Dual Xeon X5550 getting really hot.

[Methal]

Currently folding protiens in a VM running the two x5550 at 100%. I'm getting temps upwards of 90c. (way way hot if you ask me)

Wondering if there is a way to put a custom/aftermarket heat sink on these things.

Been searching on line and haven't been able to see anything yet. Is it the same 1366 socket as a PC? or did apple change that to fit these less than adequate things on here?

Thanks for any info you give me. =)

 

[Inconsequential]

90C is perfectly within spec.

Download fan control software to boost the load fan speeds if you want.

No 3rd market coolers would be better.

 

[alphaod]

I [used to] fold a lot on my Mac Pro. Same processors.

The highest temperatures I got were 70ºC.

I would suggest taking your computer apart and cleaning all the fans, and blow out some dust on these CPU heatsinks. Maybe the thermal paste on the CPU has started to dry up and cake, so consider reapplying the thermal paste. These things are very easy to do on the Mac Pro, just be careful with the processors as the 2009 models do not have integrated heatspreads on the processors and so the cores can be cracked.

Anyways I completed disassemble my Mac Pro once every 6 months or so, so far, to clean out dust, replace caked up thermal paste, clean contacts, etc. My computer works fine.

 

[Rokeneer]

Currently folding protiens in a VM running the two x5550 at 100%. I'm getting temps upwards of 90c. (way way hot if you ask me)

Wondering if there is a way to put a custom/aftermarket heat sink on these things.

Been searching on line and haven't been able to see anything yet. Is it the same 1366 socket as a PC? or did apple change that to fit these less than adequate things on here?

Thanks for any info you give me. =)

My computer runs about that temperature when running at 100% capacity. Get smc fan control and crank up the fans. I did that, and the temperature dropped durastically.
Hope this helps!

 

[nanofrog]

90C is perfectly within spec.

Download fan control software to boost the load fan speeds if you want.

No 3rd market coolers would be better.

This would depend on the specification you're looking at.

For example, 90C may fit in the short duration spec (i.e. 8hr max intervals), but not the long term spec (24/7 operation @ 100% on all cores), which is only 67C IIRC.

So entering SMC and pushing the fan speeds may be advisable, depending on the duration of OP's usage that pushes the temps that high (personally, I'd recommend it).

Another thing that hasn't been mentioned, is the CPU coolers and fans may be caked with dust (which acts like a blanket and retains heat), and need to be cleaned.

 

[Rokeneer]

This would depend on the specification you're looking at.

For example, 90C may fit in the short duration spec (i.e. 8hr max intervals), but not the long term spec (24/7 operation @ 100% on all cores), which is only 67C IIRC.

So entering SMC and pushing the fan speeds may be advisable, depending on the duration of OP's usage that pushes the temps that high (personally, I'd recommend it).

Another thing that hasn't been mentioned, is the CPU coolers and fans may be caked with dust (which acts like a blanket and retains heat), and need to be cleaned.

With the dust idea though, wouldn't getting smc and cranking it up to the full 5000 RPM clean it out itself just because of the amount of airflow?

 

[alphaod]

With the dust idea though, wouldn't getting smc and cranking it up to the full 5000 RPM clean it out itself just because of the amount of airflow?

If you ever used a vacuum, when you turn up the power, the filter only gets clogged up faster.

 

[Rokeneer]

If you ever used a vacuum, when you turn up the power, the filter only gets clogged up faster.

That makes sense. So then using some compressed air to clean out my Mac Pro every once in a while is a good idea?

 

[Inconsequential]

This would depend on the specification you're looking at.

For example, 90C may fit in the short duration spec (i.e. 8hr max intervals), but not the long term spec (24/7 operation @ 100% on all cores), which is only 67C IIRC.

So entering SMC and pushing the fan speeds may be advisable, depending on the duration of OP's usage that pushes the temps that high (personally, I'd recommend it).

Another thing that hasn't been mentioned, is the CPU coolers and fans may be caked with dust (which acts like a blanket and retains heat), and need to be cleaned.

Tcase and Tcore are two different things

I've run my Mac Pro at full load 24/7 (~95C core) for months on end before, it's still here and still working

The heatsink temp should be 70C or less but the core temp will be much, much higher.

 

[nanofrog]

That makes sense. So then using some compressed air to clean out my Mac Pro every once in a while is a good idea?

Definitely.

