[PLUGIN] GPU Statistics

[sersh]

10 minutes ago, SimonF said:

What do you have the main refresh timer set to?

I use default settings 1000. Ok. I changed it to 30000 and the load of the processor become normal.

Edited December 9, 2023 by sersh

[SimonF]

27 minutes ago, sersh said:

I use default settings 1000. Ok. I changed it to 30000 and the load of the processor become normal.

You maybe can reduce the override timers as will run for 6 secs every 30 secs

[sersh]

On 12/9/2023 at 11:11 AM, SimonF said:

You maybe can reduce the override timers as will run for 6 secs every 30 secs

Which indicator is responsible for what?

Edited December 12, 2023 by sersh

[mikeyosm]

@SimonF Disabling the fan/power for nvidia/amd in settings does not remove the fan stat from the dashboard.

[ich777]

1 minute ago, mikeyosm said:

Disabling the fan/power for nvidia/amd in settings does not remove the fan stat from the dashboard.

I think the bar that you see that says N/A is the Power Draw, is that also disabled?

[mikeyosm]

2 minutes ago, ich777 said:

I think the bar that you see that says N/A is the Power Draw, is that also disabled?

yeah i had to disable power draw to remove the fan from the dash. Maybe you could change it so disabling fan removes the label?

[SimonF]

7 hours ago, sersh said:

Which indicator is responsible for what?

it is for the linux command timeout when running intel_gpu_top command.

-k is kill

timeout -k .500 .600

 

So will run command for .6 of a second and kill after .5

[KillerVette42]

Just setup a new server. It’s a 13th gen i7. Gpu statistics used to work on my 10th gen, but now I am not seeing any stats. The device is raptor lake and is selected in the settings, but no data shows on the dashboard. Any ideas?

[ich777]

4 hours ago, KillerVette42 said:

The device is raptor lake and is selected in the settings, but no data shows on the dashboard. Any ideas?

Please provide your Diagnostigs alongside with the file: /tmp/gpujson (or simply open up a terminal and issue: cat /tmp/gpujson)

[KillerVette42]

On 12/15/2023 at 12:53 AM, ich777 said:

Please provide your Diagnostigs alongside with the file: /tmp/gpujson (or simply open up a terminal and issue: cat /tmp/gpujson)

actually, everything is looking good now. I am using qsv on a transcode and the stats are showing. Thanks. 

[Chris69]

On 12/15/2023 at 6:53 AM, ich777 said:

Please provide your Diagnostigs alongside with the file: /tmp/gpujson (or simply open up a terminal and issue: cat /tmp/gpujson)

 

Hi guys my igpu is not displayed do you have any idea what is wrong?

 

Time = Wed, 20 Dec 2023 18:08:26 -0800
{"00:02.0":{"clock":0,"fan":"N\/A","memclock":"N\/A","memutil":"N\/A","memused":"N\/A","power":"0W","powermax":"N\/A","rxutil":"N\/A","txutil":"N\/A","temp":"N\/A","tempmax":"N\/A","util":"0%","vendor":"Intel","name":"Raptor Lake-S GT1 [UHD Graphics 770]","3drender":"0%","blitter":"0%","interrupts":0,"powerutil":"0%","video":"0%","videnh":"0%","vfio":false,"vfiochk":false,"vfiochkid":"0000:00:02.0","panel":1}}

[SimonF]

That looks normal for at idle are you putting any workload thru the GPU, i.e. transcoding?

[Chris69]

The problem has been solved, I used emby and I had to assign the igpu to emby now everything is transcoded via the igpu and it is also displayed, thanks for the help😅

 

[budman789]

Hey mate, would love if the "whisper-asr-webservice" docker image could be added like Plex does under active apps. Thanks.

[LanceM26]

I see bars for a split second when going to dashboard, and then it goes to this:

 

# cat /tmp/gpujson
Time = Tue, 02 Jan 2024 03:22:08 -0800
{"00:02.0":{"clock":"N\/A","fan":"N\/A","memclock":"N\/A","memutil":"N\/A","memused":"N\/A","power":"N\/A","powermax":"N\/A","rxutil":"N\/A","txutil":"N\/A","temp":"N\/A","tempmax":"N\/A","util":"N\/A","error":[{"code":301,"message":"Vendor command returned no data.","extra":""}],"vfio":false,"vfiochk":false,"vfiochkid":"0000:00:02.0","panel":1}}

 

 

diagnostics-20240102-0323.zip

[JamesHaefner]

my GPU Stats window on the dashboard will not stay expanded to show the information. When I try to expand the view to see the specifics, it just shows the header. When I try to expand it when it thinks it is collapsed, the details show for a brief moment then disappear again, even though the window thinks it is still expanded. Any thoughts?

