2 hours ago, aeryk said:

But it does show up and get passed through to a Windows VM. Odd?

nope, normal behaviour ...

 

passthrough (or vfio bind) == gone for the host == gone for nvidia-smi == gone for this plugin

2 hours ago, alturismo said:

nope, normal behaviour ...

 

passthrough (or vfio bind) == gone for the host == gone for nvidia-smi == gone for this plugin

Oh, so how do I get my GPU UUID for the plex argument NVIDIA_VISIBLE_DEVICES?

16 minutes ago, aeryk said:

Oh, so how do I get my GPU UUID for the plex argument NVIDIA_VISIBLE_DEVICES?

same as above ... either passthrough VM ... OR ... Host usage (Dockers, ...)

 

like described on page 1 from the NV plugin, just read the instructions

5 hours ago, aeryk said:

I'm not seeing my 3080 Ti in the GPU list, and the command says: "NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running."

 

But it does show up and get passed through to a Windows VM. Odd?

As alturismo has said if you passthru a GPU to a VM it is isolated from the host and dedicated to the VM. You need another gpu if you need to transcode at the same time.

This plugin is gone from my dashboard.

I found it under failed installed plugins, so I deleted it from there.

Tried to download it again from CA, but there is no install button.

Where it usually says install, is says "Pending" now

 

How can I get this neat plugin back on my dashboard?

 

Thanks

1 hour ago, SimonF said:

As alturismo has said if you passthru a GPU to a VM it is isolated from the host and dedicated to the VM. You need another gpu if you need to transcode at the same time.

That's not what I meant. Even if the GPU isn't assigned to a VM, or all VMs are stopped, it still doesn't show my GPU in the list and the GPU info plugin doesn't work, and Plex is not using hardware acceleration (it uses all CPUs).

 

As far as this plugin, the other plugin, and the dockers are concerned, my GPU doesn't exist.

 

 

  

1 hour ago, alturismo said:

same as above ... either passthrough VM ... OR ... Host usage (Dockers, ...)

 

like described on page 1 from the NV plugin, just read the instructions

 

I read them. They only work if the GPU shows up, which mine doesn't =(

 

I'm just not sure why that is...

 

Edited February 8, 20242 yr by aeryk

1 hour ago, aeryk said:

That's not what I meant. Even if the GPU isn't assigned to a VM, or all VMs are stopped, it still doesn't show my GPU in the list and the GPU info plugin doesn't work, and Plex is not using hardware acceleration (it uses all CPUs).

 

As far as this plugin, the other plugin, and the dockers are concerned, my GPU doesn't exist.

 

 

  

 

I read them. They only work if the GPU shows up, which mine doesn't =(

 

I'm just not sure why that is...

 

Hi,

 

Can you provide cat /tmp/gpujson  I suspect you have a driver issue.

7 hours ago, aeryk said:

As far as this plugin, the other plugin, and the dockers are concerned, my GPU doesn't exist.

 

 

10 hours ago, alturismo said:

passthrough (or vfio bind) == gone for the host == gone for nvidia-smi == gone for this plugin

 

and your card is also not vfio bound ?

I'm confused. Does this plugin support 6.12 or not? Update Assistant says no, but reading this thread sounds like it does.

3 minutes ago, Iceman24 said:

I'm confused. Does this plugin support 6.12 or not? Update Assistant says no, but reading this thread sounds like it does.

Yes if you install my version from CA

 

any discussion on this issue and if it is actually related to this plugin?  I am sufferring this issue and would like to keep the plugin, if possible!  thanks

28 minutes ago, sMACk said:

 

any discussion on this issue and if it is actually related to this plugin?  I am sufferring this issue and would like to keep the plugin, if possible!  thanks

I dont believe it is related to this plugin, I not see the issue on current releases and it used to happen during a parity check for me if I remember correctly. reducing refresh time in the plugin may help. Also I was not running gpustat on the system that was having an issue.

