[Support] binhex - Plex


Recommended Posts

10 minutes ago, binhex said:

ok give this is a go, its an AUR package so needs compiling, let it finish then reboot:-

export aur_packages=intel-opencl-runtime && aur.sh

 

Alright, that generated a ton of lines in the console (again, can paste if you want), but it finished without an error afaik and I restarted the container. Transcode speed on the 4K HDR file is still 0.7 however, so no luck.

Link to comment
8 minutes ago, thatsthefrickenlightning said:

Alright, that generated a ton of lines in the console (again, can paste if you want), but it finished without an error afaik and I restarted the container. Transcode speed on the 4K HDR file is still 0.7 however, so no luck.

Is your igpu Intel IvyBridge or Haswell iGPUs?.

Link to comment
15 minutes ago, binhex said:

Is your igpu Intel IvyBridge or Haswell iGPUs?.

Are those the two options? It's an i3-7100, so if those are the two options, I'd go with Haswell, as that's newer, but it's still older than my gen. The generational name of my CPU is Kaby Lake, the GPU is an HD 630, the QSV version is (according to wiki) 6.

 

Hope that helps.

Edited by thatsthefrickenlightning
Link to comment
3 hours ago, thatsthefrickenlightning said:

Are those the two options? It's an i3-7100, so if those are the two options, I'd go with Haswell, as that's newer, but it's still older than my gen. The generational name of my CPU is Kaby Lake, the GPU is an HD 630, the QSV version is (according to wiki) 6.

 

Hope that helps.

@thatsthefrickenlightning

Ivy bridge is gen 3, haswell is gen 4. They're old and lack features the newer chips/igpus have. I'm assuming Binhex wanted to confirm you weren't using a less capable chip that may not support the required features or may do so in an non-ideal manner.

 

  • Your chip (7100) is gen 7 (KabyLake) with the HD630 igpu
  • Plex's post about HDR mapping states "Plex Media Servers using Intel Coffee Lake or newer processors on Ubuntu require version 20.04 or later". Coffee lake is gen 8 and contains the UHD630 gpu
  • For reference, my chip (Xeon E-2176G) is Gen 8 (Coffee Lake) with the HD P630 igpu

That said, I'm not sure if Plex's comment is hard requirement or maybe something specific to Ubuntu. If you look at the LinuxServer support thread you can see Kaby lake users reporting success when they implemented the change in their container last December. The (7th gen) HD630 and (8th gen) UHD630 gpus were very similar in features, I think the primary difference being a minor clock speed bump (50mhz or so). 

 

I'll test the console commands later today on my system. Before doing so I want to document the transcoding speed (as you were) so that I can confirm that not only is HDR tone mapping occurring, but that it's being handled by hardware and shows that same performance increase you noted in the LinuxServer container.

 

Cheers! 

 

 

Edited by SloppyG
Link to comment
3 hours ago, SloppyG said:

@thatsthefrickenlightning

Ivy bridge is gen 3, haswell is gen 4. They're old and lack features the newer chips/igpus have. I'm assuming Binhex wanted to confirm you weren't using a less capable chip that may not support the required features or may do so in an non-ideal manner.

 

  • Your chip (7100) is gen 7 (KabyLake) with the HD630 igpu
  • Plex's post about HDR mapping states "Plex Media Servers using Intel Coffee Lake or newer processors on Ubuntu require version 20.04 or later". Coffee lake is gen 8 and contains the UHD630 gpu
  • For reference, my chip (Xeon E-2176G) is Gen 8 (Coffee Lake) with the HD P630 igpu

That said, I'm not sure if Plex's comment is hard requirement or maybe something specific to Ubuntu. If you look at the LinuxServer support thread you can see Kaby lake users reporting success when they implemented the change in their container last December. The (7th gen) HD630 and (8th gen) UHD630 gpus were very similar in features, I think the primary difference being a minor clock speed bump (50mhz or so). 

 

I'll test the console commands later today on my system. Before doing so I want to document the transcoding speed (as you were) so that I can confirm that not only is HDR tone mapping occurring, but that it's being handled by hardware and shows that same performance increase you noted in the LinuxServer container.

 

Cheers! 

 

 

I see, I misunderstood the question a bit then - thanks. Luckily, I still gave the full and correct info.

 

I read Plex's requirement and am suspecting that Coffee Lake is not the minimum and they just didn't formulate that very well (and that it's a Ubuntu-specific minimum, for whatever reason). This suspicion is supported by the Wiki page for QSV stating Kaby, Coffee, Whiskey and Comet Lake all have the same capabilities, QSV wise. EDIT: no, that's not the case, but their differences seem to be few and are not based around encoding or decoding h.265 video.

