New Install - Playing Video causes CPU overheat.

[Yorick]

The only time the CPU goes into overdrive, is when trying to play the file that does not work.

Sure. There are two issues, and it’ll help you to keep them separate:

- Ability to play files, or anything else that spikes the CPU to full load
- Temperature of the CPU at full load

A CPU at full load in a well-cooled case should go to around 60C. If it’s an SFF case I can see 70C. Anything above that, and cooling needs to be re-thought.

CPUs can run at full load for days and weeks and months without overheating, if the system has been built correctly.

The CPU cooler you used was designed for static pressure, forced air 1U server chassis. That’s not what you built in. You built in an SFF PC case, and that means your cooling solution needs to match. I found some threads online about what coolers people use in the exact case you use, I am sure you can find them too.

Horses. Water. Can lead one to the other, can’t make them drink.

Just in case though: Try and stay curious, assume you don’t know everything. Everyone tells you your CPU is allowed to run at full blast without overheating. Maybe there’s something to it. Explore that possibility.

 

[gwaitsi]

Sure. There are two issues, and it’ll help you to keep them separate:

- Ability to play files, or anything else that spikes the CPU to full load
- Temperature of the CPU at full load

A CPU at full load in a well-cooled case should go to around 60C. If it’s an SFF case I can see 70C. Anything above that, and cooling needs to be re-thought.

CPUs can run at full load for days and weeks and months without overheating, if the system has been built correctly.

The CPU cooler you used was designed for static pressure, forced air 1U server chassis. That’s not what you built in. You built in an SFF PC case, and that means your cooling solution needs to match. I found some threads online about what coolers people use in the exact case you use, I am sure you can find them too.

Horses. Water. Can lead one to the other, can’t make them drink.

Just in case though: Try and stay curious, assume you don’t know everything. Everyone tells you your CPU is allowed to run at full blast without overheating. Maybe there’s something to it. Explore that possibility.

man, i'm not trying to be a smartass here,
- the static pressure cpu fan, seals around the CPU and forces up into the open space of the case
yes, you are correct - but
a) it is the only fan that fits in the space between the cpu and mounting case
b) i have already said, i will 3d print a flange that clip onto the extraction part and takes the airflow to the vents on the side.

heating is not an issue and i currently have zoneminder, plex and clamav running all while continue to import 300GB ntfs data sets, (while music is streaming from plex)

The main issue and the reason i posted in the plex plugin (otherwise i would have posted in hardware) is;

from three sample files: x264 + acc 5.1 (mkv) & x264 + ogg 5.1 (mkv) & x264 + ogg stereo (mkv)

x264 + ogg 5.1 (mkv) - remuxed/enconded to x264+ogg5.1 (mp4) & x264 ogg2(mp4)
= neither work - not surprised, it is not within the mp4 standard (not withstanding it works in VLC)

mpeg + ogg stereo (mkv) = remuxed/enconded to mpeg ogg2(mp4)
= neither work - not surprised, it is not within the mp4
** remixed to mpeg + aac (mp4)
= works
** remixed to mpeg + aac (mkv)
= doesn't works

x264 + acc 5.1 (mkv) - remuxed/enconded to x264+acc5.1 (mp4) & x264+acc2(mp4)
= both work without problems (or cpu overdrive)
** remixed as x264 + acc 5.1 (mkv) original but from a much newer release of mkv
= it doesn't work e.g. with cpu override.

so problems confirmed are:
1) the plugin and/or freebsd within has a problem with MKV whether old or new compiled
2) the plugin and/or freebsd within has a problem with OGG whether 2 or 5.1 channel

converting to
mp4 + acc works in all cases
mkv never works
ogg never works ** unless it is audio only.

wma audio only doesn't work

m4a, ogg, flac, wav audio only works

i have 4TB of media that worked no problems with the current version of plex under linux.

is this a freebsd generic issue, or a plug-in specific issue?

Happy to share sample files for others to play with, if you can get over the CPU cooling issue

 

[Yorick]

¯\_(ツ)_/¯

Gave it my best shot. Can't get over cooling issue because the way my brain works, "stable system" comes before any other troubleshooting.

mkv never works

Yeah no idea, maybe try Plex forums. All ~12TiB of my library are mkv, and they all play fine. So it's not the container as such, but it might be specific to how you constructed your container. Mine are straight from disk, via MakeMKV.

 

[gwaitsi]

¯\_(ツ)_/¯

Gave it my best shot. Can't get over cooling issue because the way my brain works, "stable system" comes before any other troubleshooting.



