Overclock.net banner

Why the hate on Intel's TIM?

6K views 5 replies 5 participants last post by  czin125 
#1 ·
It just seems like every thread that has to do with temperatures on Intel's mainstream CPU's that someone is knocking on intel because they use paste instead of soldering. After doing a bit of research, there are more variables than just TIM which causes high temperatures, and a whole lot of reasons why they aren't soldering their mainstream CPUs. I believe that we should all read this great article before hating on Intel's decision on using TIM.

In short, it isn't as easy as just melting on some solder to bond the die and the heatspreader. It is an extensive and expensive process to solder a cpu while having good thermal conductivity and preventing damage to the die. The smaller the die is, the higher chance that micro-cracks in the solder will occur, which can kill your cpu. And that's the reason why Intel uses TIM. Not because they're cheap and they want to maximize profits, but because Intel's engineers know this and having a functional CPU long-term is more important than any thermal advantage that soldering has. They could get away with soldering Sandy Bridge because it was on a larger node, but Ivy-Bridge and onwards use pretty small dies, so they couldn't solder them because of the risk of damage.

The reason why Intel doesn't use higher quality TIM like liquid metal is because it isn't necessary. The main reason for the large temperature difference between stock and delidded isn't mainly the TIM, but the sealant they use to stick the heatspreader to the substrate. It increases the distance between the die and the heatspreader, which lowers thermal transfer efficiency. After removing that for the delid, you shorten the distance between the two materials, which improves thermal efficiency. I can't comment on Intel's TIM because I don't know what it's composed of, but considering that most pastes on the market perform within ~5C of each other (more like 2C for the more popular ones), I think it's safe to say that it doesn't have extremely poor thermal conductivity.

And because paste doesn't stick to silicon that well, if it wasn't for the sealant raising the heatspreader away from the CPU, it would get squished out due to the increased pressure and multiple thermal cycles. This is why those who opted for a paste instead of liquid metal are experiencing higher temperatures over time.

Then, you have the manufacturing differences from heatspreader to heatspreader. Some are more convex than others, which will lower thermal transfer efficiency on the edges, and others have minor imperfections on and underneath the surface that TIM will have to compensate for. Some chips also have more leakage than others, so that is also another big contributor from the temperature variances from one CPU to another.

If I'm wrong on anything, please let me know so it can be corrected. I just want to stop the misinformation going around about Intels' TIM
thumb.gif
 
See less See more
1
#2 ·
Quote:
Originally Posted by GnarlyCharlie View Post

It's explained in this podcast. Here's a link with the timestamps of the various topics:

http://www.overclock.net/t/1590822/intels-dan-snyder-will-be-on-the-next-asus-straight-edge-podcast-ask-your-questions-here
Podcast

Timestamps (but the topics overlap on others, best to listen to the whole thing)

What determines the choice of thermal interface material for various CPUs? 26:30~29:40
What determines the gap between the CPU die and IHS? 29:40~34:21
 
#3 ·
I think the data speaks for itself. Some chips seeing over 20C improvements after a delid! No quality design should be that far off the ideal solution. Plenty of other processors, including Intel's own, have had more thermally effective packages before this generation.

You're right that the TIM might be "good enough" for base clock speeds, but Intel has gone through a lot of effort to make their processors+chipsets easy and effective to overclock. People are just frustrated they skimped or made a poor choice on the package thermals, negating all that other great engineering work to make their parks overclockable in the first place!
 
#4 ·
Quote:
Originally Posted by GnarlyCharlie View Post

Podcast

Timestamps (but the topics overlap on others, best to listen to the whole thing)

What determines the choice of thermal interface material for various CPUs? 26:30~29:40
What determines the gap between the CPU die and IHS? 29:40~34:21
That's some great information there. Thanks for posting! Glad to hear something from Intel themselves, and I learned about other things too. IIRC, I remember SiliconLottery saying that after a few years of running CLU under the heatspreader, the temperatures were basically the same as stock, so it will require frequent maintenance to keep those lower temperatures too.
Quote:
Originally Posted by briank View Post

I think the data speaks for itself. Some chips seeing over 20C improvements after a delid! No quality design should be that far off the ideal solution. Plenty of other processors, including Intel's own, have had more thermally effective packages before this generation.

You're right that the TIM might be "good enough" for base clock speeds, but Intel has gone through a lot of effort to make their processors+chipsets easy and effective to overclock. People are just frustrated they skimped or made a poor choice on the package thermals, negating all that other great engineering work to make their parks overclockable in the first place!
Yup, I completely understand, thought I meant good enough in the sense that you won't be gaining much from going with better TIM like CLU or Thermal Grizzly, not when it comes to overclocking. It just frustrates me when some people blame something they shouldn't. Much like when people blame a buggy or poor game for "lazy devs". There's a lot more to it than people think, and I think it's something that really needs to be understood.

But I think another reason for the large temp. drop is because these dies are getting smaller and smaller. Because of that, the thermal density of these CPUs become higher, and thus any methods of removing or improving thermal transfer layers give more extreme results.
 
#5 ·
Quote:
Originally Posted by airisom2 View Post

But I think another reason for the large temp. drop is because these dies are getting smaller and smaller. Because of that, the thermal density of these CPUs become higher, and thus any methods of removing or improving thermal transfer layers give more extreme results.
You don't even have to run it bare die to get extreme results, even just replacing the existing TIM with normal quality TIM and butting the IHS back on can easily give drops of ~15c. Removing the IHS or not really doesn't make much of a difference.

Intel just cheaped out on the TIM they used. Even if they couldn't solder their high end mainstream CPU's, the could've at least used proper high end paste. That's the only reason these chips run hot, not because of small dies, not because of the IHS, its because they just don't care and will shake every last penny out of consumers, enthusiast appeal be damned.

They've made it plenty clear they just don't care about enthusiasts anymore, not that it matters now that AMD is back. IMO, everyone who cares about this hobby should refrain from buying Intel now, unless they absolutely have to.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top