 

Looking for ward to more from you two. Thanks.

 

Edit 2: found this post from a Plex moderator on their forum. The table towards the bottom of his post lists the i7-7700K as capable of 'hardware transcoding h265/HEVC (8/10bit → SDR/HDR)'. My i3 is of the same generation so it should be able to do this without issue, and in my view it does do hw tone mapping without issue on LSIO's container.

Edited by thatsthefrickenlightning
Link to comment
On 10/29/2021 at 7:22 AM, binhex said:

ok give this is a go, its an AUR package so needs compiling, let it finish then reboot:-

export aur_packages=intel-opencl-runtime && aur.sh

 

Hey @binhex, quick update.

 

so I tried both the first post's suggestions:

Quote

pacman -Syy && pacman -S ocl-icd --noconfirm

pacman -Syy && pacman -S intel-compute-runtime --noconfirm

 

and

Quote

export aur_packages=intel-opencl-runtime && aur.sh

(which took a little while but I didn't notice any errors.

 

I rebooted the container between each and also my unraid server entirely at the end. My testing up until now had just been 1080p HDR files, so I wanted to better match what @thatsthefrickenlightning was doing by throwing a 4k HDR file at it. I connected by iPad to a vpn (to send the traffic outside of the local network) and forced transcoding to 1080p 10mbps (as it was happily playing via direct play till I did so.

 

At this point I could confirm that transcoding was occurring for both audio and video. I also noticed that within Plex is looked like my CPU usage had gone up from +/- 40% (it bounces around of course) to 50-60%. I also see changes on the Unraid dashboard, attached is a screenshot of that. On the right is my baseline screenshot which is the "vanilla" plex container.  Cpu usage roughly matches what plex reports and you can see that GPU usage, at least as reported via the "Intel GPU TOP" plugin has disappeared. This absence of gpu usage is similar to if I remove "/dev/dri" device from the container. I did reboot in case something about the package change required a full reboot.

 

My system can keep up with HDR in realtime (probably due to the number of cores, it's a decent chip) but it does appear those commands seemed to have removed whatever gpu offloading was already occurring for me. 

 

I did a quick pacman search to 'prove' the packages are installed now:

sh-5.1# pacman -Ss ocl-icd
extra/ocl-icd 2.3.1-1 [installed]
    OpenCL ICD Bindings

sh-5.1# pacman -Ss intel-compute
community/intel-compute-runtime 21.21.19914-1 [installed]
    Intel(R) Graphics Compute Runtime for oneAPI Level Zero and OpenCL(TM) Driver


sh-5.1# pacman -Ss opencl
extra/libclc 12.0.1-1
    Library requirements of the OpenCL C programming language
extra/ocl-icd 2.3.1-1 [installed]
    OpenCL ICD Bindings
extra/opencl-clhpp 2.0.15-1
    OpenCL C++ header files
extra/opencl-headers 2:2021.04.29-1
    OpenCL (Open Computing Language) header files
extra/opencl-mesa 21.2.1-1
    OpenCL support for AMD/ATI Radeon mesa drivers
extra/opencl-nvidia 470.63.01-1
    OpenCL implemention for NVIDIA
community/clinfo 3.0.21.02.21-1
    Simple OpenCL application that enumerates all available platform and device properties
community/clpeak 1.1.0-1
    A tool which profiles OpenCL devices to find their peak capacities
community/erlang-cl 1.2.4-3
    OpenCL binding for Erlang
community/intel-compute-runtime 21.21.19914-1 [installed]
    Intel(R) Graphics Compute Runtime for oneAPI Level Zero and OpenCL(TM) Driver
community/intel-graphics-compiler 1:1.0.7423-2 [installed]
    Intel Graphics Compiler for OpenCL
community/intel-opencl-clang 12.0.0-1 [installed]
    Wrapper library around clang that can compile OpenCL C kernels to SPIR-V modules
community/pocl 1.7-2
    Portable OpenCL is an open-source implementation of OpenCL which can be easily adapted for new targets
community/pyopencl-headers 1:2021.1.4-1
    A complete, object-oriented language binding of OpenCL to Python
community/python-pyopencl 1:2021.1.4-1
    A complete, object-oriented language binding of OpenCL to Python

 

 

BeforeAndAfter_UnraidDashboard.png

Link to comment
26 minutes ago, SloppyG said:

At this point I could confirm that transcoding was occurring for both audio and video.

so just to be clear, are you saying that after those packages are installed that hw tone mapping is taking place for your igpu?, if so im not sure why this is not the case for @thatsthefrickenlightning im assuming you guys have different cpu's.

