The Official Thermal Interface Material thread

[ShrimpBrime]

Nice! Mine wasn't nearly that bad, it only rolled the ends of the MB up off of the test bench about an inch and a half. It was quite the sight.

I've noticed this more on the ply wood I was using earlier this year, but straight out of the box brand new without a cpu in it. Like you said, better to be mounted to a case or bench table than just breadboarding it. Hold downs do really help keep the board flat. (edit: but not an inch and a half. maybe quarter inch)

The Cooler mounting holes are pretty far away from the socket. But the coolers have to be so darned big, especially on socket 1700 lol. I can see with enough flex, the cooler may actually separate some from the IHS plate and possibly separate the TIM as well.

I use only 2 of the mounting holes and you can see that picture in the post your cooling thread. I snug them down first night. Since it's a water block, I don't worry on the temps much.. But the next night or two or three, I check the torque. By feel yes. But in the past have used my inch pound torque driver. It's older than dirt, was my grandfathers actually. Accuracy unknown haha. I'll have to take a picture of that. Anyhow, my silly opinion would be to utilize some tooling to have nice accurate results if YouTubers are doing any actual real testing. Be it Der the Mr. Bauer or Steve or Linus dropped a video card.... I could see some actual evidence of good testing methods. Well, instead of watching videos, I do a lot of this testing at home, but much of it 10 years or more ago.

If I where to design a new ILM, it would be large enough to reach the cooler mounting holes for maximum rigidity. I see how they are somewhat useful now, but I feel they could be a little larger.

 

[Noci]

If I where to design a new ILM, it would be large enough to reach the cooler mounting holes for maximum rigidity. I see how they are somewhat useful now, but I feel they could be a little larger.

Just to give you some reading material to prepare you how deep this rabbit hole is , see file in this link. (download the PDF)

This subject is well studied by Intel as well as AMD engineers and is much more complex than most here realise if I read throug the posts. Nevertheless there is always room for improvement and new ideas, I'm all for that.

 

[ShrimpBrime]

Just to give you some reading material to prepare you how deep this rabbit hole is , see file in this link. (download the PDF)

This subject is well studied by Intel as well as AMD engineers and is much more complex than most here realise if I read throug the posts. Nevertheless there is always room for improvement and new ideas, I'm all for that.

Yes, looks good. Lots of useful information. Thank you

 

[KLiKzg]

If I where to design a new ILM, it would be large enough to reach the cooler mounting holes for maximum rigidity. I see how they are somewhat useful now, but I feel they could be a little larger.

Nice idea, but impractical...as there are many small & larger stuff on boards between CPU & holes...& if they got rid of those, that means more travel time for electrons to the RAM, chipset, etc.
Also, note that no 2 boards are the same...so you would need to have dozen of shapes or maybe even square of a dozen shapes...

But probably we would see twist to KPU addition on boards, implementation of KPU inside CPU & then switching CPU to all new technology...probably very soon!

 

[ShrimpBrime]

Nice idea, but impractical...as there are many small & larger stuff on boards between CPU & holes...& if they got rid of those, that means more travel time for electrons to the RAM, chipset, etc.
Also, note that no 2 boards are the same...so you would need to have dozen of shapes or maybe even square of a dozen shapes...

But probably we would see twist to KPU addition on boards, implementation of KPU inside CPU & then switching CPU to all new technology...probably very soon!

Since I'm not a manufacturer, I'd design one for my board only. Fk everyone else.

 

[TPCEA]

Reasons why YouTubers say that, is because they are paid to say positive review on the merchandise!

Some are, some aren't. Of those that are, some will shill any product, while others are selective.

The rest of us don't shill; we say it like it is.

But don't "assume" too much, as it is with the word...without checking the facts (even right here on forum), you make "ASS of yoU & not ME"!

Actually, it's "make an ass of you and me", but I like your variant.

 

[KLiKzg]

Some are, some aren't. Of those that are, some will still any product, while others are selective.

The rest of us don't shill; we say it like it is.


Actually, it's "make an ass of you and me", but I like your variant.

Well, I still like the Igor's lab far better then most of YouTubers...especially they are the ones doing a laboratory testing of this TIMs.

 

[TPCEA]

Well, I still like the Igor's lab far better then most of YouTubers...especially they are the ones doing a laboratory testing of this TIMs.

And that is perfectly fine, although some people seem to be polarized about IL. But what about the YTers and others that do testing?

 

[KLiKzg]

And that is perfectly fine, although some people seem to be polarized about IL. But what about the YTers and others that do testing?

Read more about it, several pages back (in this topic).

 

[ShrimpBrime]

Some people argue that the TIM is the weak point, but I beg to differ as the TIM might get one 2°C while a better CPU cooler might get one more.

So, I got me a new CPU cooler for $11 that should get me more than 2°C

Thermal Interface Compound IS 100% the weak point..

Even if a paste was 20wm/k, that's so much lower than solder let alone the 2 copper plates, it could be no other than the weak point with the least amount of thermal conductivity.

Another weak point is all the layers.

So SOI, then solder, then copper plate, then paste, then another cold plate, then dissipation to atmosphere.

That's why direct die and LM works so much better than all those layers mentioned. Which all of this I would think you'd know by now....

Your 11$ cooler is fine for the low wattage system you have. It would work better on direct die.

FYI, air coolers are more efficient when warmed up. Thermal paste helps transfer thermals, but is the slowest of all components in the stack otherwise.