[alturismo]

2 hours ago, JamesHaefner said:

Any thoughts?

reset the GUI settings, i dont use this plugin but had a similar behaviour ...

 

Unraid GUI, may someday it will be ...

[JamesHaefner]

2 minutes ago, alturismo said:

reset the GUI settings

Done that a few times now, still no luck. I was able to see a small line of text one time it opened briefly telling me "GPU not available, bound to VFIO or in use in a VM" which is strange because that GPU is not bound to VFIO or used in a VM. It is passed through for a few dockers to use (Plex and Tdarr) but I have a separate GPU that is passed through for a VM, not the one I am trying to display in the stats.

[alturismo]

6 minutes ago, JamesHaefner said:

Done that a few times now, still no luck.

ok, then it really may be something different  just looks exactly like it did here ... but here its been the drives when i expanded them ... may switch your GPU to show then in the settings from the plugin.

[SimonF]

8 hours ago, JamesHaefner said:

Done that a few times now, still no luck. I was able to see a small line of text one time it opened briefly telling me "GPU not available, bound to VFIO or in use in a VM" which is strange because that GPU is not bound to VFIO or used in a VM. It is passed through for a few dockers to use (Plex and Tdarr) but I have a separate GPU that is passed through for a VM, not the one I am trying to display in the stats.

Can you attach copy of gpujson from them directory.

[SimonF]

On 1/2/2024 at 11:26 AM, LanceM26 said:

I see bars for a split second when going to dashboard, and then it goes to this:

 

# cat /tmp/gpujson
Time = Tue, 02 Jan 2024 03:22:08 -0800
{"00:02.0":{"clock":"N\/A","fan":"N\/A","memclock":"N\/A","memutil":"N\/A","memused":"N\/A","power":"N\/A","powermax":"N\/A","rxutil":"N\/A","txutil":"N\/A","temp":"N\/A","tempmax":"N\/A","util":"N\/A","error":[{"code":301,"message":"Vendor command returned no data.","extra":""}],"vfio":false,"vfiochk":false,"vfiochkid":"0000:00:02.0","panel":1}}

 

 

diagnostics-20240102-0323.zip 147.26 kB · 1 download

There have been a few reports of similar to this due to data not return quickly enough. You can increase the timers here.

 

 

Does this provide any output add the PCI slot details or if you have only single gpu you can remove the -d pcislot=

 

timeout - k .500 .600 intel_gpu_top -J -s 250 -d pci:slot=

[JamesHaefner]

3 hours ago, SimonF said:

Can you attach copy of gpujson from them directory.

Where would I find that file? 

[SimonF]

3 minutes ago, JamesHaefner said:

Where would I find that file? 

sorry for the typo, /tmp/gpujson

[JamesHaefner]

1 hour ago, SimonF said:

sorry for the typo, /tmp/gpujson

  GNU nano 7.2                                                                                                                                                                                                                                                                                                                       gpujson                                                                                                                                                                                                                                                                                                                                  
Time = Sun, 07 Jan 2024 11:32:59 -0500
{"04:00.0":{"clock":"300","fan":"27%","memclock":"405","memutil":"0%","memused":"11","power":"5W","powermax":"N\/A","rxutil":"0","txutil":"0","temp":"34 \u00b0C","tempmax":"96 \u00b0C","util":"0%","vendor":"NVIDIA","name":"GTX 1660 SUPER","clockmax":"2130","memclockmax":"7001","memtotal":"6144","encutil":"0%","decutil":"0%","pciemax":1000,"perfstate":"P8","throttled":"Yes","thrtlrsn":" (gpu_idle)","pciegen":1,"pciegenmax":3,"pciewidth":4,"pciewidthmax":16,"sessions":1,"uuid":"GPU-74a690f4-8a5b-ec41-d40a-39f7687a2940","plexusing":false,"plexmem":0,"plexcount":0,"jellyfinusing":false,"jellyfinmem":0,"jellyfincount":0,"handbrakeusing":false,handbrakemem":0,"handbrakecount":0,"embyusing":false,"embymem":0,"embycount":0,"tdarrusing":false,"tdarrmem":0,"tdarrcount":0,"unmanicusing":false,"unmanicmem":0,"unmaniccount":0,"dizquetvusing":false,"dizquetvmem":0,"dizquetvcount":0,"ersatztvusing":false,"ersatztvmem":0,"ersatztvcount":0,"fileflowsusing":false,"fileflowsmem":0,"fileflowscount":0,"frigateusing":false,"frigatemem":0,"frigatecount":0,"Threadfinusing":false,"Threadfinmem":0,"Threadfincount":0,"codeprojectusing":false,"codeprojectmem":0,"codeprojectcount":0,"deepstackusing":false,"deepstackmem":0,"deepstackcount":0,"nsfminerusing":false,"nsfminermem":0,"nsfminercount":0,"shinobiprousing":false,"shinobipromem":0,"shinobiprocount":0,"foldinghomeusing":false,"foldinghomemem":0,"foldinghomecount":0,"appssupp":["plex","jellyfin","handbrake","emby","tdarr","unmanic","dizquetv","ersatztv","fileflows","frigate","Threadfin","codeproject","deepstack","nsfminer","shinobipro","foldinghome"],"vfio":false,"vfiochk":false,"vfiochkid":"0000:04:00.0","panel":1}}

