[Video Added + Watercooling Results] Replacing the internal IHS TIM of an i7 3770K - Page 14

You obviously are oblivious...
Yes quite.... There is (supposed to be) LESS leakage with IB.... But I dont think you know what that means... Leakage is not meant as heat leakage but electrical leakage. Electrical leakage = Heat... Leakage in this case is like this: electrical energy escapes... it then transforms into heat energy when it is displaced... we see hotter thermals in our chips

So not to say that the added transistors packed into a smaller area are not creating a blanketing effect because they very well might but then again it could be a load of BS (IDK for sure)

But since the die shrink and 3d transistors are supposed to help control leakage we should be seeing much lower temps even if this blanketing is somehow occurring. As we are not it does seem quite plausible there is production issues to where these chips are some how causing MORE leakage which would generate MORE heat...
Edited by Jesse D - 5/13/12 at 5:29pm

You mean like this test:


Or this test:


I don't know why you want to continue saying MX-4 is better or even comes close to the same performance, because it is nowhere near Coollaboratory Liquid Pro. The only thing MX-4 has going for it is that it is cheap and generally performs better then the older silicone-based TIMs that cost similar.

You are once again OBLIVIOUS.

INTEL states they are less leagage.....WHAT MORE PROOF DO YOU WANT THAT IT IS NOT A LEAKAGE PROBLEM?

source


And where is the proof that it is the die? Everywhere. You can see that it is not enough area to dissipate heat because the chips will be at 80C and the heatsink will be cool to the touch or the rad will not be at all warm. That means there that there is not much heat transfer. And that is the problem. The reason there is little heat transfer is either because of the surface area being too small or the 3d transistors but it is NOT a leakage problem. These chips have less leakage.

I have proof from Intel that it isn't a leakage problem but you have ZERO proof it is and yet you keep going on with your false statement.

Heat leakage? I'm not sure what that is. It is electrical leakage we are talking about. THEY HAVE LESS LEAKAGE BUT LESS AREA TO DISSIPATE THE HEAT or the 3d transistors cannot dissipate heat very well because they are not flat. So they have less leakage but higher temps because of their not so great ability to dissipate the heat.

It doesn't matter how little leakage there is, if you can't dissipate the heat that is generated be it a lot or a little, the temperatures of the chip will rise.

Once again please show some PROOF it is leakage because Intel and other sources say otherwise and I'm pretty sure Intel would be a RELIABLE source.
Edited by icehotshot - 5/13/12 at 8:07pm

Nice graphs

Except for the fact that somehow MX3 beat mx2, mx4 and prolimatech not to mention CLMP somehow walloped IE by 1.5c when any other test has ever showed them either even or with IE taking the lead... I can make a graph that shows AS5 stomping anything else in the market but that wont make it true. Find a more reliable source for your TIM info

As far me saying mx4 was better... I never did. I simply said it would be close (eg within 1-1.5c)


Go back and read this post http://www.overclock.net/t/1256249/replacing-the-internal-ihs-tim-of-an-i7-3770k-testing-done/120#post_17229080 then re-read mine and you will see what I was responding to... And I never said heat leakage, I said electrical leakage will turn into heat... any physics or electrical class will teach you how electrical energy can become heat energy... Learn to read

And you have no proof from Intel. You only have what they said. AMD said bulldozer had higher IPC than Sandy, but that didnt make it true (nor valid proof)



Back to the OP poke, poke, got a plan for your direct die yet?

You did say "heat leakage" in your statement. I was just wondering why you thought I meant heat leakage or whatever, I'm not sure what that meant cause I don't think there is such a thing as heat leakage, that is why I questioned it.

And also by your logic, no one on this forum can prove anything by any internet article.....lol. I guess we will keep arguing forever.

And seeing how the chips get hot, yet the coolers are cool or room temp to the touch would suggest that there is a heat transfer problem that has not yet been seen before by any other die shrink.

You're joking right? You must be. You can't possibly believe that an intel press release from may of last year, note that I said may of last year before the production run for the commercially released chips had even started, that talks about the design benefits of trigate transistors as what you consider "proof". How that could possibly be considered proof is beyond logical reasoning.

Let's examine your reasoning (called the effect before cause reasoning). If I were to summarize your argument it would go like this: Ivy Bridge CPU's are hotter than Sandy Bridge CPU's. Ivy Bridge CPU's are smaller than Sandy Bridge CPU's. Therefore it's smaller size must be the cause.


I can apply that same principle to everything. I'll give you two examples and let's see how it works out.

Example 1:

John got into his first accident today. John just bought a new car yesterday. Therefore John's new care must be the cause of the accident.

Example 2:

The earth's temperatures are higher today than 2000 years ago. Jesus died 2000 years ago. Therefore Jesus's death is the cause of global warming.

All 3 arguments as you can see take two pieces of information that may or many not be related and without showing any valid proof it makes a difinitive conclusion.

Oh, and as for my proof? Fact 1 - Tri-gate draws more current than a traditional planar transistor (due to increased surface area created by the "fin").
Fact 2 - At high overclocks Ivy Bridge requires even higher voltage than sandy bridge at those same speeds.

Quote:
"3.Ivy Bridge usually runs using a lower operating voltage than Sandy Bridge. But to overclock it way up to 4.9GHz, you have to bump up the voltage. In fact you have to add more voltage to an Ivy Bridge processor than a Sandy Bridge processor to reach that level. Since power consumption is strongly related to operating voltage, this means you're adding more power consumption and heat."



Conclusion - When a CPU that is designed to require lower voltage than it's predecessor is in fact using higher voltage at the same overclocked speeds as it's predecessor the only explanation is that at higher clocks it's suffering from higher voltage leakage than designed and therefore is producing higher heat.

Another fact - at stock speeds Ivy Bridge requires less voltage and produces less heat than Sandy Bridge. If die size were indeed the problem than it wouldn't matter what speed the CPU were operating at, clock for clock it would ALWAYS be higher than a non 22nm chip.
Edited by Bubba Hotepp - 5/13/12 at 10:00pm

Please keep this conversation civil. 

I almost forgot. You keep mentioning "the heatsink is cool to the touch". That's impossible. Want to test it out? Take a soldering iron, touch just the tip (much smaller contact area than the IB die surface area) leave it there for 10 minutes and then touch the heatsink and tell me that it's "cool to the touch".

Considering some soldering irons can hit an upwards of 1000C, I don't think that is a good example, even normal soldering irons hit almost 500C. The CPU will go into thermal throttle at 105C so unless he uses like a 15W soldering iron I don't think the comparison is fair..

if the solder Intel has traditionally used has a metal component to it, then I can see why TIM Is worse, since TIM usually contains less conductive metal oxides.

Solder properties: http://www.electronics-cooling.com/2006/08/thermal-conductivity-of-solders/

Thermal greases: http://en.wikipedia.org/wiki/Thermal_grease

There's about a factor of ten drop, but if we assume the thickness of the TIM is also less, then Intel could have assumed it would balance out. Apparently, it did not (basically, if k, the heat transfer coefficient, goes down by 10x, then dx, the thickness of the material, must also go down by 10x assuming the area and temperature difference remain roughly constant).