Intel’s Comet Lake-S Processors Feature Thinner Dies and Thicker Heat Spreaders for Improved Thermals

[Tsing]

Image: Intel



With their increased core counts, temperature has become a rising concern with Intel’s recently unveiled Core S-series desktop processors, but the company has made some under-the-hood tweaks that should make it easier to keep thermals in check. According to a slide from the official Comet Lake-S deck, Intel implemented a thinner die STIM, which allowed for a thicker integrated heatspreader (IHS).



“On top of a processor sits an area of inactive silicon, and silicon is a terrible heat conductor, at least when compared to copper,” explained OC3D. “With Comet Lake, Intel has thinned this...

Continue reading...

 

[Eduardo_Domingot]

Really not impressed by this effort on their part. Like I still can’t believe I was wrong and they actually went with this branding... still good on them i guess for making some thermal improvements that I really think they honestly had to make for the product to be possible.

 

[Grimlakin]

So we're talking about heavier and thicker heat spreaders on top of thinner layers... Folks are going to need to be VERY careful when mounting their coolers, or a rather ingenious mounting design will be needed to spread pressure correctly at all times. The bigger thicker heatspreader... I suppose that's all good, but does it have the ability to transfer that heat out rapidly? I'm sure they have engineered that to be a boon as opposed to a problem.

 

[Dan_D]

I was going to say the same thing. Thinner heat spreaders could mean damaged heat spreaders or dies if you aren't very careful.

 

[funkydmunky]

Article says thicker heat spreaders. One would imagine Intel has done some steps to help get the extra 100-300 extra MHz headroom they required.

 

[Riccochet]

I don't think you could apply enough force, with most cooler mounting systems, to deform the heat spreader enough to crack a chip. I say most mounting systems since most use springs and nuts that only screw down so far to prevent over pressure.

They kind of took all the guess work out of mounting a cooler.

 

[Dan_D]

I don't think you could apply enough force, with most cooler mounting systems, to deform the heat spreader enough to crack a chip. I say most mounting systems since most use springs and nuts that only screw down so far to prevent over pressure.

They kind of took all the guess work out of mounting a cooler.

Considering people have crushed 6700K's by over tightening the mounting screws on waterblocks and air coolers, it's entirely possible to crush one of these newer CPUs the same way.

 

[LeRoy_Blanchard]

Yeah let's not rule out human stupidity.

 

[Eduardo_Domingot]

Considering people have crushed 6700K's by over tightening the mounting screws on waterblocks and air coolers, it's entirely possible to crush one of these newer CPUs the same way.

But then its kinda a moot point. If they did it before, then it has been a possibility for a little while.

 

[Riccochet]

Considering people have crushed 6700K's by over tightening the mounting screws on waterblocks and air coolers, it's entirely possible to crush one of these newer CPUs the same way.

What cooler was doing that?

 

[Dan_D]

What cooler was doing that?

It wasn't any specific cooler. It can happen when you over tighten the screws on anything that uses screws and tension springs for retention. It was something people brought up awhile back and people quickly concluded that it was user error.

 

[Grimlakin]

When you start making parts thinner on the die you open them up to problems. A thicker heat spreader might help keep pressure even but unless there is a rather study shim to keep it that way there could still be issues.

Don't get me going on the time BEFORE heat spreaders and direct cooler mounting on top of CPU's. Plenty of the were cracked back then.

Of course this was back in the day when you could 'unlock' a DX feature on a SX cpu by drawing a line with a leaded pencil. Nobody would believe that today unless they were around for it back in the day.

 

[David_Schroth]

Don't even have to go back that far. Pencil trick worked to unlock the Thunderbird Athlons as well.

 

[Dan_D]

When you start making parts thinner on the die you open them up to problems. A thicker heat spreader might help keep pressure even but unless there is a rather study shim to keep it that way there could still be issues.

Don't get me going on the time BEFORE heat spreaders and direct cooler mounting on top of CPU's. Plenty of the were cracked back then.

Of course this was back in the day when you could 'unlock' a DX feature on a SX cpu by drawing a line with a leaded pencil. Nobody would believe that today unless they were around for it back in the day.

Intel and AMD both utilized heat spreaders long before there were any CPU's that allowed for direct die cooling. The original Pentium and Pentium Pro processors had heat spreaders as did the AMD K6 processors. Even Cyrix's 6x86 and 6x86 MX and MII processors all featured heat spreaders. The Thunderbird based Athlons and FC-PGA 370 Pentium III's were the first and pretty much only direct die cooled CPU's outside of the mobile segment where the practice continued for a lot longer.

What enabled direct die cooling was the introduction of flip chip packaging like what was used on the Pentium III's and Thunderbird based Athlons. Prior to that, heat spreaders existed, but the CPU's were arranged differently. The FC-PGA packaging was designed to allow for direct die cooling. However, in the desktop market this made the CPU's exceptionally fragile and many people cracked chip dies doing routine builds. Copper shims and various products were made to try and reduce the prevalence of such issues. In the end both AMD and Intel simply stopped selling CPU's without heat spreaders for desktops.

 

[Brian_B]

I cracked my share of Durons.

That being said... not sure how a thicker heat spreader helps so much, unless it's more to mechanically offset the fragility of a thinner die.

Seems like a thicker transfer medium (in this case, the spreader itself) is actually an impediment to heat transfer. I'm not sure what the spreaders are actually made of (copper? aluminum?, steel?) but anything they use has some positive non-zero value for thermal conductivity, and more of it will impede heat transfer more than less of it.

I guess more of it helps to distribute a point heat source over a larger surface area, but at the expense of impeding the heat transfer somewhat - probably some sweet spot there you can hit in terms of additional surface area vs the additional thermal resistance depending on the material used and the temperatures involved.