Anyone good with server/workstation motherboards? Thermal alarm problem :(

I finally got my flawless server/media player put together with a Supermicro X8DA6 motherboard but every time I play a 4k movie the thermal alarm goes off

[video=youtube;TxHiUpq6jRI]https://www.youtube.com/watch?v=TxHiUpq6jRI[/video]


Around 60 seconds after I stop playing the movie the thermal alarm stops going off, even though the CPU temp doesn't go down at all


There is very high cpu core usage on core 1 (92% usually on a 4k movie), not sure if the alarm is just due to cpu usage and not temp.


Bios version is R2.1 and the thermal alarm setting in the bios is set to "higher" temp. I tried it at the lower temp setting, exact same problem. The thermal alarm doesn't go off when I play 1080p movies or video games, just 4k movies.


If there's no solution I guess I'll have to cut it off the motherboard but I don't want to wreck the motherboard if it goes wrong.


Thanks for any ideas you might have.

What the heck am i looking at? Are you a space shuttle commander?
(googling now....)

Edit> I think you should tap it with the big hammer! That high frequency will melt your brain!

Assuming the CPU heatsink fan is operating nominally and the fins are completely free of dust and blockage about the only other thing that comes to mind is the thermal resistance between the processor and the heatsink: was there a thermal pad or thermal compound on the heatsink/processor interface when it was assembled? If you're not sure, obtain some new thermal compound, remove the heatsink, scrape off any old junk from both parts (heatsink and processor), clean with isopropyl alcohol and apply a small amount of new compound and re-assemble.

That heatsink looks a little small; is that the supplied piece?

Also, have you tried running the system with the side cover in place? Airflow management is pretty important and sometimes that means having the covers on.

Does the case have multiple fans, one at the front pulling cool air into the case and one at the back pulling hot air out? Are they both working? Are they both blowing in the proper directions?

Is the power supply fan also working properly? A large-capacity power supply like that one (1kW) can dump significant amounts of heat into the case.

YMMV.

I could be one of the other temperature sensors on the motherboard that isn't related to the CPU like the VRM temperature for example. Try using a program that will give you more temperature information such as CPUID HWMonitor.

Since it's an older motherboard maybe it's just defective, has it worked before or is this your first build with it?

If it was defective it would fail with 1080p also.
I would suggest the following:
take it apart and clean everything, use rubbing alcohol and cotton pads to wipe every inch
apply new thermal paste
get a bigger heatsink fan and possibly better heatsink
can you add one of those liquid cooled heatsinks?

reflash the bios, perhaps flash back to previous version just to see if the alarm kicks off again

grab a magnifying glass and check some of the connectors

consider using copper shims on some of the chips
get rid of the thermal pads and use copper shims

Just for fun run one of those stress testing programs to see what happens.

good luck

Blackfin said:

Assuming the CPU heatsink fan is operating nominally and the fins are completely free of dust and blockage

Click to expand...

It's a brand new build, liquid cooling kits/radiators are new, zero dust

Blackfin said:

about the only other thing that comes to mind is the thermal resistance between the processor and the heatsink: was there a thermal pad or thermal compound on the heatsink/processor interface when it was assembled?

Click to expand...

The heatsinks/pumps came with a thermal pad on the bottom but I removed it with isopropyl alcohol and used Coollaboratory Liquid Ultra "compound" instead (it's what I use for all my builds). The first time I got a thermal warning I took it all apart, cleaned the cpu's and pumps and re-goo'd them and re-assembled, same issue. I did it all again but used Arctic Cooling MX-4 but it was around 10 degrees more heat with that, so I removed it and went back to Coollaboratory Liquid Ultra. We're talking $68 worth of CPU thermal compounds so far My only possible next step is to de-lid the CPU's and go right on the dies, but these Xeon's are damn expensive if I wreck em

Blackfin said:

That heatsink looks a little small; is that the supplied piece?

Click to expand...

Do you mean the radiators in the top of the case? They are dual pass 120mm radiators (one per cpu). If you're talking about the devices mounted to the CPU's those are the pumps with internal waterblocks. It comes as a complete sealed kit all assembled (pump, hoses, radiators).

Blackfin said:

Also, have you tried running the system with the side cover in place? Airflow management is pretty important and sometimes that means having the covers on.

Click to expand...

It normally runs with the covers on, I just removed them for the video.

Blackfin said:

Does the case have multiple fans, one at the front pulling cool air into the case and one at the back pulling hot air out? Are they both working? Are they both blowing in the proper directions?

Click to expand...

