Delidded my i7-3770K, loaded temperatures drop by 20°C at 4.7GHz

Page 10 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
I've never seen a paste do that before. Have you?

I have not seen NT-H1 do that, and I've used NT-H1 in situations all the way up to 105°C.

There isn't anything special about sitting on the CPU die when it comes to temperature, so I am surprised you observed the separation.

I wonder if you have a potentially bad tube of NT-H1? Is it old? (>6 months)

Long ago, like 4yrs ago, I had a 2yr old tube of AS5 that had separated on me from sitting on the shelf for so long. I went to put some on my IHS and a bunch of liquid squirted out, reminded me of when I don't shake ketchup bottle when it has been sitting for a while.

But after 2yrs I imagine just about any TIM will separate under the influence of gravity as these pastes are all emulsions. That doesn't explain why yours separates in a matter of months and after application, head-scratcher there.
 

Ferzerp

Diamond Member
Oct 12, 1999
6,438
107
106
There was a bit of liquid that came out when I squirt it initially (but my AS5 does that as well, just a little less).

It may be a bad tube then. I bought it off Amazon back in May. It was Amazon direct and not a 3rd party (or even a 3rd party fulfilled via amazon), but stranger things have happened. Maybe I'll get brave and get a metallic one. The instability of my TIM under my IHS is why I haven't put my second application of IX between the IHS and the H100 block yet.
 

HondaCop

Member
Aug 4, 2012
42
0
0
Since we are on the subject of TIMs, let me tell you that if you plan on using Liquid Metal Ultra, make sure you don't plan on messing around with your setup in a long time, unless you want to re-lap your IHS and heatsink everytime you take it apart.

I had my LMU on for a week and after taking it apart, I had to re-lap both the IHS and heatsink. If you don't, then when you attempt to put LMU on again, it will not adhere to the surface. You will keep working it in with the little brush but it will just stay in one place.

Just a heads up, folks. Use LMU, prepare to re-lap each time you take it apart.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
IDC i would like to ask if you could measure the Core i7 2600K at default settings and the Core i7 3770K (better TIM between the die and IHS) at default settings but with the default Heat-sink+fan in both of them .

I would like to see whats the impact of the smaller die size in operational temps if any at default settings and cooling solutions.

Hi AtenRa,

Here are the results, for reference please see this post of mine regarding some questions I had.

StockHSF-comparing2600kvs3770k.png


For 3.8GHz I kept the Vcc the same for both processors, obviously way more voltage than necessary for the 3770k and the temps show it.

For 3.9GHz I used the optimal voltage for both processors, which reflects in the operating temperatures and power consumption numbers.

It was interesting to run these tests because one thing I learned is just how bad, truly awful, the stock HSF is. Consider these results, with the IB delidded and CPU TIM replaced but I then bolted the stock HSF onto it the temps hardly improved at all over those observed before I delidded the CPU.

Delidding the CPU appears to only help if you also replace the stock HSF with something that can take advantage of the improved thermal conductance.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
Hi AtenRa,

Here are the results, for reference please see this post of mine regarding some questions I had.

StockHSF-comparing2600kvs3770k.png


For 3.8GHz I kept the Vcc the same for both processors, obviously way more voltage than necessary for the 3770k and the temps show it.

For 3.9GHz I used the optimal voltage for both processors, which reflects in the operating temperatures and power consumption numbers.

It was interesting to run these tests because one thing I learned is just how bad, truly awful, the stock HSF is. Consider these results, with the IB delidded and CPU TIM replaced but I then bolted the stock HSF onto it the temps hardly improved at all over those observed before I delidded the CPU.

Delidding the CPU appears to only help if you also replace the stock HSF with something that can take advantage of the improved thermal conductance.


Hello IDC,

Thx for the reply and the tests.

Actually i wanted to see at default settings for both CPUs, default voltages and frequencies (turbo on) on both processors. But its ok, the more data we have the better to understand how those two CPUs behave.

Taken the data at 3,9GHz that closer resemblance the default settings of the CPUs (voltages), i will conclude that Ivy is cooler than SB with the better TIM and it would have even lower temps if it was soldered like SB is.

Higher voltages seams to have more thermal impact on Ivy than SB and that could be due to two factors.
First is Intel's Tri-Gate transistors, they are designed for lower voltage operation. Increasing voltages above 1,1V may dramatically increase current (I) and thus temperature.

http://www.anandtech.com/show/5763/undervolting-and-overclocking-on-ivy-bridge
Temperature.png


This is the reason i have asked to test at default settings (at or bellow 1.1v for Ivy), because of the difference in transistor design and electrical characteristics, testing at same voltages above 1.1v will only make things worst for processors with Tri-Gate transistors.

