[Support] ich777 - AMD Vendor Reset, CoralTPU, hpsahba,...

[ich777]

6 hours ago, Jarao said:

Thanks for your reply! So there have any optimze suggestion?

No, because it seems to me that nothing is wrong with the transcoding.

Can you maybe share more information about the file that you are transcoding and to what device you are streaming (mobile phone, PC, browser, native app,...).

 

If you have a audio stream that uses Dolby or something that the client doesn't support your server have to convert it too and this is usually the high usage on your CPU.

What you can try is that you let the movie play for about 10-20 minutes and see after that 10-20 minutes if the usage is still that high <- transcoding from the audio stream should be done by that.

 

You can also try a file which is 720p h264 and has only stereo audio, this file should not be transcoded at all.

[Jarao]

@ich777

Here is the video playback info. My client is Edge html client. I found the cpu usage will still keep high until the video stop. Only this video take so much cpu resource

 

[ich777]

38 minutes ago, Jarao said:

I found the cpu usage will still keep high until the video stop.

I really don‘t know whats going on then… 😅

Is something else using the CPU?

[Jarao]

10 hours ago, ich777 said:

I really don‘t know whats going on then… 😅

Is something else using the CPU?

No. I top the shell and ffmpeg keep the cpu usage.

When I pause the video, the usage get down, but when I play, usage come back. Maybe my pause time not enough?

[ich777]

3 hours ago, Jarao said:

No. I top the shell and ffmpeg keep the cpu usage.

When I pause the video, the usage get down, but when I play, usage come back. Maybe my pause time not enough?

This is really strange, Jellyfin usually should release the CPU after ending the playback however pausing the movie should not stop transcoding.

 

I would recommend that you post over on the Jellyfin GitHub Issue tracker about that: Click

[Jarao]

9 hours ago, ich777 said:

This is really strange, Jellyfin usually should release the CPU after ending the playback however pausing the movie should not stop transcoding.

 

I would recommend that you post over on the Jellyfin GitHub Issue tracker about that: Click

Ok, Thanks for your help  

[Lightarrow]

I just wanted to chime in here and thank ich777 for his work on this AMD Vendor reset bug. I recently got a second hand 5600XT for my build (already have a 3080) and got hit with this reset bug. After some research, I installed this plugin and it did not help for the first few tries. What ended up working was not passing the Sound card along with the GPU to my Linux VM and then I could start / stop the VM without getting the reset bug. So thank you for this ich777, I really appreciate it.

[inogueira]

Hi all! 

I'm noob with unraid and now I have a HP Proliant DL380 G7. It has a P401i SAS raid controller that doesn't support HBA. 

I've read a lot about hpsahba and patching kernel but now I've found this plugin. 

 

Does this plugin all the work? I mean changing controller mode to HBA and pat hing the kernel? Will i need to do something else? 

 

Thanks in advance. 

[ich777]

8 hours ago, inogueira said:

Does this plugin all the work? I mean changing controller mode to HBA and pat hing the kernel? Will i need to do something else? 

In general, yes and no.

I would recommend that you the readme from the repository which this plugin is based on over here.

 

You have to test if your specific controller is supported and is working with it.

 

Please read the readme carefully because it is also possible that you destroy your RAID controller or your data with this tool.

[inogueira]

On 12/19/2022 at 8:09 AM, ich777 said:

In general, yes and no.

I would recommend that you the readme from the repository which this plugin is based on over here.

 

You have to test if your specific controller is supported and is working with it.

 

Please read the readme carefully because it is also possible that you destroy your RAID controller or your data with this tool.

Thanks!

 

My controller is compatible and in the needed firmware vresion so I activated HBA mode. hpsahba -i shows that hba is activated.

The next step... patch kernel. How must I do it?

Thanks!

[ich777]

1 minute ago, inogueira said:

The next step... patch kernel. How must I do it?

You don't need to patch the Kernel, this is what the plugin does.

Or what do you mean exactly?

[inogueira]

3 minutes ago, ich777 said:

You don't need to patch the Kernel, this is what the plugin does.

Or what do you mean exactly?

I've installed th eplugin and then activated HBA mode on the controller, but I cannot see the siks yet... I'll give it a reboot

[ich777]

3 minutes ago, inogueira said:

I've installed th eplugin and then activated HBA mode on the controller, but I cannot see the siks yet... I'll give it a reboot

You have to reboot after installing the plugin for the first time.

[inogueira]

10 minutes ago, ich777 said:

You have to reboot after installing the plugin for the first time.

yes, that was the problem. Thank you so much! nice work!

[Skitals]

Re: AMD Vendor Reset

 

I have been using this wonderful plugin since it was available, and was building a custom kernel with gnif's patch before that. I've never had any issues with my 5700 XT until yesterday. It seems like it crashed hard. At first it was stalling at the Tianocore splash screen, after some reboots and troubleshooting it would boot but the GPU had "stopped working" with Code 43 (low res output, drivers wouldn't load). I tried everything under the sun to fix the code 43, but no beans.

 

The one and only fix was to boot windows on bare metal (Windows is installed on an NVME drive I pass to the VM). The GPU/AMD driver immediately worked, I rebooted into unraid and it immediately worked in the VM again.

 

My best guess is the GPU was in some unrecoverable state. The only thing that fixed it was booting it natively. A word of advice if anyone finds themselves in a similar situation where the card suddenly stops working with no change.

[yanakis]

Hi. While testing transcoding I see this strange behavior with "intel_gpu_top'" and similar in the Dashboard plugin: usage is shown for a second then dissapears and goes to zero for longer time and then shows some usage and so on. Plex dashboard shows HW transcoding, cpu usage quite low, from 2 to 14% max.

 

What going on? :D

[ich777]

1 hour ago, yanakis said:

What going on?

