Intel 18c HEDT Die Pictured, Not Soldered (7980XE)

[mynameisjuan]

2 minutes ago, Jito463 said:

The exact numbers?  I couldn't say.  I don't run servers personally, and I don't have access to the necessary hardware to test.  However, even a few degrees difference can help significantly with their cooling solutions, and could save hundreds - if not thousands - of dollars per month in power for said cooling.

I couldnt find any stock numbers either since everyone that delids OCs it right off the bat. But I cant argue temps saving money. Id be curious on how much stock temps are altered by soldered vs non-soldered.

[Drak3]

27 minutes ago, Jito463 said:

Now, maybe there's a legit reason for Intel to not solder these chips, or maybe it's nothing more than a cost cutting measure

I'm thinking that soldering was either killing too many low yield chips, or Intel found longevity issues with soldered engineering samples.

If only we had an Intel Engineer to verify.

[Jito463]

1 hour ago, Drak3 said:

I'm thinking that soldering was either killing too many low yield chips

That's my theory, which goes back to cost cutting.  I'm not saying I'd blame them if that's the case.  That would be a perfectly understandable reason for them to switch to TIM instead of solder.  However, people think that by saying Intel does it for cost cutting, it means we're lambasting Intel unfairly.  Not every cost cutting measure is done out of greed (though sometimes it can certainly be done for penny-pinching reasons).

[leadeater]

50 minutes ago, Drak3 said:

I'm thinking that soldering was either killing too many low yield chips, or Intel found longevity issues with soldered engineering samples.

If only we had an Intel Engineer to verify.

Tinfoil hat on maybe Intel is trying to kill off the used market by making their CPUs die sooner /s

 

My guess would also be yields since dies are getting very large so ever one ruined by the soldering process could start to get costly, even if it's just in unit availability. 

[mr moose]

1 hour ago, Jito463 said:

Now, maybe there's a legit reason for Intel to not solder these chips,

 

There will definitely be a legitimate reason for them not to solder, it's just that whatever that reason is, it doesn't make enthusiasts happy.   And that's when the assumptions start flying.

 

 

[Jito463]

4 minutes ago, mr moose said:

 

There will definitely be a legitimate reason for them not to solder, it's just that whatever that reason is, it doesn't make enthusiasts happy.   And that's when the assumptions start flying.

Sorry, I worded that incorrectly.  I meant to say "technical" reason (as in, the TIM performed the job better than solder for some unknown reason).

[mr moose]

4 minutes ago, Jito463 said:

Sorry, I worded that incorrectly.  I meant to say "technical" reason (as in, the TIM performed the job better than solder for some unknown reason).

if that turned out to be the case, then there should be a few embarrassed people on this forum. 

[MageTank]

4 hours ago, mynameisjuan said:

I couldnt find any stock numbers either since everyone that delids OCs it right off the bat. But I cant argue temps saving money. Id be curious on how much stock temps are altered by soldered vs non-soldered.

I provided stock numbers in my delid post: 

I used a 6700k at 4.2ghz (it's stock all-core boost with advanced turbo on), an undervolt from 1.23v down to 1.12v. The before and after result both had this undervolt applied, inside of an ITX chassis with a 45mm vapor chamber cooler (not unlike the vapor chambers you find in rack mounted chassis'). Delidding and using liquid metal brought a 22C difference in temperatures from the package. Solder, from what @done12many2 showed in this very thread using Der8auers video, is about 4C hotter on average compared to liquid metal. Still a substantial difference from traditional paste.

 

That being said, this only raises more questions. That Xeon-W IHS still has the pressure release holes in it, despite it being soldered. I cannot tell if Intel is intentionally deceiving us by making the IHS appear that it's soldered, or if it was an aesthetic they wished to keep with the Xeon lineup, but it certainly seems odd to keep those holes when they serve no functional purpose now.

 

One thing is for certain though. If they chose not to solder these massive Xeon's, they have to have a good reason for it. The penny pinching theory simply does not make sense when talking about these massive Xeon's, as they make the most margin on these chips in the first place. Companies pay extra for the premium longevity and performance that these Xeon's offer, and Intel would not intentionally cut corners to save a few bucks on a product that costs thousands of dollars. Hopefully @LinusTech or other reviewers with contacts inside of the industry can press Intel for more information about this. I cannot think of a reason as to why they would not disclose the truth, unless they think their competitors might gain from having that knowledge. Either way, I feel they stand to gain far more by mending this rift between their die-hard fans and the general consumers that are both questioning this decision. 

[X_X]

On 9/10/2017 at 2:47 AM, JurunceNK said:

The two-PCB thing started back with Haswell-E. The last generation of HEDT CPUs to have only one PCB was Ivy Bridge-E.

Thanks to @PCGuy_5960, the two PCB thing started with Broadwell-E. Haswell-E has only one PCB, as there's Google images of Haswell-E delidded 

 

Here's an old Pentium III

 

The words for today are "substrate" and "interposer"