 

edit - ahh i see reading your screenshots the wrong way around, so applying those packages makes things worse not better right?, hmm. ok this is getting a little painful, im basically shooting in the dark here guys, feel free to take a look through Arch repo and Arch User repo (AUR) and try installing any other packages, if you happen against the magic sauce then let me know.

Link to comment
22 minutes ago, binhex said:

so just to be clear, are you saying that after those packages are installed that hw tone mapping is taking place for your igpu?, if so im not sure why this is not the case for @thatsthefrickenlightning im assuming you guys have different cpu's.

 

edit - ahh i see reading your screenshots the wrong way around, so applying those packages makes things worse not better right?, hmm. ok this is getting a little painful, im basically shooting in the dark here guys, feel free to take a look through Arch repo and Arch User repo (AUR) and try installing any other packages, if you happen against the magic sauce then let me know.

 

@binhex Correct! yes sorry I should have flipped those images hah :)

 

Yes I'll dig around. I BELIEVE that i'm seeing at least partial acceleration but (based on @thatsthefrickenlightning's reporting) with your container as delivered but I think that it's not utilizing all the hw acceleration capabilities for intel gpus.  I'll keep digging around the threads 

 

 

Question to help me troubleshoot:

https://forums.plex.tv/t/package-beignet-required-for-hardware-transcoding-of-hdr-content-hdr-tonemapping-now-supported/655048/26

If you look at that post, when Plex is installed/updated it looks like it tells you if the 2 packages it needs are found? Is there a command line I can run in your container to force plex to update/install over itself? I don't THINK the install log is saved to a file or at least I couldn't find the location. I was hoping that (as shown in the post above) the install would identify which (if any) of the intel packages are found or if any are missing. I THINK the beignet package is the one it can't find.

 

I don't want to pull the container again as I think I'll  lose the changes we made and I wasn't sure of the command in Arch to do so. 

 

I'll also install the linuxserver container as a comparison against by baselines as it does appear people are having success with mapping and acceleration with intel igpu's all the way up to gen 10.

 

 

Link to comment

Hi @binhex @thatsthefrickenlightning !

 

Just a quick post, wanted to confirm that linuxserver's container shows much lower cpu usage and much improved igpu utilization with the same video file I was testing on binhex-plex.

 

This test was done:

- with the same 4k HDR 20gb video file I was testing in the binhex-plex container in the post above where I shared unraid dashboard screenshots

- forcing transcoding to 1080p 10mbps

- Transcoding is occurring in Ram (/tmp)

 

During heavy transcoding activity (filling up the 60 second buffer):

- Unraid dashboard now shows cpu usage is around 10% with linuxserver-plex, down from 40% with binhex-plex

- "Intel-GPU-TOP" plugin on the dashboard reports GPU video load of 70-80% with linuxserver-plex, up from 30% with binhex-plex

 

Once playback has buffered and transcoding is just "topping off":

- Unraid dashboard now shows cpu usage mostly idle (under 10%) with linuxserver-plex

- GPU video load has short spikes every 5 seconds to 20 or 50%  with linuxserver-plex

 

So far I'm really impressed with the performance and quality of the tone mapping being done for about 8 watts of power. I'm probably going to stay on this container for a little and see if I run into any issues. 

 

I will keep my eyes out for any more information about what the team that supports the LS container may have done to get this working. I have a feeling there are also some changes made in the intel compute or beignet packages themselves, maybe this didn't get carried over to the Arch Linux versions? I think this may have improved support for older/less common device ID's as intel's GPUs didn't change a great deal for a number of years so chips like Kaby Lake could be made to support it as well (vs Coffee Lake and above). 

 

Linuxserver-PLEX_UnraidGpuCpueUsge.png

Link to comment

Perhaps the packages simply aren't available for Arch yet, or they're not in the right working condition. With LSIO's container running on Ubuntu (if I'm not mistaken), that's a different ball game. Maybe @linuxserver.io can weigh in on this: are there Arch equivalents of ocl-icd-libopencl1 and beignet-opencl-icd available and are they working? We're having some trouble with this over here (no HW tone mapping on Intel). Thanks much.

