above 4g decoding is enabled in your BIOS.

diagnostics-20250822-0831-WITHOUT-KVM.zip

The box ran fine for 18 hours without the HDMI cord plugged in. I've attached the diagnostics pulled prior to plugging it back in. The PCIE gen is set to auto in the BIOS and is currently running gen4.

I've plugged the HDMI cord back in and will continue to monitor to see when it does it again so I can grab diagnostics at that point. I can't reboot it at the moment to check resizable bar support and above 4g decoding, but will do that later today.

Plugged the KVM in and the issues seem to have resumed.

Looking in dmesg, I see more of the same error:

[  529.708730] NVRM: rpcRmApiAlloc_GSP: GspRmAlloc failed: hClient=0xc1d00001; hParent=0x00010011; hObject=0x00010016; hClass=0x0000ca7d; paramsSize=0x00000028; paramsStatus=0x00000065; status=0x00000065

[  529.708785] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])

[  529.709959] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer

I've attached the diagnostics that include the KVM.

goober-diagnostics-20250823-0724-WITH-KVM.zip

9 minutes ago, fnhanks said:

Plugged the KVM in and the issues seem to have resumed.

Possibly an issue with your KVM?

Did you already try to plug in a real monitor?

After unplugging, still getting the errors. I haven't rebooted since unplugging, so still trying to 100% confirm what the culprit is. Wondering if plugging in the KVM causes an issue that doesn't get cleared up until a reboot, as it's still doing it even after being unplugged.

One thing that's interesting, is I think it might be somewhat related to the display going to sleep, as the KVM will say no HDMI output.

Haven't tried with a real monitor yet, as this is sitting inside a rack, so somewhat difficult to do. Will try and use one of my longer HDMI cables to test.

Just rebooted the server to see if the issue comes back with the KVM unplugged. If it does, then I know it's something else. If it doesn't, then I'll try and plug a real monitor in to see if it comes back to confirm whether it's any display or just this KVM.

Edited August 23, 2025Aug 23 by fnhanks

1 hour ago, fnhanks said:

One thing that's interesting, is I think it might be somewhat related to the display going to sleep, as the KVM will say no HDMI output.

You know that you can disable display sleep in the settings from Unraid.

1 hour ago, fnhanks said:

Haven't tried with a real monitor yet, as this is sitting inside a rack, so somewhat difficult to do. Will try and use one of my longer HDMI cables to test.

Yes, please, maybe it‘s related to the KVM since it‘s usually emulating an EDID.

Hello,

I have a quadro P4000 and I just ran the update from 7.0.1 to 7.1.4. I got two error:

I haven't rebooted yet since the upgrade because of the error. I'm unsure what to do.

Thank you

edit: I think I found the problem but now, I'm unsure what to do. Still waiting for the reboot post upgrade, it seems my usb drive is full!

I do have another usb drive that I bought about 6 months ago to replace it, never done it. Can I do it now while it's pending the reboot? What should I do?

Edited August 23, 2025Aug 23 by Nodiaque

9 minutes ago, Nodiaque said:

edit: I think I found the problem but now, I'm unsure what to do. Still waiting for the reboot post upgrade, it seems my usb drive is full!

Please upgrade you USB Boot device since you need at least about 400MB of free space on the USB Boot device for an upgarde from the Nvidia Driver.

However the plugin tries to download the new driver on boot anyways but please keep in mind that the boot will take longer.

Ok, so do I just reboot to finish the upgrade to 7.1.4 without nvidia driver, backup the flash and swap the usb or do I swap before finishing the 7.1.4 upgrade?

I did manage to put 360mb space by deleting old package. I do have 800mb that I could get from the previous folder but unsure if it's wise to delete that folder before finishing the upgrade to 7.1.4

Edited August 23, 2025Aug 23 by Nodiaque

9 hours ago, Nodiaque said:

Ok, so do I just reboot to finish the upgrade to 7.1.4 without nvidia driver, backup the flash and swap the usb or do I swap before finishing the 7.1.4 upgrade?

You can do this how you like it and it doesn‘t matter much.

9 hours ago, Nodiaque said:

I do have 800mb that I could get from the previous folder but unsure if it's wise to delete that folder before finishing the upgrade to 7.1.4

You can delete it if you want to.

How exactly do you install the Open Source Driver? I just swapped a 3060 for a 5070

45 minutes ago, lukeb873 said:

How exactly do you install the Open Source Driver? I just swapped a 3060 for a 5070

Did you yet visited the plugin page?

Wow, I feel like an idiot. This would be me straight up not reading the other options in the driver. Thank you

Edited August 24, 2025Aug 24 by lukeb873

I'm starting to wonder if the driver repo is being throttled by GitHub in certain regions btw.

https://docs.github.com/en/repositories/creating-and-managing-repositories/repository-limits

7 hours ago, Inch said:

I'm starting to wonder if the driver repo is being throttled by GitHub in certain regions btw.

Nope, this is most of the times caused by your ISP and general routing issues over all.

Especially in Europe we have to fight with low download speeds at times, yesterday I had a download with only about 100KB/s

BTW the page that you linked is for Repositories itself not the Releases which the plugin is relying on.

You can try the download through a proxy and you will most likely see that it is working just fine, as said, that‘s most of the times caused by the ISP‘s.