Mac Pro Temperature question

so if the max for the 5160, is 56c. Why are mine running over 70. I am not doing anything out of the ordinary. If i export a movie from imovie, my temps can get to over 70c. Surely apple knows this. why would they let them get so hot, if it was bad for them. Am i supposed to NEVER export a movie from quicktime, because it will fry my cpu's. I certainly hope not.

i live in nebraska, so it is not hot here.

so if the max for the 5160, is 56c. Why are mine
running over 70.

I BELIEVE that the "Core" as Apple calls it is what IS meant
by Intel when they say "Processor". Apple knows WAY WAY less
than Intel!

The reason your procs are runing so hot is because Apple
totally blew chunks when they set the speed for the fans.
They also blew chunks in getting the fans to increase speed
as the processors heated up.

Download Fan Control (linked above) and set the fan speed
to 1300 or 1400, quit the app and never revisit it. This
will keep your procs at Intel's recomended temp and also
keep your drives at a normal temp. The target for drive
temp SHOULD be 30c and a drive at 100% workload in a poor
to middelian cooled system should NEVER go over 40. In the
MacPro with OUT Fan Control they are at Over 40 more often
than not. This fact alone tells me that MacPro is NOT
getting the kind of air-flow that is REQUIRED to cool the
system. You will notice that by setting the fans to cool
the drives properly that your procs will fall into normal
standards as well.

I am not doing anything out of the
ordinary. If i export a movie from imovie, my temps
can get to over 70c. Surely apple knows this. why
would they let them get so hot, if it was bad for
them. Am i supposed to NEVER export a movie from
quicktime, because it will fry my cpu's. I certainly
hope not.

i live in nebraska, so it is not hot here.

Yeah, over 70 is crazy for these procs. And I have to say.
Apple did blow chunks. Apple has blown MANY chunks in the
past with sort of thing. The old iMacs overheat and blow
out all kinds of parts. Namely the fly-back transformers.
Apple is known for this sort of trouble. Ever heard of a
LapTop that gets so hot that the battery starts a thermal
reaction and the sucker catches on fire? Well Apple just
recalled a whole line of them that did just that. Trust
yourself. Trust common sense. About this issue, don't
trust Apple.

Here's a follow up message from Intel:

Intel:

The maximum operating temperature of the Intel(R)
Xeon(R) processor 5150 is 65 degrees Celsius.
The maximum operating temperature of the Intel(R)
Xeon(R) processor 5160 is 56.5 degrees Celsius.

Me:

OK, where exactly is this sensor located? This is case
temperature? Package Temperature? Or silicon chip
temperature?

Thanks again! I really appreciate the swift response!
Kewlness!

Intel:
Hello James,

Thank you for contacting Intel(R) Technical Support.

This temperature is for the processor it self, the sensor is located in the processor itself, but currently, we do not have information about the exact location of the sensor on the processor.

You might want to visit the following website for further technical information about the processor:
http://developer.intel.com/design/xeon/documentation.htm#designguides

Please do not hesitate to contact us again if you need further assistance.

Sincerely,

Adolfo S.
Intel(R) Technical Support

Intel(R) Processor Support Web Site:
http://support.intel.com/support/processors/index.htm

Intel is a registered trademark of Intel Corporation or its subsidiaries in the United States and other countries.
--------------------------------------------------------------------------

To defend Apple...

The whole battery issue was a Sony Issue, not an
Apple issue.

Partially true. The batteries bacame volitle at Apple's
running temperature. While this is a bit of chicken and
egg it ultimatly fell on Sony because the batteries were
rated for that maximum value. It's on Apple to some degree
that they run the systems hot enough that the batteries
ever even reach thier max heat rating.

Also this doesn't address the iMac issue. Some 40% of
several iMac models that include a CRT type monitor have
thier flyback transformers die within the 1st 18months
because the system is not properly cooled. Additionally,
both the CD/DVD drives and the HDD units die way before
thier MTBR rating for the same reasons.

These Mac Pro were designed with less
fans for a specific reason. Apple says that these
procs have better power management and do not need to
have as many fans. I would think that before
releasing this product they would have done many
tests running these procs to the max and testing
temps before putting them on store shelves.

