I just came across this interesting thread comparing different thermal pastes and other TIMs on a delidded 3770K over in the anandtech discussion forums. The tests were done on a chip die without replacing the lid, however.

 

http://forums.anandtech.com/showpost.php?p=34297433&postcount=67

 

The whole thread is worth a read. I found several useful pointers there.

One of the other things the experimenter found was that using larger amounts of TIM unexpectedly reduced temperatures by a few degrees C. (But apart from the cost, there are risks associated with doing that with electrically-conductive TIMs).

http://forums.anandtech.com/showpost.php?p=34310264&postcount=106

I get lower temps with WAY more voltage and the same thermal load (IBT maximum or Prime 95 blend for hours and hours) with a "real" water cooling solution and lapped IHS and block vs. direct mounting.  Load temps never exceed 78-80 C @ 4.9GHz and 1.5V.  Sounds like either direct die mounting is not the way to go or the H100 just isn't up to the task.  

I get lower temps with WAY more voltage and the same thermal load (IBT maximum or Prime 95 blend for hours and hours) with a "real" water cooling solution and lapped IHS and block vs. direct mounting.  Load temps never exceed 78-80 C @ 4.9GHz and 1.5V.  Sounds like either direct die mounting is not the way to go or the H100 just isn't up to the task.  

 

I was hoping to draw attention to the differences in efficacy of the various TIMs, rather than to promote direct die mounting per se.

By way of news about Haswell from Tweaktown, we have learned that Intel is acknowledging the height of the IHS on Ivy Bridge is too high above the CPU die, which is why Intel used a thermal paste in IVB rather than soldering as was common with previous parts.  I see this as a manufacturing issue which could have been addresssed with IVB, too bad Intel didn't agree but at least they're learning from their mistake and fixing it for Haswell.  

I just delidded my 3570k with a hammer and vise, couldn't have been easier. Beforehand I was getting 83c with Intel Burn Test.... now my max is 59, that is at 4.5ghz and 1.29v . Now to see what I can get out of it to even further my fsx enjoyment.

 

 

 

 

 

 

Sent from my SPH-L900 using Tapatalk 2

Still remember we used to sand down the 486 DX2 processors to gain a couple of Mhz

By way of news about Haswell from Tweaktown, we have learned that Intel is acknowledging the height of the IHS on Ivy Bridge is too high above the CPU die, which is why Intel used a thermal paste in IVB rather than soldering as was common with previous parts.  I see this as a manufacturing issue which could have been addresssed with IVB, too bad Intel didn't agree but at least they're learning from their mistake and fixing it for Haswell.  

 

Wow I think you just solidified my decision to wait for Haswell and NOT buy an Ivy Bridge right now.

AFAIK Haswell uses the same heatsink mounting as IB.  IOW not a soldered heatsink.  That said the same inferior thermal paste is probably still used whether the heatsink is soldered or not.  

 

A side note to this is that there are commercial coolers available that mount directly on the die without the heatsink.  Since I already have a unlidded IB I may just pop off the heatsink and give it a go.  On the other hand I believe that the heatsink may serve a function and that is to spread heat evenly to the cooler head.

But still is it just masking and would it be better to have the die in direct contact with the cooler cores?

 

Decisions decisions...

 

Cheers

jja

AFAIK Haswell uses the same heatsink mounting as IB. IOW not a soldered heatsink. That said the same inferior thermal paste is probably still used whether the heatsink is soldered or not.

 

 

 

My impression from Techguy's post was that Intel basically "measured wrong" on the clearance of the ihs on ivy bridge and have fixed it on Haswell. Maybe I misunderstood the tweak town link though.

Hey! Where'd all the pictures go?

 

Still working on putting the pieces together for a 3570k system and I am wondering if anyone has used the Indigo Extreme products for between the IHS and heat sink (water block for me)? I'm pondering over using Liquid Ultra on the cpu die and then the Indigo sitting on top. Workable or not?

Good question, I don't see them either and I can't edit the original post.  I'll see if Stephen can fix it.  

That is a definite maybe!

AFAIK Haswell uses the same heatsink mounting as IB. IOW not a soldered heatsink. That said the same inferior thermal paste is probably still used whether the heatsink is soldered or not.

 

A side note to this is that there are commercial coolers available that mount directly on the die without the heatsink. Since I already have a unlidded IB I may just pop off the heatsink and give it a go. On the other hand I believe that the heatsink may serve a function and that is to spread heat evenly to the cooler head.

But still is it just masking and would it be better to have the die in direct contact with the cooler cores?

 

Decisions decisions...

 

Cheers

jja

I'd just mix some SPF150 suntan lotion, WD40 and spread it liberally all over the top of the chip, why stick with paste?

 

Still working on it! Phew!

Hey! Where'd all the pictures go?

 

 

Good question, I don't see them either and I can't edit the original post.  I'll see if Stephen can fix it.

 

 

Still working on it! Phew!

There you are sirs. Almost as good as new, thanks to Paul J who had copies in his Avsim Guide!

 

Kind regards,