The second factor is the smaller die size that creates warmer hot spots. Increasing Voltages elevates hot spot temperatures only making things worst and the default cooling solution (TIM, Heat-Sink Fan) is not adequate for that.

As a conclusion, i could say that Tri-Gate mechanical (shape of fins) and electrical (V and I) characteristics have bigger impact in thermal behavior of Ivy than die size.

http://www.chipworks.com/en/technic...4/intel’s-22-nm-tri-gate-transistors-exposed/

FIN-410K-c_branded-300x300.png


current_density_animation.gif


rolloff.gif


If you remove the IHS, the die size importance in thermal characteristics will elevate due to non uniform power maps created on the die surface. The IHS plays a big role eliminating the non uniform power map.
 

Akantus

Member
Apr 13, 2011
80
0
0
If you remove the IHS, the die size importance in thermal characteristics will elevate due to non uniform power maps created on the die surface. The IHS plays a big role eliminating the non uniform power map.

I would say, this could be true, if you were running w/o heatsink. But if you install heatsink on top of bare die it will act exactly as IHS, only the temps will be better, because it doesn't have to go through two more layers of "insulation" (tim under IHS and IHS itself).
Unless there are big differences between materials of IHS and HSF - e.g. copper and wood. :eek:
 

Plimogz

Senior member
Oct 3, 2009
678
0
71
I'd expect that the IHS definitely has practical thermal advantages when using the all aluminum heatsinks that are commonly packaged with lower price/TDP parts. And coupled with the physical protection it affords the die from improper or heavy handed installation, I would guess that it makes good economic sense for both manufacturers to prefer installing them, even on SKU's that are bundled with better heatsinks which already have a copper base or slug installed to serve the same purpose as the IHS's name suggests.
 

Ferzerp

Diamond Member
Oct 12, 1999
6,438
107
106
I would say, this could be true, if you were running w/o heatsink. But if you install heatsink on top of bare die it will act exactly as IHS, only the temps will be better, because it doesn't have to go through two more layers of "insulation" (tim under IHS and IHS itself).
Unless there are big differences between materials of IHS and HSF - e.g. copper and wood. :eek:


Correct. If IHS with paste helped cooling at all, then why not just stack IHS and make it IHSs all the way down. Absurd, yes, but it points out that flaw in thinking. If 1 IHS helps, 10 would help a whole lot... Of course, an IHS doesn't help. It's a convenience and durability thing.

*IHS can help if soldered on, but not with paste. The only reason this is the case is the much lower thermal resistance of the solder used paste. If you could solder the heatsink/waterblock on to the die, performance would be even better. Given equal TIMs, an IHS is always a cooling detriment, not a benefit because it forces you to create two areas of high resistance (layers of TIM) than just one (one layer of TIM)
 
Last edited:

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
I'd expect that the IHS definitely has practical thermal advantages when using the all aluminum heatsinks that are commonly packaged with lower price/TDP parts. And coupled with the physical protection it affords the die from improper or heavy handed installation, I would guess that it makes good economic sense for both manufacturers to prefer installing them, even on SKU's that are bundled with better heatsinks which already have a copper base or slug installed to serve the same purpose as the IHS's name suggests.

That is correct, the IHS was made in order for a more economical cooling solution like the Intel default CPU coolers. You would need a much bigger and more expensive heat-sink to get the same results without the IHS.
 

Akantus

Member
Apr 13, 2011
80
0
0
That is correct, the IHS was made in order for a more economical cooling solution like the Intel default CPU coolers. You would need a much bigger and more expensive heat-sink to get the same results without the IHS.

Not sure about that.
Thermal conductivity of copper is ~400 W/(m·K), aluminum is ~250 W/(m·K). For example cured AS5 (whis is better than what intel uses in IB) is 8.7 W/(m·K), so really IHS is mainly just about protecting the die.

IHS could help with heat dissipation if you had improperly installed heatsink (one pin not pushed, so it isn't sitting even), where bare die would hit thermal limits almost instantly.
 

Denithor

Diamond Member
Apr 11, 2004
6,300
23
81
Not sure about that.
Thermal conductivity of copper is ~400 W/(m·K), aluminum is ~250 W/(m·K). For example cured AS5 (whis is better than what intel uses in IB) is 8.7 W/(m·K), so really IHS is mainly just about protecting the die.

IHS could help with heat dissipation if you had improperly installed heatsink (one pin not pushed, so it isn't sitting even), where bare die would hit thermal limits almost instantly.

