[PLUGIN] GPU Statistics


Recommended Posts

On 5/12/2021 at 1:36 PM, ich777 said:

Just ignore the message this is just a information from the driver itself when nvidia-smi is called and should only appear if you are on the Dashboard page (you also got at least one such message in your syslog when you open up the Nvidia-Driver plugin page because I also have to call nvidia-smi to get the driver details and also the UUID of the card/s).

Simply ignoring this message isnt that easy when you have to look through your log files and see hundreds if not thousands of these messages.  Its a lot to sort through.  Is there a way to keep this out of the log files at least?

Link to comment
34 minutes ago, danktankk said:

Simply ignoring this message isnt that easy when you have to look through your log files and see hundreds if not thousands of these messages.  Its a lot to sort through.  Is there a way to keep this out of the log files at least?

 

If you use syslog server you can add them to a blocklist.

 

Here's an excerpt from my go file:

# Suppress time capsule cifs errors
echo ":msg,contains,\"bogus file nlink value\" stop" >> /etc/rsyslog.d/01-blocklist.conf
echo ":msg,contains,\"cifs_all_info_to_fattr\" stop" >> /etc/rsyslog.d/01-blocklist.conf
/etc/rc.d/rc.rsyslogd restart

 

  • Thanks 1
Link to comment
On 8/4/2021 at 2:55 PM, CS01-HS said:

 

If you use syslog server you can add them to a blocklist.

 

Here's an excerpt from my go file:


# Suppress time capsule cifs errors
echo ":msg,contains,\"bogus file nlink value\" stop" >> /etc/rsyslog.d/01-blocklist.conf
echo ":msg,contains,\"cifs_all_info_to_fattr\" stop" >> /etc/rsyslog.d/01-blocklist.conf
/etc/rc.d/rc.rsyslogd restart

 

Thats definitely an option, but I think I will just remove the plugin.  Thanks for the reply.

  • Like 1
Link to comment
32 minutes ago, danktankk said:

Thats definitely an option, but I think I will just remove the plugin.  Thanks for the reply.

 

I think you can disable UI Automatic Refresh in the plugin's settings to mostly solve it. Dashboard reporting's less useful than stats to track usage, e.g. through Grafana.

 

1865156904_ScreenShot2021-08-06at3_15_34PM.thumb.png.1babc30faa8d571a22e04a28ac6d0290.png

Link to comment
On 8/6/2021 at 3:23 PM, CS01-HS said:

 

I think you can disable UI Automatic Refresh in the plugin's settings to mostly solve it. Dashboard reporting's less useful than stats to track usage, e.g. through Grafana.

 

1865156904_ScreenShot2021-08-06at3_15_34PM.thumb.png.1babc30faa8d571a22e04a28ac6d0290.png

thanks ill give that a try

Link to comment
46 minutes ago, danktankk said:

thanks ill give that a try

 

If I remember right you need the intel-gpu-telegraf docker to collect the stats in telegraf.

 

EDIT: Now that I think about it this plugin may not be necessary for stats collection, just the docker. Not sure, sorry it's been a while since I set it up. 

 

EDIT 2: Ha, apparently I had my plugins confused. It's the Intel GPU TOP plugin that works with the docker to collect stats.

Edited by CS01-HS
Link to comment
On 8/9/2021 at 4:03 PM, CS01-HS said:

 

If I remember right you need the intel-gpu-telegraf docker to collect the stats in telegraf.

 

EDIT: Now that I think about it this plugin may not be necessary for stats collection, just the docker. Not sure, sorry it's been a while since I set it up. 

 

EDIT 2: Ha, apparently I had my plugins confused. It's the Intel GPU TOP plugin that works with the docker to collect stats.

lol ok thanks for the update :)

  • Haha 1
Link to comment
  • 2 weeks later...
2 hours ago, ds123 said:

IMC Bus Utilization is always active, is this normal?

Well that's the Integrated Memory Controller... Hope that makes it a little bit clearer what it is and what it does.

 

2 hours ago, ds123 said:

Power Draw also seems to be "full" even though 0W is consumed

That's a bug in the intel_gpu_top executable and nothing that could be easily solved for now...

Can you share your diagnostics so that I can send it to the developers from the intel_gpu_tools so that they can take a look at it?

Link to comment
13 hours ago, ich777 said:

Well that's the Integrated Memory Controller... Hope that makes it a little bit clearer what it is and what it does.

 

That's a bug in the intel_gpu_top executable and nothing that could be easily solved for now...

Can you share your diagnostics so that I can send it to the developers from the intel_gpu_tools so that they can take a look at it?

 

Sure, thanks.

 

tower-diagnostics-20210821-1129.zip

  • Like 1
Link to comment
  • 2 weeks later...

It is normal to see a constant higher CPU usage when using this plugin on a nvidia card? Because with this plugin enabled I see on htop regularly cpu usage of 20% on nvidia-smi. After disabling this plugin I see the CPU usage back to almost zero.

My system specs:

i5-6600 with Asus P10S-M WS

GTX1650

nvidia driver 470.63.01

 

EDIT:

After starting plex and the GPU finally went to idle mode when nothing running and the CPU usage went down.

 

EDIT2:

When the GTX1650 went back to P0 state, the CPU usage rises again. I disabled the automatic refresh for now.

Edited by bergi9
Link to comment
On 9/2/2021 at 6:49 PM, shimi269 said:

GTX 760

18 minutes ago, QuasimodosHump said:

I have the same problem on my recent install of the GPU Stats app.

