What is the recommended temperature for a MacPro 2009 upgraded with a Intel Xeon W3680?

Studio, doing I litlle reading on the intel site, both processors W3680 and yours W3565 have the same Tcase of 67.9C = 154.22F max. So I belive we are within the specs right?

I also read your post below

Regarding Tcase (which is 67.9C), that is Intels' upper limit for 'normal' usage. The following is an explanation by 'Intel_Karla' (presumably an Intel employee) responding to someone on the Intel Support forum.....

"Anything from the Tcase and below will be the expected temperature of the processor in normal use, anything that doesn't stress out the processor (watching movies, burning CDs, browsing the internet, creating documents, etc.) When the processor is stressed out meaning that you are running heavy processor applications that take control of the CPU or uses it at 100% the temperature will go beyond the Tcase. It can perfectly reach 90 to 95 degrees and the processor will still be OK. The cooling fan is in charge to keep that temperature there.

If the processor temperature reaches 100°C or more it will send a signal to the motherboard to shut down to prevent mayor damages and most likely it won't be possible to turn the computer back in until it cools down."

I'll be back tonight with the test results, I'll try screen shoots to post also.

Once I'm confortable that the processor "Surgery" is ok, I'll end my questions and will start to racke the leaves of my backyard 🙂

I'm using the W3570, but I believe Tcase is still the same.

It's likely that CPU-intense activity (like stress-testing) is going to drive the CPU Tdiode temp higher than 67.9C.

Keep in mind that bombarding the system with the 'yes' command 10 times (or even using a video encoder like Handbrake) is not considered 'normal' usage.

Tcase, I believe, refers to the temperature around the CPU case---not the core temps, which are higher. When Intel Karla says it can reach 90 to 95 degrees and still be OK, I am hoping that she if referring to the Core Temps (which are not measured by iStat) and not the Tcase. A Tdiode temp of 95 would mean your cores are close to 110C---which is not good at all. A core temp of 95C would mean the TDiode is around 83C.

Anyway.....be prepared to see Tdiode exceed 67.9C. HOw much it exceeds is what I'm curious to know. And at what point the Fans begin spinning.

I look forward to hearing your results.

Your 2009 should never have run with 10.5.7 - 10.6.2 or iTunes and any audio, as the core temps would hit 80-85*C immediately and stay there.

Like saying "Yes" or "Are you free?" or "Are you idle?" would constantly wake up the processor core and keep it up.

There was a Facebook page on the problem. No one it seems tested for how the system and audio code would play on Nehalem processors (and it didn't happen running Windows).

Just launching apps, opening files, can see a brief but 10-20*C jump in core temp but immediately falls back, which is what you want and expect.

I think you get better core temp readings either iStatPro or what I favor, Hardware Monitor or free version Temperature Monitor.

Mavericks is suppose to be more aggressive in idling and reducing processor usage to reduce heat, lengthen battery life. But have to see if they treat Xeon cpu differently. Even servers are very interested in being able to reduce power and cost, reduce the need for aggressive cooling and thereby save money by allowing servers to run even 5* warmer. Also with disk drives.

I always read that running a stress test for stability and reliability that 85*C Core Temp was more or less the cut off for safe running or overclock (Windows) - and then back off, you proved how far and how much you could get out of your case and system with this processor.

Yes Hatter, I noticed the jump as you said from 5 to 10 F from the idle temp as I launch an app like photoshop but it then drops back another 3 degrees with the application opened. As far as Maverick, I 'll hold off on that for now even if it improves the temps. One thing I forgot to mention is that I get the same temps in ML and SL.

Thanks Hatter for your imput as usual.

Ok, here are the results

Let me know if I miss a step. If anyone would like to comment on this please do so. First (Iddle), Second (Hanbrake), Third (Cores Maxed Out for 20min).

I couldn't hear the fan going much when all the cores were maxed out.

Regards

It appears that, with the W3680 installed, your CPU temps peak at 79C (diode) and 10C from Prochot (which implies core temps at 90C). Once you are at those levels, the temperature holds steady.

My quad-core behaves differently. The temperature will continue to rise beyond your peak until the fans finally spin up and stablilize things. My diode will reach 85C and 2C from Prochot before the INTAKE, EXHAUST, and BOOST fans start to move from their minimum default rpms.

This will prevent the temperature from rising any further and bring it down to 4C from Prochot where it will remain until the stress test ends.

So 2C from Prochot appears to be the trigger for the CPU fans to take action. Since you never went beyond 10C from Prochot, the fans did nothing. That's why you did not hear much fan activity.....because there was none.