Edited by thatsthefrickenlightning
Link to comment
On 10/28/2021 at 1:26 AM, thatsthefrickenlightning said:

I bit the bullet late last night and installed the linuxserver PMS container to see if I could garner more info. Lo and behold, after giving that container access to the Intel iGPU, I could transcode three whole streams of my 4K HDR file to 1080p simultaneously with a speed of ~1.5.  How much ram does Plex have available for you, Arturo? Because while it's performing much better, you're still doing better than that.

That's great news.  I have 32gb ram on the system and 10gb Ramdisk dedicated for transcode.

  • Thanks 1
Link to comment

Hi all,

 

So i have an issue where movies will only play if I use the shuffle option. When I try to manually select a movie the loading wheel spins for a bit until the movie fails and I get the message "Couldn't retrieve the play queue for this item. Please try again."

 

I've tried restarting the container and even did a completely fresh install from scratch but still ran into this issue. All other media such as TV, 4k movies, and music play without issue. Also checked the logs but didn't see anything indicating what the problem is.

Link to comment
12 hours ago, jayephx said:

Does anyone know if there has been any progress on HDR transcoding with Rocket Lake iGPU?

You'd need to read the last ~2 pages to get up to speed on that, but in a nutshell: no HW tone mapping on Intel in this container at the moment. LSIO's Plex Server works for this. Not sure about your specific Intel gen, but in general, it works there.

Link to comment
On 7/1/2018 at 5:22 AM, binhex said:

One thought, does your new router have IP spoofing protection? this might be triggered by the fact docker uses its own network range, try turning off any lan firewalling.

Sent from my SM-G935F using Tapatalk
 

I got the reason, you must use ***.local to access plex and you can claim it. Do not you IP address!!!

Link to comment

I love my binhex-Plex, but it stop opening the GUI. Below is the error, related to the xml.Any ideas? Thanks 

 

https://hub.docker.com/u/binhex/


2021-11-17 13:49:27.267749 [info] Host is running unRAID
2021-11-17 13:49:27.322172 [info] System information Linux Galaxy 5.10.28-Unraid #1 SMP Wed Apr 7 08:23:18 PDT 2021 x86_64 GNU/Linux
2021-11-17 13:49:27.389044 [info] OS_ARCH defined as 'x86-64'
2021-11-17 13:49:27.444993 [info] PUID defined as '99'
2021-11-17 13:49:27.575445 [info] PGID defined as '100'
2021-11-17 13:49:27.734656 [info] UMASK defined as '000'
2021-11-17 13:49:27.801616 [info] Permissions already set for '/config'
2021-11-17 13:49:27.876916 [info] Deleting files in /tmp (non recursive)...
2021-11-17 13:49:27.952235 [info] TRANS_DIR defined as '/tmp'
2021-11-17 13:49:28.139885 [info] Starting Supervisor...
2021-11-17 13:49:28,807 INFO Included extra file "/etc/supervisor/conf.d/plexmediaserver.conf" during parsing
2021-11-17 13:49:28,807 INFO Set uid to user 0 succeeded
2021-11-17 13:49:28,814 INFO supervisord started with pid 7
2021-11-17 13:49:29,817 INFO spawned: 'plexmediaserver' with pid 71
2021-11-17 13:49:29,820 INFO spawned: 'shutdown-script' with pid 72
2021-11-17 13:49:29,820 INFO reaped unknown pid 8 (exit status 0)
2021-11-17 13:49:30,154 DEBG 'plexmediaserver' stderr output:
Failed to load preferences at /config/Plex Media Server/Preferences.xml

2021-11-17 13:49:30,155 INFO success: shutdown-script entered RUNNING state, process has stayed up for > than 0 seconds (startsecs)
2021-11-17 13:49:31,156 INFO success: plexmediaserver entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2021-11-17 13:49:32,154 DEBG 'plexmediaserver' stderr output:
Failed to load preferences at /config/Plex Media Server/Preferences.xml

Link to comment
22 hours ago, Profezor said:

I love my binhex-Plex, but it stop opening the GUI. Below is the error, related to the xml.Any ideas? Thanks 

