Xeon X5690 temperatures and fan speed

[Squuiid]

Looking to find out what normal temps are for the X5690 CPU.
Running Geekbench stress test on OS X I'm seeing 84 degrees C and running Prime95 in Windows I'm seeing 86 degrees C.
The CPU fan is not ramping up beyond default speed (BOOSTA is 800-856 RPM) in either OS however. And yes, I've reset SMC.

Is this expected fan behaviour?

 

[bokkow]

Well, yes, kind of. My X5680 idles at 37-38 degrees C and can go all the way too 80+, where 78.5 degrees C is the Intel claimed max temperature, if I would let the SMC manage fan speed.

But I don't, I let Macs Fan Control take over from startup (available for OSX and Bootcamp). I set BOOSTA to start increasing when hitting 55 degrees and max RPM at 75 degrees C, same for INTAKE and EXHAUST.

Now what happens at full load is the fans turn on from 55 degrees and gets a bit audible but nothing too annoying and the max temp is 65 degrees at slightly increased fan speed. You will have to play around a bit to find your sweet spot perhaps. 99% of the time my Pro is whisper quiet and cool enough.

 

[Squuiid]

Well, yes, kind of. My X5680 idles at 37-38 degrees C and can go all the way too 80+ if I would let the SMC manage fan speed. But I don't, I let Macs Fan Control take over from startup (available for OSX and Bootcamp). I set BOOSTA to start increasing when hitting 55 degrees and max RPM at 75 degrees C, same for INTAKE and EXHAUST.

Now what happens at full load is the fans turn on from 55 degrees and gets a bit audible but nothing too annoying and the max temp is 65 degrees at slightly increased fan speed. You will have to play around a bit to find your sweet spot perhaps. 99% of the time my Pro is whisper quiet and cool enough.

Yes, that's what I did in Windows, thinking this was strictly a BootCamp SMC issue, but then I realised OS X does the same thing!
I'd rather not use a utility. Surely the SMC should keep temps within a reasonable range by ramping up the fan. That's what it is there for afterall! Is there an issue in El Cap specifically perhaps, or is the cMP 2010's SMC not suited to handling this CPU?
Or do all cMP 2010 CPUs just run hotter than Intel's stated maximum temp under heavy load?

 

[MacUser2525]

Yes, that's what I did in Windows, thinking this was strictly a BootCamp SMC issue, but then I realised OS X does the same thing!
I'd rather not use a utility. Surely the SMC should keep temps within a reasonable range by ramping up the fan. That's what it is there for afterall! Is there an issue in El Cap specifically perhaps, or is the cMP 2010's SMC not suited to handling this CPU?

It is Apple's default behavior on all their machines run them hot as hell rather than have the fan kick in. It has been this way for many years through all their OS releases you either live with it that way or use a utility to modify the defaults.

 

[bokkow]

Second that ^^^

I feel the same way as you Squuiid, regarding 3rd party tools to take over the tasks the Mac Pro should do on its own by default. However, after setting up Macs Fan Control once and let it run at startup I did not look back anymore. For me it proved itself more than usefull and I find it essential especially when using CPU's other than stock.

By the way, Apple did ship MP5.1 with W3680 (3.33GHz 6 core with 130W TDP) so it is kind of crazy that the fans do not kick in by default through the SMC to keep temps below 78.5 degrees.

 

[flehman]

It is Apple's default behavior on all their machines run them hot as hell rather than have the fan kick in. It has been this way for many years through all their OS releases you either live with it that way or use a utility to modify the defaults.

4,1->5,1 with a single X5690 here. Typical idle temp range is between 30-35 Celsius. It usually reads mid to high 20s when waking from sleep and then creeps up above 30.

 

[MacUser2525]

4,1->5,1 with a single X5690 here. Typical idle temp range is between 30-35 Celsius. It usually reads mid to high 20s when waking from sleep and then creeps up above 30.

So what has that to do with the price of tea in China? The OP is talking high under load temperature not idle. The only way to get lower in that case is to use third party utility, Apple has chosen defaults that are not comfortable for some people.

 

[flehman]

So what has that to do with the price of tea in China? The OP is talking high under load temperature not idle. The only way to get lower in that case is to use third party utility, Apple has chosen defaults that are not comfortable for some people.