Yeah no idea, maybe try Plex forums. All ~12TiB of my library are mkv, and they all play fine. So it's not the container as such, but it might be specific to how you constructed your container. Mine are straight from disk, via MakeMKV.

don't get me wrong. Appreciate the input on the cooling and as i say, i will work in parallel on that to get the cpu air out of the box, rather than circulating within the box. For now, i will leave the lid off and let the cpu cooler blow into open space.

nearly all of the mkv i have were created via Handbrake over the last 10yrs.
Except the test ones from today, i used avidemux to re-mux or encode to mp4 and mkv

I was hoping the plugin creator might have some input, but i will pop on to plex, because as i say the library was working fine under linux.
i assume plex uses the codecs belong to the system and not their own, should be something within freebsd - i guess

 

[HoneyBadger]

Not to go back to the cooling thing, but can you tell me if the HSF itself is getting hot? Check with a spot thermometer - if we're talking 90C potential, I wouldn't suggest a finger-test here.

If the HSF is getting really hot then it's an in-case airflow problem and you can solve it with some additional airflow or ducting. But what I suspect is that the HSF isn't going to be nearly as hot as the CPU which indicates poor contact between CPU/HSF - this could be a bubble in the TIM between CPU/HSF (solve by repasting) a poor latch on the mounts (they look like pushpin style - repaste and make sure they're fully seated) or an internal failure between CPU cores and heatspreader (which you don't solve without delidding)

But the "CPU gets immediately very hot and hitting thermal throttle" strongly points to bad contact between CPU/HSF. That's where I'd look, not at airflow in case.

 

[gwaitsi]

Not to go back to the cooling thing, but can you tell me if the HSF itself is getting hot? Check with a spot thermometer - if we're talking 90C potential, I wouldn't suggest a finger-test here.

If the HSF is getting really hot then it's an in-case airflow problem and you can solve it with some additional airflow or ducting. But what I suspect is that the HSF isn't going to be nearly as hot as the CPU which indicates poor contact between CPU/HSF - this could be a bubble in the TIM between CPU/HSF (solve by repasting) a poor latch on the mounts (they look like pushpin style - repaste and make sure they're fully seated) or an internal failure between CPU cores and heatspreader (which you don't solve without delidding)

But the "CPU gets immediately very hot and hitting thermal throttle" strongly points to bad contact between CPU/HSF. That's where I'd look, not at airflow in case.

like i say, the only time the CPU goes above 65 is when i try to run one of those mkv with/without ogg

right now, i am importing NTFS, zone minder running and playing mp4 video from the test files.
CPU 55, PCH 67 HSF 44

now runing a mkv file - kills
CPU 76, PCH 65 HSF 43

i've tried a number of times to push it now, but can't get above 76 at the moment.
so definitely blowing in the cpu air into the tight enclosed space is the problem.

So leaving the lid open is the workaround for now.

1) Two solutions are an exhaust fan (2x) along the vents.
2) custom flange/funnel to get the air from the cpu flush against the vent. rather than inside the box.

the case is a great little HO case and supermicro x11 is one of the few boards that fit the bill for this case and freenas.
option 2) would be the cleanest.

but need to come back to why the mkv as not able to run through the plex plugin.

 

[Fredda]

like i say, the only time the CPU goes above 65 is when i try to run one of those mkv with/without ogg

And that is probably the only time the CPU is maxed out. Transcoding with ffmpeg uses all the CPU power available, while in all other situations the CPU is mostly idling.

 

[gwaitsi]

so i installed emby and it has no problems to recognize all my media library. when playing a HD movie, it transcodes at about 85deg but plays smoothly and the system doesn't cause any warnings, etc.

I checked around on the plex forum and loads of people report their mkv no longer play after some plex versions.

looks like i'm an emby convert

 

[Fredda]

Glad it worked out for you. I think Emby is as good as Plex. My only impression is a native Plex TV app is available for more TVs compared to a native Emby app.