This. Except in that one case (improperly attached HSF) the IHS actually slows the transfer of heat out of the core because you have to use TIM which is much, much less efficient than metal at passing through the heat to the HSF.

Direct contact of a copper HSF to the die is the optimal situation for cooling. Next best would be die->solder->IHS->solder/liquid metal TIM->HSF. The best we can expect with IB while using the IHS is die->LM->IHS->LM->HSF.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,699
1,448
126
I hope to see more; I could bet that I will, since there was work-in-progress and the OP has proven to be better than his word. But just in case, I've bookmarked this thread for future reference.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Yeah more is coming, right now I'm working on constructing the 0.2mm thick metal shim so we can generate the right kind of apples-to-apples data for comparison to the stock CPU TIM.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,699
1,448
126
Yeah more is coming, right now I'm working on constructing the 0.2mm thick metal shim so we can generate the right kind of apples-to-apples data for comparison to the stock CPU TIM.

I see your point about that. That's a pretty thin wafer of metal. I guess you could almost call it "foil" . . . .
 

Ferzerp

Diamond Member
Oct 12, 1999
6,438
107
106
What I would like is an IHS that instead of being open and flat, actually made contact with the sides of the die too. Something that had a die shaped notch in it, and touched the die from all 5 sides. Sure, you couldn't do much with this as the die doesn't stick up very far, and there is that rim of whatever material around it, but every little bit helps, right?

Or since there is nothing conductive in there, fill it up with liquid metal, and seal the IHS down.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,699
1,448
126
What I would like is an IHS that instead of being open and flat, actually made contact with the sides of the die too. Something that had a die shaped notch in it, and touched the die from all 5 sides. Sure, you couldn't do much with this as the die doesn't stick up very far, and there is that rim of whatever material around it, but every little bit helps, right?

Or since there is nothing conductive in there, fill it up with liquid metal, and seal the IHS down.

I was thinking you could stuff it with diamond paste instead of the liquid metal concoctions. I'm just not sure that it would improve anything: it would only increase the area of contact by an area of 1/8" x 2L x 2W. Those surfaces may not radiate nearly as much heat as the flat surface.

Best to focus on the contact with the IHS or HSF-base, and the TIMs used.
 

Phil L

Member
Jun 12, 2011
41
1
66
Yeah more is coming, right now I'm working on constructing the 0.2mm thick metal shim so we can generate the right kind of apples-to-apples data for comparison to the stock CPU TIM.

200 um is very close to 8 mil (i.e. 8 thousandths of an inch, or 203.2 um), and you can buy shim stocks of various thickness and different materials (copper, steel, stainless steel, aluminum, titanium, and so on). You can buy a uniform thickness roll (typically in 6" x 50" or 15.24 cm x 127 cm), or an assortment of shim stock package (multiple sheets of different thickness, at 6" x 12" or 15.24 cm x 30.48 cm size). Of course in either case you'll end up with far more shim stock then you would ever need for just cpu shimming, but it's an option if you want to explore it.

Just a thought,
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
200 um is very close to 8 mil (i.e. 8 thousandths of an inch, or 203.2 um), and you can buy shim stocks of various thickness and different materials (copper, steel, stainless steel, aluminum, titanium, and so on). You can buy a uniform thickness roll (typically in 6" x 50" or 15.24 cm x 127 cm), or an assortment of shim stock package (multiple sheets of different thickness, at 6" x 12" or 15.24 cm x 30.48 cm size). Of course in either case you'll end up with far more shim stock then you would ever need for just cpu shimming, but it's an option if you want to explore it.

Just a thought,

Oh yeah, you and yottabit must have great minds that think alike :thumbsup: Based on yottabit's earlier, but essentially identical, tip regarding the shim stock, I previously purchased some.

It is here already, actually, and I've given it a go once already to make a shim template but the shim stock is very soft and bendy. It reminds me of really really thick aluminum foil. Cuts easily with a pair of scissors. (I did confirm it it 0.20mm shim with my digital calipers)

So today I will make another, gentler, attempt. LOL @ me, I assumed this stuff was going to be really arduous like the steel stock we worked with in college so I started out with a hack saw :eek: Imagine a hack saw on aluminum foil - shredded mess ensues - and that pretty much captures what happened for me with the first attempt at making my IHS shim.
 

Rvenger

Elite Member <br> Super Moderator <br> Video Cards
Apr 6, 2004
6,283
5
81
IDC and Ferzerp. Found this over at HardOCP.





Quote:
Originally Posted by ocbaud
a huge problem i'm having after doing this to my 3570k @ 4.5ghz

using noctura paste on the core and the IHS with good water cooling.