Actually the OP asked for "normal temps," which could be construed to encompass idle or load. The first poster offered his idle temps and you didn't jump down his throat. I was trying to be helpful..

 

[MacUser2525]

Actually the OP asked for "normal temps," which could be construed to encompass idle or load. The first poster offered his idle temps and you didn't jump down his throat. I was trying to be helpful.

"Running Geekbench stress test on OS X I'm seeing 84 degrees C and running Prime95 in Windows I'm seeing 86 degrees C.
The CPU fan is not ramping up beyond default speed (BOOSTA is 800-856 RPM) in either OS however. And yes, I've reset SMC.

Is this expected fan behaviour?"

 

[h9826790]

That 78.5 is the T-case Max, not the max core temperature. If the CPU is an engine, it's a bit like Intel only tells you that the max exhaust pipe temperature is 78.5, and now you are worrying about the cooling system can't keep the engine core temperature at or below 78.5. Both temperatures are related, but no absolute relationship, and no way to accurately work out one temperature from the others.

OP's fan behavior is absolutely normal. The SMC is working as expected. Apple decided to keep the fan at idle until the CPU reach ~83C, then spin up the fan only a little bit to keep the temperature at ~85C. That's why the fan RPM go up to 856, but not stay at 800.

AFAIK, Apple treat the W3680 with exactly the same fan profile. If the W3680 can survive, there will be nothing to worry about for the X5690, because it's T-case max is even 10C higher than the W3680.

Of course, you may argue that this temperature will shorten the CPU's life. I don't have any knowledge in this area, and I didn't read any reliable source about how much life will be shorten. If you are not happy with this temperature, you may use fan control software to custom make the fan behaviour that you want. But IMO, it's more like to keep yourself happy, rather than the CPU really need it.

The T-case Max is almost completely useless for normal end user. The CPU itself has the thermal throttling function to avoid overheat which base on it's core temperature. This core temperature can be way higher than the Tcase Max. Above that, there is another critical temperature which will cause self shutdown to prevent thermal damage.

i.e. A CPU that has a 65C Tcase Max, may have 100C max core (thermal throttling start), and self shut down at 120C. Therefore, the CPU not only can work under 80C, but also with the turbo boost still available.

If you monitor the CPU clock speed in Windows, then most likely you will find that the CPU is still running at it's full speed when you read that 86C. It can't even reach the thermal throttling temperature. Assuming the thermal paste is correctly applied, and the heatsink's performance is not greatly degraded by dust etc. Let the SMC to manage the fan speed should be very safe.

 

[flehman]

"Looking to find out what normal temps are for the X5690 CPU. Running Geekbench stress test on OS X I'm seeing 84 degrees C and running Prime95 in Windows I'm seeing 86 degrees C.
The CPU fan is not ramping up beyond default speed (BOOSTA is 800-856 RPM) in either OS however. And yes, I've reset SMC.

Is this expected fan behaviour?"

You conveniently did not quote the first line of the OP's post.

 

[Squuiid]

Folks, thank you very much. Really useful info here.
I would rather the fans stay quiet, and as long as the high temp doesn't bring instability (which hasn't happened) then I don't really mind how hot the CPU gets. Definitely not concerned about longevity and, as I mentioned, I would rather not use a utility and just let SMC do its thing.
As long as this is by design then I'm happy. Just wanted to make sure there wasn't an issue I needed to address.
Thanks again everyone.

 

[gonnok]

Just swapped my CPU's for X5690s, idle temperatures are 33(A) and 40C(B).

 

[ActionableMango]

Just swapped my CPU's for X5690s, idle temperatures are 33(A) and 40C(B).

That's odd. Did you do the thermal paste differently on the two processors?

 

[gonnok]

That's odd. Did you do the thermal paste differently on the two processors?

Nope.
I was a bit concerned first as well, but I saw similar differences in the mac fan control thread. I guess its becouse the CPUs are in line, and the rear cooler takes in warmer air.

 

[carpsafari]

I guess its becouse the CPUs are in line, and the rear cooler takes in warmer air.

Nope, has nothing to do with it!

CPU A is always hotter than CPU B, because the heatsink of CPU A also has to cool the Northbridge module.