 

[lexluthermiester]

A comment from @Shrek was here a moment ago that stated;
"Some people argue that the TIM is the weak point, but I beg to differ as the TIM might get one 2°C while a better CPU cooler might get one more.

So, I got me a new CPU cooler for $11 that should get me more than 2°C"
I happen to agree. No TIM can or will ever overcome the limitations of the cooling mechanism. Improve the cooler, improve the cooling. Basic cooling physics.

 

[ShrimpBrime]

A comment from @Shrek was here a moment ago that stated;
"Some people argue that the TIM is the weak point, but I beg to differ as the TIM might get one 2°C while a better CPU cooler might get one more.

So, I got me a new CPU cooler for $11 that should get me more than 2°C"
I happen to agree. No TIM can or will ever overcome the limitations of the cooling mechanism. Improve the cooler, improve the cooling. Basic cooling physics.

Wish it was something to agree too... but we where de-lidding 8700K processors to replace Intel stock TIM and was seeing average of 15c to 20c temp drops. And it didn't matter what cooler either. That's all the proof you need that TIM is a weak point. So processors soldered after ward never to return using thermal grease under an IHS plate.

 

[lexluthermiester]

Wish it was something to agree too... but we where de-lidding 8700K processors to replace Intel stock TIM and was seeing average of 15c to 20c temp drops. And it didn't matter what cooler either. That's all the proof you need that TIM is a weak point. So processors soldered after ward never to return using thermal grease under an IHS plate.

No, that is an example of the IHS causing an impedance to heat transfer. That has nothing to do with the TIM. At all. Full stop, end of discussion. This is why Sony is using LMoD TIM for cooling the PS5. Direct die contact will always improve cooling, always.

 

[Shrek]

A comment from @Shrek was here a moment ago that stated;
"Some people argue that the TIM is the weak point, but I beg to differ as the TIM might get one 2°C while a better CPU cooler might get one more.

So, I got me a new CPU cooler for $11 that should get me more than 2°C"
I happen to agree. No TIM can or will ever overcome the limitations of the cooling mechanism. Improve the cooler, improve the cooling. Basic cooling physics.

My apologies... I wrote it and then thought it might be off topic, so deleted it.

 

[Count von Schwalbe]

Direct die contact will always improve cooling, always.

True, but

replace Intel stock TIM

And then put the IHS back on. So the crappy Intel TIM was to blame in that case.

Thing is, cooling is a chain. One weak component on the chain makes it all go bad, whether it be TIM, IHS, or cooler.

 

[lexluthermiester]

True, but

And then put the IHS back on. So the crappy Intel TIM was to blame in that case.

Thing is, cooling is a chain. One weak component on the chain makes it all go bad, whether it be TIM, IHS, or cooler.

Ok, while that's a fair point we're talking about less than 5C difference. However, Shrimp said it's "100% the weak point", which is complete hogwash and moose muffins. Quality of TIM can be a factor but not 100%. I proved years ago in other threads here on TPU that materials other than purpose crafted TIMs can act as a reasonable and even very good TIM.

 

[R-T-B]

I do not think so, as it would be so - then output from CPU will be only heat & no work done.

Not necessarily. It can both produce heat and do useful work.

 

[ShrimpBrime]

What proof?

Remove the part to need thermal paste between 2 surfaces.

True, but

And then put the IHS back on. So the crappy Intel TIM was to blame in that case.

Thing is, cooling is a chain. One weak component on the chain makes it all go bad, whether it be TIM, IHS, or cooler.

Well some people did. Some people didn't

And still to this day, many run lid-less LGA/PGA (with LM)

And many use SS units daily, but the thermal paste must hold up to the low temps. Again, the weak point. Once the paste cracks from being frozen, it's not thermally conductive, weakest point.

 

[jayjr1105]

Ok, while that's a fair point we're talking about less than 5C difference. However, Shrimp said it's "100% the weak point", which is complete hogwash and moose muffins. Quality of TIM can be a factor but not 100%. I proved years ago in other threads here on TPU that materials other than purpose crafted TIMs can act as a reasonable and even very good TIM.

It's a lot more than 5C if you use liquid metal. Back in my 3770K days, I dropped 19c from hottest core delidding and changing paste for liquid metal.

 

[lexluthermiester]

Back in my 3770K days, I dropped 19c from hottest core delidding and changing paste for liquid metal.

And I'll bet most of that improvement was the delidding. LM makes a solid difference, I've personally seen it. The IHS can be a serious restriction for heat conduction.

 

[jayjr1105]

And I'll bet most of that improvement was the delidding. LM makes a solid difference, I've personally seen it. The IHS can be a serious restriction for heat conduction.

The lid went back on, with LM on the die under the IHS. This wasn't direct die cooling.

 

[lexluthermiester]

The lid went back on, with LM on the die under the IHS. This wasn't direct die cooling.

And you got a 19C improvement? Damn!

 

[jayjr1105]

I had one stupidly hot core on my 3770K, the rest were like 12-13c difference.

 

[lexluthermiester]

I had one stupidly hot core on my 3770K, the rest were like 12-13c difference.

Ok, sure! That was one crazy a$$ hot core!

 

[freeagent]

I didn't delid my 3770K, but there is 12c between 4 cores at best.. crazy. With my old X5690 I could do a 3c spread between the 6 cores.