The case has a front 320mm intake fan (it's the red LED fan you can see in the video). The side panel has a 320mm exhaust fan. There is a 120mm exhaust fan on the rear of the case (visible in the video). I've tried with the fans on the radiators both sucking in and blowing out. I get a few degrees better CPU cooling with them drawing air from the outside, through the radiators, into the case, but then my video card temps go up a few degrees.

Blackfin said:

Is the power supply fan also working properly? A large-capacity power supply like that one (1kW) can dump significant amounts of heat into the case.

Click to expand...

This could be a potential problem because I've modified the power supply a bit. The power supply doesn't dump any heat into the case as far as I can tell. It sucks air from inside the case (I've added an extra 80mm thermally controlled fan on the intake spot, should be visable in the video) and also upgraded the exhaust fan where it blows the hot air directly to out side of the case.

SWH said:

I could be one of the other temperature sensors on the motherboard that isn't related to the CPU like the VRM temperature for example. Try using a program that will give you more temperature information such as CPUID HWMonitor.

Click to expand...

I ran the trial of CPUID Hardware Monitor Pro, didn't see anything out of the normal:



It doesn't look like it can read the chipset temperature or other temperatures but now that you mention it I do remember that the chipset fan was pretty loud (like bad bearings) when I first got the motherboard and I didn't hear it when I was recording the video (but the alarm is pretty loud), I'll check that! That could be it!!

SWH said:

Since it's an older motherboard maybe it's just defective, has it worked before or is this your first build with it?

Click to expand...

I pulled it out of a server. It was working fine I would assume (I did run some prime95 and occt on it) but its cpu fans were the loudest fans on earth, seriously. So I built it into a new case with the liquid cooling kits.

-D- said:

If it was defective it would fail with 1080p also.

Click to expand...

The 1080p videos don't use any CPU and just a hair of the video card core (like 8%. The 4k uses the first CPU core at 92% and 89% of the video card core so I think that's why I don't see the problem on 1080p.

-D- said:

I would suggest the following: take it apart and clean everything, use rubbing alcohol and cotton pads to wipe every inch, apply new thermal paste, get a bigger heatsink fan and possibly better heatsink, can you add one of those liquid cooled heatsinks?

Click to expand...

I've re-thermal paste it too many times so far lol. I'm absolutely sure it's not the thermal paste. Someone did mention that my Xeon CPU's put out WAY WAY too much heat for 120mm radiators and they said to upgrade to a 240mm per CPU and possibly a better flow pump and better quality water heatsink block. That's about all the upgrades I could do (will be super annoying to have one radiator on the top of the case and the other on the side panel ).

-D- said:

reflash the bios, perhaps flash back to previous version just to see if the alarm kicks off again

Click to expand...

When I put it into the new case I flashed the bios with the newest version. Great idea going to an old bios and see if it has the same problem Thanks.

-D- said:

Just for fun run one of those stress testing programs to see what happens.

Click to expand...

I ran Prime95 on it before for hours, no thermal alarm. Goes up to about 60 degrees. I just ran it again for about 10 minutes, no thermal alarm:



When I start that 4k movie within about 45 seconds I get the alarm I tried Furmark to stress test the video card, again, no alarm.

I'm pulling out some stuff to check the motherboards chipset fan, I'm assuming that's GOT to be the issue.

Thanks for all the ideas guys.

That's weird. Try and RMA it if all else fails because that's definitely not normal.

SWH said:

I could be one of the other temperature sensors on the motherboard that isn't related to the CPU like the VRM temperature for example.

Click to expand...

You're a genius The chipset fan WASN'T EVEN SPINNING! I thought it was seized up so I took the video card out and gave it a spin and it seemed fine!?? I checked the wiring and connector, also looked fine. I powered up the computer without the video card installed, chipset fan was spinning fine!??! I put the video card back in, FAN DOESN'T SPIN! It looks like part of the plastic shroud around the video card touches the chipset fan I guess when I'm playing a 4k movie I'm stressing the video card just enough to blow hot air onto the chipset heatsink and probably stressing the chipset a bit too so that's why it sets off the alarm!!

I'm going to see if I can add a spacer under the video card and still get it in the slot enough for it to work properly. Not sure how good that chipset fan is going to breathe anyway with the video card almost completely covering it up and blowing hot air on it



Thanks again guys

Sleep on it, leave it alone for tonight.
Google for spacer ideas, you can't be the only one with the problem.
Covering that fan and trapping heat I don't think will work out for you.
Can you add another fan that blows across that area?

Perhaps you can get a lower profile heatsink and fan combo???
Can you trim off some plastic from the video card casing?