 

[gonnok]

Nope, has nothing to do with it!

CPU A is always hotter than CPU B, because the heatsink of CPU A also has to cool the Northbridge module.

Thanks for the head up.
So these are normal temperatures after all.

 

[pastrychef]

The Northbridge has its own heatsink.

 

[carpsafari]

So?

Better stick to your Hackintosh.

 

[pastrychef]

What the hell does the CPU heatsink have to do with the Northbridge heatsink?

Stop spreading BS.

 

[carpsafari]

Have you upgraded ±100 Mac Pro's?
I did, and each and every 8-core /12 core you will see a hotter CPU A.

The fact that the NB is placed under the heatsink of the CPU is for a reason!

 

[pastrychef]

Yes, it's under the heatsink. That does not mean the CPU heatsink is cooling the Northbridge or its heatsink.

The Northbridge and it's heatsink are also under the heatsink of single CPU MacPro4,1/5,1s, why don't those CPUs run as hot as CPU A of a dual CPU model?

It doesn't take a genius to do a CPU swap.

 

[h9826790]

If fact, we may do some test to know more about the reason of this temperature difference.

Lets start with the facts

1) NB has it's own heatsink
2) NB heatsink is under heatsink A
3) NB heatsink is at the upstream of CPU A's socket (the contact point between CPU and heatsink)
4) CPU A is at the downstream
5) heatsink A and heatsink B is not identical (e.g. size ,weight, shape...)

And the theory

a) CPU A's heat sink has to cool down the NB as well
b) CPU A's heatsink is cooled down by warmer air (air pre-warm by CPU B).

To test theory A. We may...

Is there any operation that can stress the NB a lot but not the CPU? e.g. PCIe SSD RAID 0 speed test? If yes, we should able to see the temperature difference increase. To make a control test. We should record the CPU usage, and run another control test which only stress the CPU to the same level, but not the NB, and then compare the results.

To test theory B. We need more tool. e.g.

Measure the airflow temperature just before it go into each heatsinks (by a thermometer). And compare if that temperature difference is similar with the CPU heatsink's temperature difference.

Anyway, my own guess is that's the resultant temperature difference with more than one reason. e.g.

A) the airflow is warmed up by CPU B
B) the airflow is further warm up by the NB heatsink
C) the CPU A heatsink is a bit smaller than heatsink B because it has to leave some space for the NB heatsink
D) .....

The above is just my theory, zero proof. However, I think it's hard to conclude that heatsink A is warmer because the NB is there. This is simply not enough data to support this conclusion.

 

[seveej]

It is Apple's default behavior on all their machines run them hot as hell rather than have the fan kick in. It has been this way for many years through all their OS releases you either live with it that way or use a utility to modify the defaults.

Just to point to the exception which validates the rule:
There has once been a mac "workstation" which definitely was not noise-optimized. For those who were not around back then, the PowerMac G4 "Mirrored drive doors" (or MDD) (http://www.everymac.com/systems/apple/powermac_g4/specs/powermac_g4_1.0_dp_mdd.html) was so noisy, it received the semi-official name "Wind tunnel".

That design was seriously flawed, with two (these were all dual CPU systems) very hot G4 processors, bad resonance proofing, and a power source crammed into an oblong box (which both made air throughput low, and necessitated two small and (per default) hideously noisy fans for the PSU). Any half decent tinkerer could get the noise level more decent by switching fans and adding some dampeners.

The flak Apple received for the noisiness of the MDD was one of the central motivating factors for the PMG5's elaborate air "thermal zone" design (the G5 had similar performance per watt as the G4 - Brilliant if you're looking for a design heater, but not otherwise.)

RGDS,

 

[MacUser2525]

There has once been a mac "workstation" which definitely was not noise-optimized. For those who were not around back then, the PowerMac G4 "Mirrored drive doors" (or MDD) (http://www.everymac.com/systems/apple/powermac_g4/specs/powermac_g4_1.0_dp_mdd.html) was so noisy, it received the semi-official name "Wind tunnel".

I used to think of it as a vacuum cleaner when I owned a dual 1.25, nothing could be done even by replacing the fans in it to quiet it down and it still got hot as hell.