 

https://hub.docker.com/u/binhex/


2021-11-17 13:49:27.267749 [info] Host is running unRAID
2021-11-17 13:49:27.322172 [info] System information Linux Galaxy 5.10.28-Unraid #1 SMP Wed Apr 7 08:23:18 PDT 2021 x86_64 GNU/Linux
2021-11-17 13:49:27.389044 [info] OS_ARCH defined as 'x86-64'
2021-11-17 13:49:27.444993 [info] PUID defined as '99'
2021-11-17 13:49:27.575445 [info] PGID defined as '100'
2021-11-17 13:49:27.734656 [info] UMASK defined as '000'
2021-11-17 13:49:27.801616 [info] Permissions already set for '/config'
2021-11-17 13:49:27.876916 [info] Deleting files in /tmp (non recursive)...
2021-11-17 13:49:27.952235 [info] TRANS_DIR defined as '/tmp'
2021-11-17 13:49:28.139885 [info] Starting Supervisor...
2021-11-17 13:49:28,807 INFO Included extra file "/etc/supervisor/conf.d/plexmediaserver.conf" during parsing
2021-11-17 13:49:28,807 INFO Set uid to user 0 succeeded
2021-11-17 13:49:28,814 INFO supervisord started with pid 7
2021-11-17 13:49:29,817 INFO spawned: 'plexmediaserver' with pid 71
2021-11-17 13:49:29,820 INFO spawned: 'shutdown-script' with pid 72
2021-11-17 13:49:29,820 INFO reaped unknown pid 8 (exit status 0)
2021-11-17 13:49:30,154 DEBG 'plexmediaserver' stderr output:
Failed to load preferences at /config/Plex Media Server/Preferences.xml

2021-11-17 13:49:30,155 INFO success: shutdown-script entered RUNNING state, process has stayed up for > than 0 seconds (startsecs)
2021-11-17 13:49:31,156 INFO success: plexmediaserver entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2021-11-17 13:49:32,154 DEBG 'plexmediaserver' stderr output:
Failed to load preferences at /config/Plex Media Server/Preferences.xml

 

I was seeing the same. This Plex release takes ages on first boot as it has to do a DB migration afaik. Leave it alone for a bit and it should come back up. See: https://forums.plex.tv/t/plex-media-server/30447/467 "Database changes may cause longer than normal start-up times after initial installation of this update, especially on larger libraries. Please be patient and do not interrupt this process."

Link to comment
11 hours ago, lphmedia said:

 

I was seeing the same. This Plex release takes ages on first boot as it has to do a DB migration afaik. Leave it alone for a bit and it should come back up. See: https://forums.plex.tv/t/plex-media-server/30447/467 "Database changes may cause longer than normal start-up times after initial installation of this update, especially on larger libraries. Please be patient and do not interrupt this process."

I switched to Plex-Pass and turnoff the the old one. Works fine now. Although I am not sure of the advantages of Plex-pass yet other than it works.

Link to comment

Hi Everybody

 

as my previous posts, sorry for my bad english and for my poor know how... i'm a beginner!

 

I'm trying to use better Plex docker bu i have tghe followig problem:

When i try to find the directory of my file in the webGUI i can't find my shares folder of my unraid. 

 

I would like to use Plex instead to Kodi, because Kodi has always a lot of issue, mostly with my favorite skin!! 

 

Tks for help me! 

Link to comment

Hi guys,

 

I added some extra drives (removed old ones) for my media but I am finding that within the PLEX WEBUI when I go to /mnt the shares I have created are not there? I'm not doing anything different to what I did before albeit its been a long time since I needed to do any of this so not sure if something has changed.

 

Can someone please assist? 

Link to comment

Hey @binhex, I'm having trouble getting all my tv channel scanned. I tried it with the windows version of plex and it worked. I assume it's just a version issue. Any hint then you'll be upgrading to 1.25?

 

Edit: nevermind, just realized the plexpass version fits my needs.

Edited by epp
Link to comment

Hi, so i am investigating what uses up lots of space on my cache drive and i found that my binhex-plex docker currently uses 261GB. i found the size by doing the following command

du -a -h --max-depth=1 | sort -hr
278G    .
261G    ./binhex-plex

 

Is this normal? It does take care of lots of files but i find that a bit much? Is there anything i can do about it?

Link to comment
23 minutes ago, likesboc said:

Hi, so i am investigating what uses up lots of space on my cache drive and i found that my binhex-plex docker currently uses 261GB. i found the size by doing the following command

du -a -h --max-depth=1 | sort -hr
278G    .
261G    ./binhex-plex

 

Is this normal? It does take care of lots of files but i find that a bit much? Is there anything i can do about it?

Many possible reasons for that.  Main biggie reason though would be if you copied the appdata somewhere and then placed it back and didn't structure the commands appropriately to handle all the symlinks etc present which will seriously bloat the appdata

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.