Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

SDAVE

macrumors 68040
Original poster
Jun 16, 2007
3,579
601
Nowhere
I searched and people seem to be doing the pea-sized ones with spreading it with a credit card and sometimes not spreading it. I tried both and it seems that I idle about 55-60c for dual x5680s 12core. I have to use iStat menu to up the intake/exhaust & cpu cooler fans to get it down to 42c idle.

Should I reapply the paste with the X diagonal spread with my arctic silver 5?


Cheers
 
Don't overthink this. As long as IHS is fully covered with the compound and its layer isn't too thick, you're good to go. No matter which method you'd use.
 
If you look at my mod thread, under CPUs item 4 I have a direct link to Arctic Silver's pdf with their instructions.

Nice I think I followed the AS pdf...

Don't overthink this. As long as IHS is fully covered with the compound and its layer isn't too thick, you're good to go. No matter which method you'd use.

I tried 2 approaches so far, and they're both equal in terms of thermal compounds.

One is peasized just dabbed on and allowed heatsink to flatten it. The other was peasized and then spread across with no bubbles with a credit card.

Do you guys think I should use another compound? AS (reading online) seems that they dry too fast.

I'm open to re-applying I just want a surefire method this time. I don't think 55-60c is good for idling at Mac Pro default fan speeds.

I also think I scratched the heatsink surface a bit so maybe there are bubbles??
 
AS needs curing time (see specs for how long it should be, I don't remember now) to get its full performance. There's a few other Arctic products which do not require curing. I'm recently using MX-2, because I didn't see the difference between its newer versions (MX-3, MX-4). According to MX specs, it should last 8 years, while AS less than 5 IIRC.
 
Yes, AS states around 200 hours burn in (but that means normal operation, no need to benchmark or anything like that).
 
AS needs curing time (see specs for how long it should be, I don't remember now) to get its full performance. There's a few other Arctic products which do not require curing. I'm recently using MX-2, because I didn't see the difference between its newer versions (MX-3, MX-4). According to MX specs, it should last 8 years, while AS less than 5 IIRC.

I see, I've been using AS since like early 2000 when I used to build PC's and use large heatsinks. Never had a problem.

I do have the MX2 I think somewhere laying around, should I use that instead?

I used the Apple technician's manual technique, seems to work fine.

I don't get what they mean here, do they mean just fill that whole area or just a straight line in the middle?
 

Attachments

  • Screen Shot 2015-03-29 at 12.34.53 AM.png
    Screen Shot 2015-03-29 at 12.34.53 AM.png
    638.2 KB · Views: 127
As you see in the AS guide, the IHS covers a lot more area than the underlying cores. Still assuming that the IHS does its work and distributes the heat, you'll want a thin even layer over the entire IHS.

The guide you show means that you should apply just the thin line and the circle. The preassure of combining the heatsink with the IHS will automatically distribute the compound.

You will want the long thin line of paste to be oriented correctly. See AS guide for reference.
 
The Apple guide means clean up the CPU surface, and apply the thermal paste like a stop sign (N.B. the CPU is rotated in 90 degree).

Anyway, the method of applying thermal paste isn't that important. Tests show that the difference usually within 1 degree, that's well within normal variation.

If you try different method and still get more or less the same number. That should be something else to cause the high idle temp. e.g. Poor air flow, high ambient temp, etc.
 
As you see in the AS guide, the IHS covers a lot more area than the underlying cores. Still assuming that the IHS does its work and distributes the heat, you'll want a thin even layer over the entire IHS.

The guide you show means that you should apply just the thin line and the circle. The preassure of combining the heatsink with the IHS will automatically distribute the compound.

You will want the long thin line of paste to be oriented correctly. See AS guide for reference.

Ah so just draw with lines...I thought fill the line or do a line and spread around :)

The Apple guide means clean up the CPU surface, and apply the thermal paste like a stop sign (N.B. the CPU is rotated in 90 degree).

Anyway, the method of applying thermal paste isn't that important. Tests show that the difference usually within 1 degree, that's well within normal variation.

If you try different method and still get more or less the same number. That should be something else to cause the high idle temp. e.g. Poor air flow, high ambient temp, etc.

The machine is out in the open, room isn't that hot. I kicked up the fans in iStat menu so they are a bit audible now, but I'd like to turn off iStat menu and leave it as it is, but the speeds jump up too much.

Weren't the x5680s and 90s never supported? since the TDP was so high on them. I wish i had the x5690s though to max this machine out...these CPU's are so speedy (they are about 5 years old now).