For completeness sake, just last week I looked a little bit deeper into transcoding when playing a full HD x265 source, which my TV does not play natively. You can tune the effort Plex puts into transcoding. So if Plex is not able to transcode, one should check the settings.
"Make my CPU hurt" setting might cause problems. But "Prefer higher quality encoding" works for me perfectly.
(Linux version on a I7-5500U CPU, which is not a very fast one)

 

[gwaitsi]

i did try playing with that, but didn't make any difference. i think it is the version of ffmpg they packaged in the plugin,
but like i say a quick google shows heaps of people struggling with mkv from the current plex major version onwards.
definitely repackaging them to mp4 works - but that is not an option due to qty.

i only occasionally use in the browser, but the more tell tail aspect is; emby had no problem reading my mkvs and obtaining the meta info and was able to pull the fan art, etc. plex wasn't able to do that, it simply couldn't work with the mkv at all. But from the various forum comments, it seems that this issue has hit multiple platforms so is not restricted to freebsd.

i use android tv boxes e.g. x96, t95m, and i replace the android dead weight with coreelec.
they all support x265, so direct streaming works best (haven't tried the emby plugin for kodi yet)
i don't run tv apps (in fact i block tvs,etc - since i discovered my samsung bluray loaded with malware 2mths before vault7 leaks confirmed samsung had loaded backdoors in their firmware below 1118 - mine was 1113).

shame. i used plex for many years. by

 

[Apollo]

I don't want you guys to think i am ungrateful.

I have the case open and will 3D print a flange to fit the cooler and channel it to the vents on the left, as of course the open case is running 5-10deg cooler. However, i am afraid this is a distraction from the real issue.

I have remux'd 3 different files into mp4 (because the files direct playing from youtube without problems are mp4)
i will report back separately, so we can focus on what stream, or container is broken

p.s. i know under the zoneminder that x264 creation has an issue under the version of freebsd in the package.

View attachment 41085

I think there is too much obstruction around the CPU to provide decent cooling.
It's not enough providing room temperature air intake when blown out hot air has no where to go.
In a nutshell, the CPU is being cooled off mostly by the heat it just released.

To validate cooling capability, I would take the motherboard out of the case and run it without air obstruction on the inlet and the outlet of the fan and heatsink.

 

[Apollo]

i did try playing with that, but didn't make any difference. i think it is the version of ffmpg they packaged in the plugin,
but like i say a quick google shows heaps of people struggling with mkv from the current plex major version onwards.
definitely repackaging them to mp4 works - but that is not an option due to qty.

i only occasionally use in the browser, but the more tell tail aspect is; emby had no problem reading my mkvs and obtaining the meta info and was able to pull the fan art, etc. plex wasn't able to do that, it simply couldn't work with the mkv at all. But from the various forum comments, it seems that this issue has hit multiple platforms so is not restricted to freebsd.

i use android tv boxes e.g. x96, t95m, and i replace the android dead weight with coreelec.
they all support x265, so direct streaming works best (haven't tried the emby plugin for kodi yet)
i don't run tv apps (in fact i block tvs,etc - since i discovered my samsung bluray loaded with malware 2mths before vault7 leaks confirmed samsung had loaded backdoors in their firmware below 1118 - mine was 1113).

shame. i used plex for many years. by

You can use ffmpeg to convert the mkv format to mp4 without re-encoding.

The basic command should be:

ffmpeg -i file.mkv -codec copy file.mp4

 

[gwaitsi]

You can use ffmpeg to convert the mkv format to mp4 without re-encoding.

The basic command should be:

ffmpeg -i file.mkv -codec copy file.mp4

x264 + ogg or x265 + ogg are not valid formats for an mp4 container.

i have now loaded the emby plugin on kodi and the tested playing the videos that were failing on plex,
all is no problem, no load on cpu, etc. i.e. they are direct streaming.

so plex is obsoleted for me.

re your previous post, did you look at the #18. there is no obstruction for the cpu cooler.
it is a static pressure channeling the hot air into the vertical position with no obstructions.

i will 3D print a bezel/funnel to bring the ait from the cpu cooler to the vents so it does not circulate inside the case.

 

[Apollo]

re your previous post, did you look at the #18. there is no obstruction for the cpu cooler.
it is a static pressure channeling the hot air into the vertical position with no obstructions.

i will 3D print a bezel/funnel to bring the ait from the cpu cooler to the vents so it does not circulate inside the case.

This is the post I replied to.
I don't think "static pressure" is the right terminology to define the cooling process.
Regardless, what comes in must come out and on the picture, I believe I see the side of the fan but where is the hot air coming out from? It seems to me it goes somewhere underneath or the side of the PSU stand.