Both of you, can you please open up a terminal from unRAID and issue the command 'nvidia-smi' (without quotes) and post the output as a screenshot here?

 

The main problem is that this two cards are really old by nowadays standards and it is very common that these cards report not much anymore through nvidia-smi, this is a common problem with older Nvidia cards and there is nothing @b3rs3rk or I can do about this.

 

May I ask for what both of you are using the card? Also keep in mind that Nvidia dropped support for 6xx & 7xx series cards and will not be supported with the next driver release (Nvidia driver 470.xx.xx is the last one that supports those cards).

Link to comment
On 9/6/2021 at 3:01 PM, bergi9 said:

It is normal to see a constant higher CPU usage when using this plugin on a nvidia card? Because with this plugin enabled I see on htop regularly cpu usage of 20% on nvidia-smi.

Should only happen when you are on the Dashboard in unRAID or am I wrong?

 

On 9/6/2021 at 3:01 PM, bergi9 said:

When the GTX1650 went back to P0 state, the CPU usage rises again. I disabled the automatic refresh for now.

This is very unlikely to be caused by the plugin, also consider enabling nvidia-persistenced mode.

  • Thanks 1
Link to comment
  • 4 weeks later...

Hello and thanks for this Plugin!

I want to import the Nvidia Data into influxDB with telegraf. But some variables has their Units shown, like Power(W) or temperature, so telegraf cant read them as values.

Would it be possible to provide a clean Version with only ints/floats (where if it is possible) without any Unit?

As a workaround for now, I changed the /lib/Nvidia.php and removed them, but that's nasty and only works till the next update.

 

I'm also wondering why the plugin only shows the Nvidia data. Besides an Nvidia 2070 I'm also having a Radeon RX 580 in the unraid system and radeontop runs fine.

Is there a way to get the data from both GPU as JSON output?

Link to comment
7 hours ago, corgan said:
Hello and thanks for this Plugin!
I want to import the Nvidia Data into influxDB with telegraf. But some variables has their Units shown, like Power(W) or temperature, so telegraf cant read them as values.
Would it be possible to provide a clean Version with only ints/floats (where if it is possible) without any Unit?
As a workaround for now, I changed the /lib/Nvidia.php and removed them, but that's nasty and only works till the next update.
 
I'm also wondering why the plugin only shows the Nvidia data. Besides an Nvidia 2070 I'm also having a Radeon RX 580 in the unraid system and radeontop runs fine.
Is there a way to get the data from both GPU as JSON output?

Would it be also possible to use Prometheus?
I created a plugin for that that you can fond in the CA App.

If you want to do this manually you have to create a script that exports the data from nvidia-smi and/or radeontop.

 

Nvidia:

nvidia-smi -q -x

(this will give you information in XML - JSON is not available)

 

AMD:

radeontop -d - -l 1

(this should give you informtion about the GPU but I'm not 100% sure because I don't own any AMD APUs/GPUs)

Sent from my C64

 

Link to comment
On 9/27/2021 at 5:45 AM, spacezmonkey said:

Hello, After installing, dashboard show the fan % , but encoding and decoding show NA... and the load show NA as well... may i know is there anything i have done wrongly? 

 

image.png.2ba090994c861855d9784fd6a1adf232.png

 

image.png.cbc59a31679284e0dda2f8a950bcd5b7.png

image.png.4ec508a457709916790eccbf6d80dc84.png

The same as in the the post that I've linked (a few posts above) applies to your card:

 

Link to comment
On 9/27/2021 at 12:42 AM, mihcox said:

Im having issues with my p2200. It works and transcodes as excpected, but no statistics load on the homepage. NVIDIA-SMI still works in the command line though.

 

Any ideas?

Have you selected the right card on the plugin settings page?

Link to comment
2 hours ago, ich777 said:

Would it be also possible to use Prometheus?
I created a plugin for that that you can fond in the CA App.

If you want to do this manually you have to create a script that exports the data from nvidia-smi and/or radeontop.

 

Nvidia:

nvidia-smi -q -x

(this will give you information in XML - JSON is not available)

 

AMD:

radeontop -d - -l 1

(this should give you informtion about the GPU but I'm not 100% sure because I don't own any AMD APUs/GPUs)

Sent from my C64

 

 

Prometheus is an option. Thanks for the tip.

But you only have this for Nvidia? Maybe I can create an AMD version of the Prometheus exporter.

Link to comment

Hello again,

 

I just made a PR on Github, in which I added support for two more App, Shinobi and nfsminer.

Sadly the third app is ffmpeg coming from frigate, but there is no clear identifier to be sure that this ffmpeg binary comes from frigate.

 

But this PR was "by accident", because mainly I was troubleshooting why my AMD Card don't get shown.

On Paper the sensor data are there. They also get exported via telegrafs sensor plugin.

But on the unraid dashboard, the panel is empty if I select AMD Vendor in the Settings.

Radeontop is installed and running.

 

Any Ideas?

 

 

grafik.thumb.png.30e5d07165fa8e3ed7899c2bec081a62.png

 

grafik.png.6cbc646d5e262886b795f4eb3378c837.png

 

 

grafik.png.a5afa87feef96fe4d2be563715c0e5cd.png

 

 

 

 

grafik.thumb.png.90dcf088f0f2c84ffcfb8964002e7239.png

 

 

 

grafik.png

grafik.png

Link to comment

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.