Tcase and Tcore are two different things

I assume you mean Tjunction in terms of the cores.

Unfortunately, neither is all that accurate anyway as they're usually calculated off of the Thermal Diode Sensor (seen it referred to as DTS), which aren't the most accurate. The only way to get better accuracy, is to use a calibrated thermocouple (usually found in the heatsink's CPU when present).


In the case of the MP, it has a thermocouple in the heatsink, but it's possible the OP is reading the wrong value (been awhile since I had access to a MP, so my memory is a bit rusty on what temp values it shows).

That said, the first thing to do is clean the dust out of the heatsinks (the OP's last post gives me the impression they've never been cleaned).

BTW, take a look at the following image:

• Ta = Temp of air
• Ts = Temp of heatsink (base)

As you can see, Ts and Tcase are only separated by the thermal paste (TIM).

I know the image is for an LGA775, but it does apply to the more recent CPU's as well.

 

[Methal]

Its completely dust free. Both coretemp and cpuid are telling me that the current temperatures are between 91c and 89c for cpu 0 (nearest the back of the tower) and 86c and 81c on cpu 1 (nearest the front of the tower)

The fans are not speeding up as the cpus get hotter. <-- thats got to be bad.

I have noticed with cpu 0-the one at the back- the rear fan only covers 2/3rds of the HS. The rest is obscured by the back of the case. Its also pulling hot air off the front HS.

The air going into the front case is currently 6c I need to find a way to either get better Heatsinks on this thing, or increase the air flow, since the later is the only viable option, looks like i'll be getting a 2nd PSU and putting another fan in there.

been reading up on it, the Tj is 95c for the x5550, however anything above 90c is really bad for it.

Unless I got some Peltier plates and somehow inserted them into the HS...would have to find a way to insulate the cpu/ram board from moisture though.

BTW Part of that diagram is wrong. Apple removes the IHS off the top of the CPUs. Its said to increase the thermal conductivity between the HS and the Die, however it does not work as well as apple intended...bless their hearts for trying. The IHS should be made of pure copper,(albieit with counterproductive nickel or zinc coating so they can lazer names/models etc on it. can be easily sanded off with 2k grit and a glass panel)which will spread the heat over a greater area giving the HS more surface contact, more surface contact = more heat moved. Intel did that for a reason. Apple removed it to cut costs.

 

[thekev]

This would depend on the specification you're looking at.

For example, 90C may fit in the short duration spec (i.e. 8hr max intervals), but not the long term spec (24/7 operation @ 100% on all cores), which is only 67C IIRC.

So entering SMC and pushing the fan speeds may be advisable, depending on the duration of OP's usage that pushes the temps that high (personally, I'd recommend it).

Another thing that hasn't been mentioned, is the CPU coolers and fans may be caked with dust (which acts like a blanket and retains heat), and need to be cleaned.

Where are you finding your data for short vs. long term spec? I'd like to be able to look that up myself as sometimes I do have to run things that go on for a very long time.

With the dust idea though, wouldn't getting smc and cranking it up to the full 5000 RPM clean it out itself just because of the amount of airflow?

It doesn't work that way.

 

[nanofrog]

BTW Part of that diagram is wrong. Apple removes the IHS off the top of the CPUs.

Only on the 2009 DP systems. The SP and both SP and DP 2010 models ALL use the Integrated Heat Sink.

In theory, their claims are valid (removing the IHS can improve thermal transfer between the semiconductor and the cooler), but it's not drastic, and puts the CPU in greater danger physically (i.e. over-tighten the cooler screws, and break the actual CPU).

In the case of it being needed in a MP, it's total BS, as evidenced by the 2010 models that use IHS in the DP models as well.

As per why they did it, Apple (may have been Foxconn's initiative) was being cheap, not Intel (if you've noticed, the retaining mechanism that's usually part of the socket is missing as well). Apple had to custom order those parts without the IHS.

In the case of Intel, it may have saved a few cents per unit on production, but they required special handling that easily could have consumed those savings (i.e. must be kept separately from the rest of the production so they weren't mistakenly sent to another vendor).

Where are you finding your data for short vs. long term spec? I'd like to be able to look that up myself as sometimes I do have to run things that go on for a very long time.

It's in one of the datasheet downloads, and usually including those available publicly (obscure, so it will take some time to find, but it's usually there IIRC).

If separate, check the Thermal/Mechanical Data Sheet first. Otherwise, it's usually found in Part II.