Question Alder Lake - Official Thread

Page 113 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

igor_kavinski

Platinum Member
Jul 27, 2020
2,536
1,292
96
Like spoons? o_O'

The best result can be achieved in our tests with a thickness of 1.0 mm with 70.88 °C and thus an improvement of 5.76 °C
At least, it can be solved easily, if you are willing to put a little effort into it.
 

Exist50

Senior member
Aug 18, 2016
623
555
136
Wait what?
What do you mean? This addresses the same thing as the washer mod. Both have been tested to be, at best, a few degrees difference with a particular flat waterblock. More mainstream cooling like AIOs and air coolers will see negligible difference.
 

igor_kavinski

Platinum Member
Jul 27, 2020
2,536
1,292
96
I mean where are you getting the lapped heatsink thing? It's not a bowed IHS, it's a bent socket.
The IHS is getting bent, leaving an airgap. It should be serious but only around a 5 degrees celsius improvement with washer mod? Doesn't make sense. The gap should reduce heat dissipation significantly but apparently, that's not happening.
 

coercitiv

Diamond Member
Jan 24, 2014
4,884
7,137
136
I mean where are you getting the lapped heatsink thing? It's not a bowed IHS, it's a bent socket.
The bent heatsink is only a problem if you have a perfectly flat heatsink, which likely means a lapped heatsink. Most consumer heatsinks are already curved (convex) for exactly this reason, to maximize contact area with the slight curvature of the Intel heatsink. This is not something new, heatsink curving on Intel (and sometimes AMD) sockets has been around for a long while now. This includes DIFFERENT types of bending depending on socket (convex, concave, double wave), it all depends on the type of socket retention.

Here's one example from observations in a NH D15 review:
The Noctua NH-D15 heatsink has a machined nickel-plated copper baseplate with a surface roughness of approximately ~16 microinches. The base plate is machined perfectly flat in one axis and very slightly convex in the opposite.

This differs from the NH-D14 observation, which is good news for me considering I have the habbit of mounting it at a 90 degree angle:
The Noctua NH-D14 heatsink has a machined nickel-plated copper base with a surface roughness of approximately ~16 microinches, which is considered very good. The base plate is machined flat in both axis.

And here's what Noctua has to say about it:
Why is the cooler’s contact surface slightly convex?
As the Integrated Heat Spreaders (IHS) of today’s CPUs are slightly concave, the cooler’s contact surface has been deliberately designed to be slightly convex in order to ensure optimal contact. This way, more contact pressure will be applied at the centre of the IHS directly above the DIE, which results in better heat transfer and improved overall performance.
and another comment from Noctua support on lapping, dated 5 years ago:
Apart from that, lapping the CPU/ IHS outside the socket doesn't help, as it gets deformed once installed inside the socket. If you would like to have the best performance with a lapped heatsink, you would have to flatten the CPU when it is installed inside the socket.
 
  • Like
Reactions: A///

DrMrLordX

Lifer
Apr 27, 2000
19,163
7,920
136
The IHS is getting bent, leaving an airgap.
The entire IHS is bending? I doubt that! It's more likely that the socket retention mechanism is changing the lie of the HSF due to the bent socket.

Also again, who said anything about this being a problem exclusively for lapped heatsinks? I've not seen one report indicting lapped heatsinks.
 

igor_kavinski

Platinum Member
Jul 27, 2020
2,536
1,292
96
The entire IHS is bending? I doubt that!

Our internal data show that the IHS on 12th Gen desktop processors may have slight deflection after installation in the socket.
Reading between the lines here, they are essentially saying “it might bend a little, sure, but that won’t cause it to run outside of its specification.”
It's bending, Oh God, it's bending!
 

DrMrLordX

Lifer
Apr 27, 2000
19,163
7,920
136
Wouldn't there be more chance of the IHS adhesion getting compromised, essentially delidding the CPU?
The PCB would probably crack before that happened. I don't know if you've ever tried to chop through that stuff, but it's stout. Really if the IHS were to bend to any considerable degree, the first thing to go would be the solder.
 
  • Like
Reactions: A///

MadRat

Lifer
Oct 14, 1999
11,767
117
106
If you have thermal paste then it fill that concavity. You are talking at a microinches scale. I'm honestly surprised they aren't grooved to mate up even better, but some numbskull would figure out how to screw that up, too.
 

DrMrLordX

Lifer
Apr 27, 2000
19,163
7,920
136
@MadRat

Getting a bit off-topic, but IBM did some research related to that which they posted . . . wow, 15 years ago:


I think it was more intended from the point-of-view of a die<->IHS thermal interface, but it could also have been applied to an IHS<->HSF interface as well. I'm not sure whether all that research bore fruit.
 
  • Like
Reactions: MadRat

jpiniero

Lifer
Oct 1, 2010
11,320
3,052
136

Alder Lake HX's final specs leak. Max turbo power is 157 W. Launch is on Tuesday.
 

ASK THE COMMUNITY