I've been having buffering/transcoding issues and I noticed that my gpu statistics plugin shows "Vendor data valid, but not enough received." most of the time. I tried removing intel gpu top and gpu statistics, rebooting, and reinstalling them, but my issues seem to persist. I'm attaching all of the imagery and debug info people seem to be requesting and would really appreciate help!

The only numbers that seem to change in the statistics panel are the 'IMC utilization' and 'interrupts/sec' fields.

pms-diagnostics-20240213-2151.zip

1 hour ago, picky said:

I've been having buffering/transcoding issues and I noticed that my gpu statistics plugin shows

may rather take a look on the Media Server App logs (Plex, Emby, Jellyfin, whatever ...) what it states there and look for the errors happening there while transcoding.

 

this plugin is just reading the usage and has usually nothing todo with your issues above, so might be the wrong place to ask ...

4 hours ago, picky said:

I've been having buffering/transcoding issues and I noticed that my gpu statistics plugin shows "Vendor data valid, but not enough received." most of the time. I tried removing intel gpu top and gpu statistics, rebooting, and reinstalling them, but my issues seem to persist. I'm attaching all of the imagery and debug info people seem to be requesting and would really appreciate help!

The only numbers that seem to change in the statistics panel are the 'IMC utilization' and 'interrupts/sec' fields.

pms-diagnostics-20240213-2151.zip 88.23 kB · 1 download

running gpustatus.php is pre 6.12 can you provide cat /tmp/gpujson

9 hours ago, SimonF said:

running gpustatus.php is pre 6.12 can you provide cat /tmp/gpujson

root@PMS:~# cat /tmp/gpujson
Time = Wed, 14 Feb 2024 10:40:44 -0500
{"00:02.0":{"clock":0,"fan":"N\/A","memclock":"N\/A","memutil":"N\/A","memused":"N\/A","power":"0W","powermax":"N\/A","rxutil":"186.29 MB\/s","txutil":"24.97 MB\/s","temp":"N\/A","tempmax":"N\/A","util":"0%","vendor":"Intel","name":"HD Graphics 630","3drender":"0%","blitter":"0%","interrupts":0,"powerutil":"0%","video":"0%","videnh":"0%","vfio":false,"vfiochk":false,"vfiochkid":"0000:00:02.0","panel":1}}

 

12 hours ago, alturismo said:

may rather take a look on the Media Server App logs (Plex, Emby, Jellyfin, whatever ...) what it states there and look for the errors happening there while transcoding.

 

this plugin is just reading the usage and has usually nothing todo with your issues above, so might be the wrong place to ask ...

I don't doubt this could be the case. I do feel like people generally have some activity reported on gpu statistics while idle but in my case it's always stark... I assumed my GPU just isn't interfacing with the OS let alone plex. I've tried looking at plex and tautulli logs but honestly I don't know how to decipher much and figured I'd start here.

Edited February 14, 20242 yr by picky

6 hours ago, picky said:

root@PMS:~# cat /tmp/gpujson
Time = Wed, 14 Feb 2024 10:40:44 -0500
{"00:02.0":{"clock":0,"fan":"N\/A","memclock":"N\/A","memutil":"N\/A","memused":"N\/A","power":"0W","powermax":"N\/A","rxutil":"186.29 MB\/s","txutil":"24.97 MB\/s","temp":"N\/A","tempmax":"N\/A","util":"0%","vendor":"Intel","name":"HD Graphics 630","3drender":"0%","blitter":"0%","interrupts":0,"powerutil":"0%","video":"0%","videnh":"0%","vfio":false,"vfiochk":false,"vfiochkid":"0000:00:02.0","panel":1}}

This looks correct, it seems to be the output of IntelGPUTop is not providing any info apart from IMC from the other screen shots you have provided.

On 2/13/2024 at 8:11 PM, picky said:

I've been having buffering/transcoding issues and I noticed that my gpu statistics plugin shows "Vendor data valid, but not enough received." most of the time. I tried removing intel gpu top and gpu statistics, rebooting, and reinstalling them, but my issues seem to persist. I'm attaching all of the imagery and debug info people seem to be requesting and would really appreciate help!