Yet, surely Apple tested those iMacs I spoke of above?

Don't get me wrong. Apple is offering us a $3,500 to
$8,000 (depending on which brand you research) system
for $2,500. That's even $400 or $500 less than I can
build one with the same spec, on my own, shopping for the
very cheapest parts available. I love them for that!

Go Apple! Do it again! Do it again!

But on this fan speed issue and considering thier past
history with heat issues in general, I think it's safe to
assume that they blew it here as well. <shrugs>

Intel still is not giving you straight answers
regarding the placement of sensors...therefore we
have no idea what we should realy be concluding.

True. I rephrased the question. Let's see if the air gets
cleared from my final effort with this. <crosses fingers>

yeah, i am not going to change anything with the fans. I just don't think that apple released the mac pro without ever encoding video. they know how hot it gets, and they if it is good for it or not. they know that some people probaly encode video everyday for hours and hours and hours. maybe for days at a time.

How do we even know if hardware monitor is accurate?

I got 3 years of applecare, so i'm not really worried.

I think this is just rubbish. You berate Apple for getting their thermal design wrong without pointing out any testable facts. I suggest you read the "Thermal/Mechanical Design Guidelines" for the 5100 series Xeon processors, found here: http://download.intel.com/design/Xeon/guides/31335701.pdf

Intel doesn't specify the maximum core/junction temperatures anywhere, they only give the maximum case temperatures (and there is no ambiguity about this term, it is well defined as the temperature at the center of the heat spreader).

We haven't got sensors at the heat spreader, though, only in the cores and in the heat sink. The case temperature is somewhere in between (see figure 2-13 in the above document).

The fact that there are two chips with different thermal design powers (5150 and 5160) lets us take a stab at guestimating on-die temperatures that correspond to the maximum case temperatures (65°C and 56.5°C, respectively). If we assume that the allowable die temperatures would be the same for both chips, and also that the thermal resistance of the material coupling the die to the heat spreader is the same (both reasonable assumptions I think) we can extrapolate that a die temperature of 92.6°C corresponds to a case temperature of 65°C at 65W dissipation, and to a case temperature of 56.5°C at 85W dissipation. This figure seems very believable to me, it is in the ballpark I would have expected.

As long as the core temperatures stay clear of this limit there is no reason to cry panic and ramp up the fans. Besides, Intel uses a multi-stage thermal monitoring/management on these CPUs (also described in the above document) that won't allow the chips to overheat, regardless of what the fans do, and regardless of what size chunks Apple supposedly blows.

Cheers
Steffen.

i agree with you, apple and intel, is not going to let there cpus fry themselves. which is why i am not altering my fans.

I think this is just rubbish. You berate Apple for
getting their thermal design wrong without pointing
out any testable facts.

I'm trying to understand the facts. So far everything my
pee-brain can grasp points only to Apple setting thier fan
speed too low. AGAIN I might add! LOL the iMacs I mention
don't even HAVE fans. Like THAT'S going to work?!? Right
guys, let's put a harddrive and DVD player inside a TV set
with a hot CPU and NOT add a fan. Sorry that doesn't make
any sense to me and my logic bares out by the numbers of
broken ones linning the streets here in Japan. And yes,
Broken iMac actually did line the streets in neighborhoods
all across Japan. But Apple techs will tell you the same
thing about damaged iMacs from over-heating.

I suggest you read the
"Thermal/Mechanical Design Guidelines" for the 5100
series Xeon processors, found here:
http://download.intel.com/design/Xeon/guides/31335701.pdf

I read every word of it when it was first posted something
like 3 days ago. Many paragraphs I read multiple times.

Intel doesn't specify the maximum core/junction
temperatures anywhere, they only give the maximum
case temperatures (and there is no ambiguity about
this term, it is well defined as the temperature at
the center of the heat spreader).

Okay, so "case" temp means "Processor casing". OK, that also
follows with the discription I posted in msg #16 of this
thread.

We haven't got sensors at the heat spreader, though,
only in the cores and in the heat sink. The case
temperature is somewhere in between (see figure 2-13
in the above document).