After first doing this, i get temps idling around 31c-32c on all cores.
Load temps with intel burn test would show 62c max on any given core.

after a few weeks of this computer running almost 24/7 (not under load, mostly idle)
I get the same idle temps, but load temps hit 90c on a couple cores(i stop IBT before it goes on longer than a couple seconds)

this has happened 3 times now and each time, after i pull the ihs off and replace the thermal paste, the temps come down again.

it makes great contact at first, but given enough time it stops working well.

what could i be doing wrong?
 

BonzaiDuck

Lifer
Jun 30, 2004
15,699
1,448
126
IDC and Ferzerp. Found this over at HardOCP.


Quote:
Originally Posted by ocbaud
a huge problem i'm having after doing this to my 3570k @ 4.5ghz

using noctura paste on the core and the IHS with good water cooling.

After first doing this, i get temps idling around 31c-32c on all cores.
Load temps with intel burn test would show 62c max on any given core.

after a few weeks of this computer running almost 24/7 (not under load, mostly idle)
I get the same idle temps, but load temps hit 90c on a couple cores(i stop IBT before it goes on longer than a couple seconds)

this has happened 3 times now and each time, after i pull the ihs off and replace the thermal paste, the temps come down again.

it makes great contact at first, but given enough time it stops working well.

what could i be doing wrong?

How is the Noctua TIM more of a grease than a paste? The original Intel stuff that IDontCare shows looks like stiff putty. I think somebody on the HardOCP forum mentioned IC Diamond, which is already like stiff cement or putty, and dries out. The diamond stuff doesn't flow, though.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
IDC and Ferzerp. Found this over at HardOCP.

Quote:
Originally Posted by ocbaud
a huge problem i'm having after doing this to my 3570k @ 4.5ghz

using noctura paste on the core and the IHS with good water cooling.

After first doing this, i get temps idling around 31c-32c on all cores.
Load temps with intel burn test would show 62c max on any given core.

after a few weeks of this computer running almost 24/7 (not under load, mostly idle)
I get the same idle temps, but load temps hit 90c on a couple cores(i stop IBT before it goes on longer than a couple seconds)

this has happened 3 times now and each time, after i pull the ihs off and replace the thermal paste, the temps come down again.

it makes great contact at first, but given enough time it stops working well.

what could i be doing wrong?

I don't discount it, I believe it. What I take from this is that NT-H1 is simply incompatible with the thermomechanical environment that exists in direct-die applications.

For example we know the mismatch in coefficients of thermal expansion will be much larger between that of the silicon Die and the metal IHS versus that which exists between a metal IHS and the metal base of an attached HSF/water-block.

Additionally we know the wettability of the silicon die is markedly different than that of the metal surfaces for which NT-H1 was developed to adhere to. This makes issues of viscosity and surface adhesion a factor when matters of thermomechanical forces are a concern, which they are.

So what does this boil down to in terms of why NT-H1 could be giving rise to these results? Two things can be happening, and possibly in combination.

One is the push-pull pumping effect that comes with the mechanical expansion and contraction of the silicon, the TIM, and the overhead IHS. Eventually, given time, the TIM itself will be squeezed out of the gap that it occupies between the die and the IHS and an air gap will be left in its place. The air gap most likely forms when the CPU cools down.

Push-pumpTIMgap.png


Then when the CPU heats back up, the air gap prevents the IHS from heating up in concert with the die and the TIM, resulting in an IHS that is not expanding to fill in the air gap, leaving the CPU much hotter than before.

The second point of concern is that of emulsion separation - the molecular components that comprise NT-H1 may experience separation at the behest of the silicon die interaction (it will actually be a silicon nitride at the surface). NT-H1 was probably not designed to remain an emulsion when in contact with silicon as it was designed to remain an emulsion when in contact with copper, aluminum, or nickle (more specifically, the metal oxides of those elements).

So a chemical separation may be the culprit as well, and likely it is a combination of both.

If it is just chemical separation then choosing a different TIM will be the solution.

If it is is thermomechanical push-pump that is the culprit then going with the more permanent metal-TIMs (Indigo Xtreme or Liquid Ultra) would be the solution.

In either case, going with the more permanent metal-TIMs would appear to be the one-size-fits-all solution provided the thermomechanical stresses that will then arise from having the mismatch in coefficients thermal expansion do not cause issues for the IB packaging itself (ala bumpgate dejavu).
 

BonzaiDuck

Lifer
Jun 30, 2004
15,699
1,448
126
I figured that if you spent the $25-bucks, you're going to try the diamond stuff? One wonders if that would leave a stable result . . .