I don't know why my quad runs hotter than your hex. Either the hex simply generates less heat, or your thermal compound is better than mine. Who knows?

I am rather shocked that the CPU temps are permitted to rise to such levels. 98C core temps is right at the brink. It may be within specifications, but only just barely. Apple's got all those fans in there, and they barely use them. You can nearly double the speed of those fans before they become audible.

This forces users to rely on 3rd-party fan control software. At least for those who choose to worry about these things.

I like the Hatter's views on CPU temperatures. I wish he was on Apple's engineering team.

Well, some of your explanations about the test that I posted went over my head a bit. The tests were made in fahrenheit, on your post are talking celsius?

And yes, I could not hear the fans at all on the stress test.

This is how I did the upgrade and since it was a used computer I did this:

Remove all drives, graphic card, optical drive.

Vaccum using a small brush at same time to help to get all dust out.

Held all the fans (fins) with a pencil and used the Metro air blower without letting the fans to spin.

remove the daughter card and memories.

removed the heatsink.

cleaned the heatsink with Artcic pro cleaner untill it was shinny (careffuly not to touch the sensor wires next to the heatsink.

Cleaned older processor and removed from the cage.

Inserted the new processor.

tinted the heatsink with artic pro 5 and removed all the the access (like waxing a car)

applied 1mm line of the compound (artic silver 5) on top of the processor starting 1/8 from the edge of it (use both hands to hold the little sering to get a perfect straight line right on the middle of the processor from left to the right (looking at the daughter card horizontly).

Put heatsink back without overtight the 5 screws, few turns on each screw ramdonly and evenly untill it stops.

put everything back (memory, etc).

turn the computer on and do pram and vram.

Installed a fresh copy of mountain lion.

Thats what I did

I would like to hear Hatter's opinion as well

Just want to mention that I'm not an expert on this at all. Just to see if the temps I got are within the range (normal).

Thanks

Yes, you posted your Temps in fahrenheit and I translated them into Celcius because that's what I relate to.

Your peak temp of 176F = 80C more or less. This was the stress-test maximum temperature at the CPU Diode.

My peak temp is 85C when doing this test. What this means is that my quad gets hotter than your hex.

This is good news for you. Lower temperatures are preferred. I wish I had your numbers.

The fans never sped up on your machine because it never got hot enough to trigger them. BOOST fan was 857 rpms on all three istat images you posted. This also good news for you.

But I feel that Apple should set it up that the fans speed up sooner (like when the cores are in the 70'sC) rather than waiting until the upper 90's. One's CPU cores should never approach or surpass 100C so easily. very undesirable.

As far as I can tell, Your Mac Pro is behaving normally. It may even be running cooler now that you've put a hex in there.

But in gerneral, I think Mac pro's are permitted to run to hot.

Studio K, thanks for your input.

Thats good to know that my temps are ok since I wasn't sure. I should've cheked the temp on my computer when I first bough it while I still had the old processor to compare with the new one now.

One thing that I was really pick about was the prep work. I tried to describe the as best as I could what I did.

Mac Pro collects lots of dust inside and that's the nature of any desktop I guess.

One important thing I noticed was that when I removed the heatsink to have access to the processor I saw some gunk on top o the daughter card. This gunk came from a thermal tape that is placed around the aluminum cage. It looks like it melted a bit on top some of the components of the daughter card. I cleaned those components the best as I could without removing the thermal tape. It looked like a spill on the daughter card. Not sure if it has anything to do with the temp problem but just want to share.

Studio, you can redo the processor swap again just in case.

Best Regards and thank you again

The gunk on the daughter card is normal, I think. I had it on mine, too. Something to do with that pink strip of tape.

I've re-done my thermal application 3 times, trying different methods. Same temperature behaviors each time.

I, too, wish I had studied the temperatures before the CPU swap, but I was not interested. Why worry about such things when the machine is stock? It's only when you mess around with things yourself that you begin to think about it all.

As far as 'normal' temperature ranges go, I can't say for sure what Apple's Specs are for the Nehalem CPu. I don't know that it is published officially.

Even Intel does not publish the actual 'Max' temperature for their CPU's (that is, the hottest the cores should get before throttling or shutdown). Tcase is not the 'Max' temperature, by the way. It's only a reference temp like 'Intel Karla' explained.

From what I've read over time and observed by comparing notes with others like yourself, the temperature ranges on our machines are typical of a Nehalem 2009 Mac Pro. Nehalems are hot, I guess.