Really? What is that illustration on page 82 of that PDF
you linked? That doesn't show a sensor in or on the IHS?

But you're already losing me here. First you say "it is
well defined as the temperature at the center of the heat
spreader [IHS]" and now you're saying "We haven't got
sensors at the heat spreader [IHS]"?

Ummm... oO

The fact that there are two chips with different
thermal design powers (5150 and 5160) lets us take a
stab at guestimating on-die temperatures that
correspond to the maximum case temperatures (65°C and
56.5°C, respectively). If we assume that the
allowable die temperatures would be the same for both
chips,

Die: http://www.simhq.com/technology2/images/technology_090a002.jpg
ok.

and also that the thermal resistance of the
material coupling the die to the heat spreader is the
same (both reasonable assumptions I think) we can
extrapolate

We can? How?

that a die temperature of 92.6°C
corresponds to a case temperature of 65°C at 65W
dissipation,

Hmmm... Ok, I'll trust you on that. 😉

and to a case temperature of 56.5°C at
85W dissipation. This figure seems very believable to
me, it is in the ballpark I would have expected.

Where's my quadratic equation calculator? 😀 But 90 is
absolutly NOT what I would expect! That's almost 200F
for a proc that is supposedly supposed to have an exremely
low thermal profile. The "CPU Temp" as told by "Fan Speed
Version 4.28" for my 3400+ AMD is only 100F. 120F (49c) at
extended peroids of 100% operation. At ~130F (55c) it loses
it's brains and the system crashes. The Pentium D dual
core CPUs have a thermal design power of 130W and a maximum
operating temperature of less than 70°C. You're saying that
the Xeon is supposed to run TWICE as hot as an 3400+ and
the system fans in a MacPro are supposed to be running at
ONE TENTH the speed as a typical PC case?

Well, it might be true. It sure doesn't sound right!

As long as the core temperatures stay clear of this
limit there is no reason to cry panic and ramp up the
fans.

Well, yes, actualy there is. The drives are running too
hot as well. And the only fan control app I know of just
happens to increase the air-flow in such a way that both
the drives AND the proc temperatures ruduce. So currently
if you wanna cool one you have to cool the otther. And
unless you're going to tell me that there's even more
strange "interpolations" when observing drive temperature
they do indeed need to be running cooler than they are.

Besides, Intel uses a multi-stage thermal
monitoring/management on these CPUs (also described
in the above document) that won't allow the chips to
overheat, regardless of what the fans do, and
regardless of what size chunks Apple supposedly
blows.

Hehe, but that document said that the staged cooling
profile was optional and not recomended. Or at least
that's how I read it.

Cheers
Steffen.

Well Steffen, I appreciate you taking the time with this.
I hope in my struggle to learn more about this system that
I'm not pizzing off too many of you readers here. 😉

All things considered I STILL recomend using Fan Control
if nothing else, for the drive temps - but also for the
CPU, DVD, MEMORY, and etc. Hatter said it right. Cool is
good. This much I know! Errr... I think... :P

I think i'm leaning towards your opinion of the issue
as well dotnet. Apple is selling this machine making
the point that it is much quieter and has better heat
and power managment. Still want to know from apple or
intel what their max core temps should be.

Me too. My last reply to the "tech" at Intel got kicked
upstairs so maybe we will know soon? I hope.

all across Japan. But Apple techs will tell you the
same
thing about damaged iMacs from over-heating.

Knowing how hot and humid it can get there in the summer (especially August since I've been there several times during that month), I can understand why,

Personally, I had two Power Mac G4s whose multi-processors fried because of overheating (thank goodness they were covered under warranty). The Radeon video card in a Power Mac Cube also died from overheating because of that issue where Apple mistakingly sent some of those cards out without the fan. Said drive in Cube also developed problems which were also diagnosed as heat related. I've a bunch of hard drives used in poorly ventilated external cases from places like LaCie and OWC which suffered from premature failure as a result of heat (and because of that, I now stay away from using external drives for long periods of time).