Your iGPU is transcoding…

 

What is wrong or what do you think isn‘t working?

In Plex throttling is enabled by default and this causes such behavior, but pleasr let it enabled, it‘s enabled for a reason…

[yanakis]

2 hours ago, ich777 said:

Your iGPU is transcoding…

 

What is wrong or what do you think isn‘t working?

In Plex throttling is enabled by default and this causes such behavior, but pleasr let it enabled, it‘s enabled for a reason…

 

I know is transcoding, that's why I have this setup but to me it seems it does not transcode in continuous mode...again, both GPU dasboard plugin and intel top gpu app shows usage for 1-2 seconds then all parameters go to 0 then after 5-6 seconds gpu usage appears again then go to zero...

 

Can I upload a video in here?

Edited December 28, 2022 by yanakis
both

[ich777]

9 minutes ago, yanakis said:

I know is transcoding, that's why I have this setup but to me it seems it does not transcode in continuous mode...

Yes, because as I wrote, throttlig is enabled in Plex.

 

BTW that is not an issue at all and this question would be better suited in the Plex support thread, but this is default behavior.

 

11 minutes ago, yanakis said:

shows usage for 1-2 seconds then all parameters go to 0 then after 5-6 seconds gpu usage appears again then go to zero...

As said above, expected behavior.

[alturismo]

16 minutes ago, yanakis said:

Can I upload a video in here?

i dont think anyone needs a video here from a working and normal behaviour

 

may to give you an idea about Plex, its working in chunks and not transcoding the whole video in one run to the end ... so a "heartbeat" behaviour is normal, if you want to see another behaviour, install ffmpeg, read how to use it and you will see a steady running 100% encode process

 

also, i guess you are in the wrong thread overall ?

[yanakis]

5 minutes ago, alturismo said:

i dont think anyone needs a video here from a working and normal behaviour

 

may to give you an idea about Plex, its working in chunks and not transcoding the whole video in one run to the end ... so a "heartbeat" behaviour is normal, if you want to see another behaviour, install ffmpeg, read how to use it and you will see a steady running 100% encode process

 

also, i guess you are in the wrong thread overall ?

Ok I see :), I got the same "heart beat" profile, must be fine after all.

 

Well, about the thread: I just pressed the "support" link from "Intel GPU TOP" plugin because I started from its "issue" :).

 

Sorry if I am in the wrong thread and many thanks for your clarifications.

Edited December 28, 2022 by yanakis

[ich777]

Just now, yanakis said:

Sorry if I am in the wrong thread and many thanks for your clarifications.

But why should this be an issue with intel_gpu_top? Your issue is related to Plex…

[yanakis]

53 minutes ago, ich777 said:

But why should this be an issue with intel_gpu_top? Your issue is related to Plex…

well....I had no idea, hence I am here :).

[ich777]

1 hour ago, yanakis said:

well....I had no idea, hence I am here :).

I can only tell you that intel_gpu_top is working since it reports something on your Dashboard and also in your terminal, this would indicate that the next thing to do is ask on the Plex support thread.

[gregzee]

I have a TS-470+. I moved over to unraid and am trying to get fan controll to work on unraid the way it worked on QTS. I am not sure how to get the QNAP-EC plugin to work on unraid.

When I install the plugin -- I get this

It says it failed loading the module. 

Is there any other steps I need to do, to get this working?

 

plugin: installing: qnap-ec.plg
Executing hook script: pre_plugin_checks
plugin: downloading: qnap-ec.plg ... done

Executing hook script: pre_plugin_checks

-----------------------Downloading QNAP-EC package!-------------------------
--------This could take some time, please don't close this window!----------

-------------------Sucessfully downloaded QNAP-EC package-------------------
-Error loading module, please post your Diagnostics on the support thread!-

---------------Installation of QNAP-EC package successful-------------------

plugin: qnap-ec.plg installed
Executing hook script: post_plugin_checks

 

 

Diag ran not sure if this is the right thing to do but posting it.

 

xxxxx-diagnostics-20221231-1258/logs/syslog.txt:
 ....
 2246  Dec 31 09:34:24 xxxxx  emhttpd: cmd: /usr/local/emhttp/plugins/dynamix/scripts/share_size isos ssz1 cache,cachetwo
 2247  Dec 31 09:34:24 xxxxx  emhttpd: cmd: /usr/local/emhttp/plugins/dynamix/scripts/share_size system ssz1 cache,cachetwo
 2248: Dec 31 09:34:40 xxxxx root: plugin: creating: /usr/local/emhttp/plugins/qnap-ec/README.md - from INLINE content
 2249  Dec 31 09:34:40 xxxxx root: plugin: running: anonymous
 2250: Dec 31 09:34:43 xxxxx root: plugin: qnap-ec.plg installed
 2251  Dec 31 09:39:12 xxxxx kernel: docker0: port 1(veth9d1d4a7) entered blocking state
 2252  Dec 31 09:39:12 xxxxx kernel: docker0: port 1(veth9d1d4a7) entered disabled state
 ....
 2259  Dec 31 09:39:12 xxxxx kernel: docker0: port 1(veth9d1d4a7) entered disabled state
 2260  Dec 31 09:42:38 xxxxx root: plugin: running: anonymous
 2261: Dec 31 09:42:38 xxxxx root: plugin: qnap-ec.plg removed
 2262: Dec 31 09:43:02 xxxxx root: plugin: creating: /usr/local/emhttp/plugins/qnap-ec/README.md - from INLINE content
 2263  Dec 31 09:43:02 xxxxx root: plugin: running: anonymous
 2264: Dec 31 09:43:03 xxxxx root: plugin: qnap-ec.plg installed
 2265  

 

Edited December 31, 2022 by gregzee