 

[gwaitsi]

This is the post I replied to.
I don't think "static pressure" is the right terminology to define the cooling process.
Regardless, what comes in must come out and on the picture, I believe I see the side of the fan but where is the hot air coming out from? It seems to me it goes somewhere underneath or the side of the PSU stand.

that fan completely encloses around the CPU fan and the hole you see is the extraction part.
If i put the lid on the case, there is nothing to direct the hot air out of the case it will accumulate inside the case adding to the heating.
Hence why there is a 10deg drop in temperature when the lid is off

Supermicro 1U Active Proprietary CPU Heat Sink Socket LGA1155/1150/1151 (SNK-P0049A4)

1U Active Proprietary CPU Heat Sink Intel Xeon Processor E3-1200 v2/v3/v4/v5 Series Socket LGA1155/1150/1151 Supermicro Certified

store.supermicro.com

 

[Newfoundland.Republic]

that fan completely encloses around the CPU fan and the hole you see is the extraction part.

@gwaitsi - I'll give this another shot: The SNK-P0049A4 is designed for a Supermicro 1U case. If you take a look at the Supermicro's store for this cooler, the page provides you with the option to make sure that the cooler will work with a Supermicro case. Supermicro cases (no different that HP, Dell, Lenovo, etc.) for 1U (and higher) also have many fans pulling air in the front of the 1U case and out out of the back. In my case, I would love to replace the case fans with something more quiet (e.g. Noctura) but in talking with Noctura their fans do not produce enough static pressure (as compared to air flow - they are different) to replace the datacentre-grade fans. That is a decision I made when selecting my solution.

Just because the fan is made for a 1U case (and a specific set at that) does not mean it will work in your build.

--Edit: I just noticed that you seem to say that you are using a Xeon E2234. That CPU cooler is made for a Xeon E3-1200. Check the specs on the TDP.

 

[gwaitsi]

@gwaitsi - I'll give this another shot: The SNK-P0049A4 is designed for a Supermicro 1U case. If you take a look at the Supermicro's store for this cooler, the page provides you with the option to make sure that the cooler will work with a Supermicro case. Supermicro cases (no different that HP, Dell, Lenovo, etc.) for 1U (and higher) also have many fans pulling air in the front of the 1U case and out out of the back. In my case, I would love to replace the case fans with something more quiet (e.g. Noctura) but in talking with Noctura their fans do not produce enough static pressure (as compared to air flow - they are different) to replace the datacentre-grade fans. That is a decision I made when selecting my solution.

Just because the fan is made for a 1U case (and a specific set at that) does not mean it will work in your build.

you are buying into the marketing. the dynatron 1U fan can be bought from any number of manufacturers for the CPU socket type.
It forces the air in a single direction out the vent, so yes, you need to either provide a funnel to a vent on the case to bring the air outside the case, or have extraction fans in the case to get the air out. at the momement, you can see from my photo #18 the exhaust part of the fan points up, and the case top is not on. Last time i looked, hot air rises (unless you are telling me there is a change to physics laws i'm not aware of), so the escaping air goes straight up and is not restricted. i have possible two possible solutions. 1) stick a couple of small exhaust fans on the side vent of the case (which i think would be less efficient) 2) 3D print a funnel with a bezel that clips onto the fan case and directs the air to the vents on the case - this is the solution i am working on

 

[Newfoundland.Republic]

@gwaitsi - I'm used to enterprise equipment (both at work and at home) so I tend to follow manufacturer recommendations (especially at work) where downtime must be minimised. Supermicro has other CPU coolers for their different chassis that they know will work.

Anyway, I'm done with this thread. Good luck.

 

[gwaitsi]

@gwaitsi - I'm used to enterprise equipment (both at work and at home) so I tend to follow manufacturer recommendations (especially at work) where downtime must be minimised. Supermicro has other CPU coolers for their different chassis that they know will work.

Anyway, I'm done with this thread. Good luck.

if you like specifications, the fan produces an airflow of 14.7CFM at 100% duty cycle. The CPU requires 11.6 CFM at 100%.
the fan is not the problem. once the duct is printed to bring the airflow directly out of the case, i will be able to put the lid back on.
(i could now anyway because the heat increase is nominal. the original problem is plex and (most likely the ffpmeg encoded). transcoding on emby does not cause the CPU to go over 80deg, where as plex transcoding of the same file does.

 

[Fredda]

No. IMHO you still get it wrong.

Of course there is something wrong with the Plex setup. It should not run in full blown trancoding mode. Especially as your client is
capable of playing h264 and HEVC. It might also be a problem of the plex client advertising wrong capabilities, so the server
thinks it does not have to transcode, while the Emby plugin advertises correctly its capabilities.

BUT this shows a problem with your box. You do have a cooling resp. airflow problem. Plex has no hidden tricks to overheat a CPU when it's running at full power. Any application causing all your CPU cores running at 100% over a longer time (e.g. recoding video with handbrake, doing a mass conversion of music from flac to mp3) will cause the same effect as Plex with transcoding.