In otherwords, common logic dictates that higher heat levels can shorten the life of any component and if those heat levels can be reduced to lower points, it can only help increase the longetivity of those components. If this is hard for some to understand, consider what effect you would feel if you were sitting in a car with all its windows up, the internal temperature was 120F/49C and your choices were to just sit there and do nothing, fanning yourself with a hand fan, having a small electric fan blowing at 3500rpm, doing one of those with the windows open to let the heat out, or running the A/C. Common sense would provide the most appropriate choice of action. This also isn't a single choice covers all solution but Apple's choice of keeping the fans at the same speed despite the temperature increases of certain components follows exactly that same line of thinking. And yes, there are "in the lab" power management and cooling parameters which are great in the lab but may not necessarily be the best in real life.

Compared to the Power Mac G4 in its later incarnation, the Mac Pro at least has much better ventilation so heat build up isn't as big an issue at least to me. What does bothers me with the Mac Pro is the temperature variation swings with the default fan settings.

I think this is just rubbish. You berate Apple for
getting their thermal design wrong without pointing
out any testable facts.

Some of this is common sense and doesn't require testing. Stick a 500GB drive into an external case with no heatsink for passive cooling or even a fan, close it, run it all the time, and see how long it lasts. Also, try opening the case after one day and feel how hot the drive is (place your hand flat on it and see how long you can keep it there. Try the same test but leave the case open and face a fan like a 7" Vornado running at the lowest speed directly at the drive. You'll find the drive cool that you can keep your hand directly on it without feeling like your going to burn yourself.

Regardless, I did a bunch of stress inducing tasks (noted below) the past few days and found under longer heavy loads, it was possible to get my cores into the 168F/76C range fairly consistently. At this level, the heatsinks were reading 140F/60C. The memory modules got really hot; 184F/84C. My drive bays averaged 104F/40C. Room temperature was around 84F/29C. Once idle, the CPU core temperatures dropped back to the 115F/46C range (heatsinks to the 105F/41C range), memory modules to 156F/69C range.

Logically I would think this kind of dynamic range could eventually have some sort of impact on the longetivity of components but yeah, maybe Apple knows much better just like with my Cube or G4s (or like what MacBook users have concerns with). Based upon my most recent experiences with Apple's newer machines, the last thing I want is downtime regardless of whether or not I have AppleCare due to Apple's desire to emphasize quietness over more airflow which can only help to reduce the sort of conditions which could possibly lead to premature component failure (and especially of concern to those whose machines are in warmer locations to begin with).

In my case, I found a relatively good sweet spot with the CPU and memory fan at 980rpm, the exhaust at 1080rpm, and the expansion slot fan at 900rpm. The power supply fan does a good job at its automatic setting so I left it as is. Again, I'm adjusting the SMC fan variables directly instead of with Fan Control (which isn't Mac Pro aware in terms of the number of fans). First of all, there is no increase in the noise level at those speeds (the system is right on my desk) but the temperature variation especially with the processor cores, heatsinks, and memory modules does not vary as much as before and temperatures are naturally all lower across the board.

At idle, my heatsinks averaged 91F/33C, processor cores averaged 95F/35C, memory modules averaged 130F/54C. All 4 drive bays averaged 88F/31C. I ran the same tests as before doing a video encode (a 3.5 hour HD MPEG Transport Stream to a 640x480 H.264 file) , looping an hour long HD H.264 video in WMP on Windows XP via Parallels, and looping an 90 minute HD H.264 video using that cpu hog Quicktime Player in OS X saturating all 4 cores for at least 2.5 hours. The highest the processor heatsinks got up to was 98F/37C, processor cores averaged 118F/48C, memory modules averaged 140F/60C, drive bays 91F/33C. Increasing the CPU and exhaust fan speeds beyond 1100rpm resulted in diminishing returns at the current room temperature and I kept the fans under the 1200rpm "Fan Safe Speed" SMC parameter (i.e. F#Sf).

Personally, I prefer having this increased airflow based upon my own observations with the lower readings but aside from this posting to share what I've come across, I'm in no position to convince anyone to do the same since it all comes down to personal choice based upon prior experiences with heat related issues.

Mac OS X (10.4.8)

Hi,

How do you set the fan speeds manually?

Thanks for any help

Andy