Here is everything that is in there, right now looks like just a bunch of apps stating they aren't using the GPU, but at the end I see some entries, mainly "vfio":false, which again verifies that the GPU is not bound to vfio and is not used in a VM. if I run nvidia-smi the GPU shows up there with the same idle state details, just won't show properly in the GPU Stats dashboard tile. 

[SimonF]

41 minutes ago, JamesHaefner said:

  GNU nano 7.2                                                                                                                                                                                                                                                                                                                       gpujson                                                                                                                                                                                                                                                                                                                                  
Time = Sun, 07 Jan 2024 11:32:59 -0500
{"04:00.0":{"clock":"300","fan":"27%","memclock":"405","memutil":"0%","memused":"11","power":"5W","powermax":"N\/A","rxutil":"0","txutil":"0","temp":"34 \u00b0C","tempmax":"96 \u00b0C","util":"0%","vendor":"NVIDIA","name":"GTX 1660 SUPER","clockmax":"2130","memclockmax":"7001","memtotal":"6144","encutil":"0%","decutil":"0%","pciemax":1000,"perfstate":"P8","throttled":"Yes","thrtlrsn":" (gpu_idle)","pciegen":1,"pciegenmax":3,"pciewidth":4,"pciewidthmax":16,"sessions":1,"uuid":"GPU-74a690f4-8a5b-ec41-d40a-39f7687a2940","plexusing":false,"plexmem":0,"plexcount":0,"jellyfinusing":false,"jellyfinmem":0,"jellyfincount":0,"handbrakeusing":false,handbrakemem":0,"handbrakecount":0,"embyusing":false,"embymem":0,"embycount":0,"tdarrusing":false,"tdarrmem":0,"tdarrcount":0,"unmanicusing":false,"unmanicmem":0,"unmaniccount":0,"dizquetvusing":false,"dizquetvmem":0,"dizquetvcount":0,"ersatztvusing":false,"ersatztvmem":0,"ersatztvcount":0,"fileflowsusing":false,"fileflowsmem":0,"fileflowscount":0,"frigateusing":false,"frigatemem":0,"frigatecount":0,"Threadfinusing":false,"Threadfinmem":0,"Threadfincount":0,"codeprojectusing":false,"codeprojectmem":0,"codeprojectcount":0,"deepstackusing":false,"deepstackmem":0,"deepstackcount":0,"nsfminerusing":false,"nsfminermem":0,"nsfminercount":0,"shinobiprousing":false,"shinobipromem":0,"shinobiprocount":0,"foldinghomeusing":false,"foldinghomemem":0,"foldinghomecount":0,"appssupp":["plex","jellyfin","handbrake","emby","tdarr","unmanic","dizquetv","ersatztv","fileflows","frigate","Threadfin","codeproject","deepstack","nsfminer","shinobipro","foldinghome"],"vfio":false,"vfiochk":false,"vfiochkid":"0000:04:00.0","panel":1}}

Here is everything that is in there, right now looks like just a bunch of apps stating they aren't using the GPU, but at the end I see some entries, mainly "vfio":false, which again verifies that the GPU is not bound to vfio and is not used in a VM. if I run nvidia-smi the GPU shows up there with the same idle state details, just won't show properly in the GPU Stats dashboard tile. 

Can you provide the output of this nvidia-smi -q -x -g GPU-74a690f4-8a5b-ec41-d40a-39f7687a2940