The only numbers that seem to change in the statistics panel are the 'IMC utilization' and 'interrupts/sec' fields.

 

 

I too am experiencing the exact same issue.   The most recent change I have made is I upgraded to 6.12.8.  I wonder if there is some driver related issue on the igpu side of things?

12 hours ago, KimoB said:

 

I too am experiencing the exact same issue.   The most recent change I have made is I upgraded to 6.12.8.  I wonder if there is some driver related issue on the igpu side of things?

Probably worth noting I switched to plex due to transcode/buffer issues with jellyfin. Same issues with plex so I'm assuming they arn't the issue.

Is the 14700 supported by this plugin?

I recently upgraded from an 11900K but now GPU stats shows nothing in the dashboard

5 minutes ago, mikeyosm said:

Is the 14700 supported by this plugin?

I think the Kernel doesn‘t support your 14700, you have to wait for a new RC/beta version with a higher Kernel than 6.1.x

1 minute ago, ich777 said:

I think the Kernel doesn‘t support your 14700, you have to wait for a new RC/beta version with a higher Kernel than 6.1.x

Bugger, also noticed that Plex doesnt see the iGPU for transcoding 😞 

Just now, mikeyosm said:

Bugger, also noticed that Plex doesnt see the iGPU for transcoding 😞 

Bleeding edge on Linux is never a good idea…

As said wait a few weeks and you will maybe see a new RC with a recent Kernel.

1 hour ago, mikeyosm said:

Bugger, also noticed that Plex doesnt see the iGPU for transcoding 😞 

Do you get any messages?

 

root@computenode:~# dmesg |grep i915
[   50.985890] i915 0000:00:02.0: [drm] VT-d active for gfx access
[   50.985939] i915 0000:00:02.0: [drm] Using Transparent Hugepages
[   50.986572] i915 0000:00:02.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem
[   50.992549] i915 0000:00:02.0: [drm] Finished loading DMC firmware i915/adls_dmc_ver2_01.bin (v2.1)
[   51.021951] i915 0000:00:02.0: [drm] GT0: GuC firmware i915/tgl_guc_70.bin version 70.13.1
[   51.021955] i915 0000:00:02.0: [drm] GT0: HuC firmware i915/tgl_huc.bin version 7.9.3
[   51.035422] i915 0000:00:02.0: [drm] GT0: HuC: authenticated for all workloads
[   51.035427] i915 0000:00:02.0: [drm] GT0: GUC: submission disabled
[   51.035429] i915 0000:00:02.0: [drm] GT0: GUC: SLPC disabled
[   51.036259] i915 0000:00:02.0: [drm] Protected Xe Path (PXP) protected content support initialized
[   52.537318] [drm] Initialized i915 1.6.0 20201103 for 0000:00:02.0 on minor 1
[   52.539040] i915 0000:00:02.0: [drm] Cannot find any crtc or sizes
[   52.539530] i915 0000:00:02.0: [drm] Cannot find any crtc or sizes
[   52.539600] i915 0000:00:02.0: [drm] Cannot find any crtc or sizes
[   52.674124] mei_pxp 0000:00:16.0-fbf6fcf1-96cf-4e2e-a6a6-1bab8cbe36b1: bound 0000:00:02.0 (ops i915_pxp_tee_component_ops [i915])
[   52.677122] mei_hdcp 0000:00:16.0-b638ab7e-94e2-4ea2-a552-d1c54b627f04: bound 0000:00:02.0 (ops i915_hdcp_ops [i915])

For the "Vendor data valid, but not enough received." folks that are also Intel iGPUs, try this in the GPU Stats configuration UI...

 

Set (or confirm) that "UI Automatic Refresh / Interval (Milliseconds):" is not lower than 1000.

Set "Intel_top_command_timeout:" to "1 .9"

 

My system is getting older (Coffee Lake / Gen8) and I've noticed that the stats panel is having trouble when my system is really bogged down.

 

Hope this helps someone!