I also have a GTX780 in this machine so I assume that's also causing some heat even at idle.
 
I've never used a credit card. I've put a tiny amount on, half pea size then put a glove on (medical type) and spead the paste over the core. The only time I'd use a pea size amount is on a for with ihs installed.
 
Last edited:
I searched and people seem to be doing the pea-sized ones with spreading it with a credit card and sometimes not spreading it. I tried both and it seems that I idle about 55-60c for dual x5680s 12core. I have to use iStat menu to up the intake/exhaust & cpu cooler fans to get it down to 42c idle.

Should I reapply the paste with the X diagonal spread with my arctic silver 5?


Cheers

Would you mind posting the full set of your temperature numbers from iStat? They can sometimes be confusing, and I'm wondering if you are actually idling that hot.
 
A small dab and spreading with a credit card has always worked well for me.

Lou
 
Would you mind posting the full set of your temperature numbers from iStat? They can sometimes be confusing, and I'm wondering if you are actually idling that hot.

This is with about 65-75% CPU usage. And I have manually increased the CPU fans and intake/exhaust to a point where it's fairly audible.

I think I just have to reapply, somethings definitely wrong.
 

Attachments

  • Screen Shot 2015-03-29 at 4.53.20 PM.png
    Screen Shot 2015-03-29 at 4.53.20 PM.png
    103.5 KB · Views: 119
The method I have used for decades is to put a small amount on both surfaces and use a credit card to sweep it full across in both the x and y directions to fill the micro-pits in the metal, leaving as thin a film as possible ... then mate the parts and secure with whatever clips, screws, clamps are involved. The parts must be clean before starting.

This hasn't let me down yet!
 
So it looks like i made a doodoo and I accidentally swapped the heatsinks so CPU A and CPU B were using the wrong heatsinks!

Cleaned everything up and just applied a pea-sized amount in the middle of CPU and followed instructions in 2010MP service manual of screwing the heatsink via 1,2,3,4 screws (X) and all is ok now. I removed iStat and it's idling at around 40C for both CPUs (one is a few C's off but that's understandable.)

Anywho thanks guys for the help!

Also after removing iStat menu the CPU is around 81C at 80% load. MPro just automatically handles it. Ok tired of the technical stuff, now back to work :)

Current temps at 80% load. Are these good? It can run at these temps and fan speeds all day long.
 

Attachments

  • Screen Shot 2015-03-29 at 8.06.05 PM.png
    Screen Shot 2015-03-29 at 8.06.05 PM.png
    149.9 KB · Views: 94
Last edited:
Does anyone have a similar setup they can post the idle temps and 80+% usage temps?

Mine are like this at idle now, it fluctuates. I don't know why CPU B and CPU A are so different. I applied AS the same way. Maybe I should let them burn in 200 hours.
 

Attachments

  • Screen Shot 2015-03-30 at 12.44.46 AM.png
    Screen Shot 2015-03-30 at 12.44.46 AM.png
    51.7 KB · Views: 106
Difference between A and B is perfectly normal. See how CPUs are located on dboard and you'll understand ;)
 
Difference between A and B is perfectly normal. See how CPUs are located on dboard and you'll understand ;)

So these look ok to you?
Don't the CPU's burn out at different times? :)

I don't care though, this machine will probably sold soon anyway. It's too big :D

But it's a beast nonetheless....
 
That's a new one on me:eek: I've never seen that recommended. Seems like to much paste.

Lou
You need to fill the micro-pits from the machining on both surfaces. The film you leave is really thin, just a "molecule" thick film is all you need. The "goup" you see in the disassembly pictures posted here is way too thick, but probably lends itself to fast mass-assembly.
 
You need to fill the micro-pits from the machining on both surfaces. The film you leave is really thin, just a "molecule" thick film is all you need. The "goup" you see in the disassembly pictures posted here is way too thick, but probably lends itself to fast mass-assembly.

I'm about to upgrade my 4core 1,1 to an 8core. Here is a good youtube video I found. He uses your method to fill the micro-pits on both surfaces.

https://www.youtube.com/watch?v=wmz7gPw31L0
 
Can anyone elaborate on the best technique for CPU's that have the IHS removed? I just purchased a pair of x5690's for my upgrade, and I spent a little extra money to get a pair that have the IHS removed. I'd imagine that a small dot of thermal paste directly on the die would be best..

Andrew
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.