If anyone wonders if the IHS and bad TIM are why IB can't be cooled. (hint, yes)

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

MisterMac

Senior member
Sep 16, 2011
777
0
0
I guess the summary so far - really is, the tim choice may not be optimal but at the same time - the not soldering the ihs to the die to create more snug fit is really balling on high voltage\ high clocks.


It looks like from ferzerp's adventures it's more non-soldering than TIM tho.
 

Kenmitch

Diamond Member
Oct 10, 1999
8,505
2,250
136
Exactly. But when he went bare die, he got roughly 15.75C better, so that's a big improvement.

It also kinda points the finger at the gap more than the TIM it looks like. I guess it could vary from chip to chip somewhat.
 

Avalon

Diamond Member
Jul 16, 2001
7,571
178
106
It makes me wonder if the TIM is fine, but the IHS sucks?

No, the TIM definitely sucks. I had a 5-6C drop just replacing the TIM with MX2 and putting the IHS back on. Ferzerp had even better success not only replacing the TIM but sanding down the contact area on the IHS to lower it closer to the die. I will probably do this myself later on down the road.
 

bononos

Diamond Member
Aug 21, 2011
3,939
190
106
One thing to keep in mind is the length of time that the thermal paste needs to be reapplied because of oil etc separation issues. I wouldn't want to have to dig out the cpu and delid every year just to keep things working.

PK-1 recommends reapplying every year but their paste is thick and viscous iirc like Noctua's HT-N1 (recommended reapply every 3yrs) or AS5 so it probably doesn't separate that easily and the 1yr recommendation is for best performance. The compound that I know with the longest shelf life is mx-4 (8yrs) and Artic says it does not need reapplication because it is different from metal/silicon compounds.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
One thing to keep in mind is the length of time that the thermal paste needs to be reapplied because of oil etc separation issues. I wouldn't want to have to dig out the cpu and delid every year just to keep things working.

PK-1 recommends reapplying every year but their paste is thick and viscous iirc like Noctua's HT-N1 (recommended reapply every 3yrs) or AS5 so it probably doesn't separate that easily and the 1yr recommendation is for best performance. The compound that I know with the longest shelf life is mx-4 (8yrs) and Artic says it does not need reapplication because it is different from metal/silicon compounds.

Good point, I have some PK-1 coming in tomorrow (thanks Newegg for fast shipping) but after learning about the 1 year thing, I am no longer considering de-lidding. Heck I might just not even apply the PK-1 until I really need to in later years when stock is no longer fast enough.
 

bononos

Diamond Member
Aug 21, 2011
3,939
190
106
Good point, I have some PK-1 coming in tomorrow (thanks Newegg for fast shipping) but after learning about the 1 year thing, I am no longer considering de-lidding. Heck I might just not even apply the PK-1 until I really need to in later years when stock is no longer fast enough.
What I meant in my post is that the really short 1yr PK-1 recommendation could be for the best performance only and it could still be pretty good after a few years.

How difficult is procuring a very small amount of solder paste and applying+baking it?
 

heymrdj

Diamond Member
May 28, 2007
3,999
63
91
One thing to keep in mind is the length of time that the thermal paste needs to be reapplied because of oil etc separation issues. I wouldn't want to have to dig out the cpu and delid every year just to keep things working.

PK-1 recommends reapplying every year but their paste is thick and viscous iirc like Noctua's HT-N1 (recommended reapply every 3yrs) or AS5 so it probably doesn't separate that easily and the 1yr recommendation is for best performance. The compound that I know with the longest shelf life is mx-4 (8yrs) and Artic says it does not need reapplication because it is different from metal/silicon compounds.

Glad someone is looking at it this way instead of just raw clock speed whining.

I could run a dragster engine, and get from Point A to Point B (on empty roads) faster...if it wasn't for that issue of needing a rebuild every 120 seconds.

This new process is a new beast, and the things we understood before may not necessarily work the same now. The IB heat cycle is definitely increased, old TIM may not have been capable of handling it for years on end. Remember Intel isn't building something to work for a year until the next processor comes out just so you can get your overclocking jollies off, they are building 99% for the OEM's, the ones that sell these computers by the thousands day in and day out and their users (especially business) expect a 5 year life cycle out of them. Intel does care about the enthusiast, but not as much as you all would like them to. They are still in it for the profit and the mainstream, where most of the profit lies. I trust the Intel engineers on their choice of TIM over some armchair engineers. As for the gap, my guess is the shrunken die caused the gap and they didn't want to change from the IHS height that most of the coolers are designed for because numbskulls would tighten them down too much. Complete shot in the dark. If they did it for that reason, then Ivy still performs exactly as it is supposed to. Intel isn't there for the overclocking 1%, they are there for the mainstream. And these processors will be reliable work horses day in and day out, and that's all that matters.