Intel ‘Devil’s Canyon’ chips’ NGPTIM is still not efficient

[HeathCliff]

Intel Corp.’s NGPTIM [next-generation polymer thermal interface material], which was supposed to bring back record overclocking capabilities to the company’s processors is better than its predecessor, but is significantly worse than commercially available thermal interfaces, a new research has found out.

 

Earlier this year Intel Corp.’s promised that its Core i7-4790K and Core i7-4690K “Devil’s Canyon” microprocessors will have considerably better overclocking potential compared to the original chips based on the “Haswell” micro-architecture released last year thanks to improved thermal interface between the die and the heat-spreader as well as revamped power supply circuity. However, actual overclocking attempts of a Core i7-4790K central processing unit by KitGuru’s reviewer Luke Hill did not reveal any breakthroughs. Apparently, the reason for that is quite simple: the new thermal interface that Intel uses still has limited efficiency.

 

All Intel processors for desktop computers in the recent 15+ years have featured heat-spreaders (which many call lids) on their dies to protect the latter from damages and enable more efficient heat dissipation. Back in the days Intel soldered dies to heat-spreaders using a special Indium-based alloy, which enabled very efficient transfer of heat. However, starting from the code-named Ivy Bridge microprocessors Intel replaced the solder with a layer of thermal paste, which is considerably less efficient than the previously used alloy.

 

Many overclockers complained that in order to unlock maximum overclocking potential of the Core i7-3770K and the Core i7-4770K microprocessors they had to remove the heat-spreaders (de-lid the chips) and change thermal interface. With the latest “Devil’s Canyon” central processing units Intel started to use a new substance called the NGPTIM, which is more efficient than its predecessors. However, a research conducted by 3DNews web-site has revealed that the NGPTIM is far less efficient than even affordable thermal pastes available on the market.

 

 

 

So I'm not really that surprised. Not sure why Intel can't go back to soldering the IHS but it seems that like before, you'll need to de-lid it before you can push your chip any further.

 

http://www.kitguru.net/gaming/uncategorized/anton-shilov/intels-devils-canyon-chips-ngptim-is-still-not-efficient-research/

[ZacDaMan72]

Sigh... Stahp Intel...

[WyvernAbstracts]

#delidintelcpu2k15 join the movement or get left behind 

[JuztBe]

devil's canyon weren't suppose to be anything special. They are even selling it for same price. I expected even less from it.

[Opsilon]

So I'm not really that surprised. Not sure why Intel can't go back to soldering the IHS but it seems that like before, you'll need to de-lid it before you can push your chip any further.

 

http://www.kitguru.net/gaming/uncategorized/anton-shilov/intels-devils-canyon-chips-ngptim-is-still-not-efficient-research/

It's not that they can't, they will not do it, it is more expensive for them, and until they have AMD or anyone else beating them, they will not give us soldered IHS on 11xx platforms, only xeons and i believe 2011 desktops and successors will have soldered IHS

[Biased Opinion]

I guess i will just delid mine then.

[Manny Calavera]

16°C Wow...

[Silentknight]

If you watched the Intel stream with Linus, you would already know why there not soldering them in place..

[Socky]

-.- really Intel? Cmon AMD we need you to get your shit together and throw it at intel, so they can smell like shit.

[Admire]

-.- really Intel? Cmon AMD we need you to get your shit together and throw it at intel, so they can smell like shit.

 

I agree but AMD has long since shown that they aren't capable to lifting that task....Sadly

[KamiKatze]

I hope intel includes an 'enthusiast' CPU in their next generation with soldered heatspreaders for what? 10-20 bucks more? I think the amount of people who would buy that exact model over others would be remarkable.

[flibberdipper]

It would only cost $20 more probably to do what they did before Ivy. Come on Intel, get your shit together.

[Levent]

ffs if you arent going to fix it just release a "new" enthusiast line cpus without lids

[warmmilk]

I agree with the whole solder vs TIM thing... but as far as TIM, Intel also has to consider the life of the TIM.  Most users (99.99999999999%) are never gonna delid their cpu's when the TIM wears out in 2-3 years.  Intel's new TIM is better than their old TIM, but I never expected it to be better than any of the top rated aftermarket TIM... I know enthusiast's get all up and arms about this sort of thing, but come on, you gotta be realistic

 

 

 

I hope intel includes an 'enthusiast' CPU in their next generation with soldered heatspreaders for what? 10-20 bucks more? I think the amount of people who would buy that exact model over others would be remarkable.

 

I do like this idea, I'd be willing to pay ~$20 more for a soldered cpu.  But if anyone watched Linus's Intel interview, Intel doesn't test the chips till they have the IHS already on, so they don't know which skew it will be till after that.  If they were to have a soldered K part available, that means they'd have to remove the IHS and install the new soldered IHS.  This means more than $20 increase...  or you can say they change their testing so they know before applying the IHS, but that means completely redesigning the process, so much much more than $20 per chip increase...

[BookerDeWitt]

If you watched the Intel stream with Linus, you would already know why there not soldering them in place..

link

[warmmilk]

link

http://www.twitch.tv/linustech/b/542046820 

[KamiKatze]

I agree with the whole solder vs TIM thing... but as far as TIM, Intel also has to consider the life of the TIM.  Most users (99.99999999999%) are never gonna delid their cpu's when the TIM wears out in 2-3 years.  Intel's new TIM is better than their old TIM, but I never expected it to be better than any of the top rated aftermarket TIM... I know enthusiast's get all up and arms about this sort of thing, but come on, you gotta be realistic

 

 

 

 

I do like this idea, I'd be willing to pay ~$20 more for a soldered cpu.  But if anyone watched Linus's Intel interview, Intel doesn't test the chips till they have the IHS already on, so they don't know which skew it will be till after that.  If they were to have a soldered K part available, that means they'd have to remove the IHS and install the new soldered IHS.  This means more than $20 increase...  or you can say they change their testing so they know before applying the IHS, but that means completely redesigning the process, so much much more than $20 per chip increase...

 

 

Good point. I watched the stream but didn't know how Intel tests their chips. If they test the done product then yes, it's too late.

 

So i guess we have to rely on LGA 2011 for high quality soldering? Very sad that intel has to go that route.

[HeathCliff]

ffs if you arent going to fix it just release a "new" enthusiast line cpus without lids

 

This. Intel please, for the love of god, do this. This will save so much time and effort in not having to de-lid them and hopefully, it should cut the cost down slightly. 

 

Also, on a side note, I wonder how well a de-lidded 4690K/4790K can overclock with a bit of exotic watercooling. Anyone willing to do it?

[topsu]

It would only cost $20 more probably to do what they did before Ivy. Come on Intel, get your shit together.

 

As far as I have heard they aren't using TIM because it is cheaper.

[warmmilk]

As far as I have heard they aren't using TIM because it is cheaper.

 

Where'd you hear this?  If not cost, then why?

[topsu]

Where'd you hear this?  If not cost, then why?

 

http://www.overclock.net/t/1398586/a-possible-reason-why-intel-was-forced-to-use-tim-instead-of-solder-with-ib-haswell

 

I heard one intel employee say this too but cannot find the source anymore because it was ages ago. =(

[Kamina]

Maybe they should just solder the K variants. 

[Ren]

I like this post from the link topsu provided:

 

solder was designed for Cpus with TDP over 100W, especially in the 130W TDP range, ie the enthusiast cpus. As Ivy and Haswell TDP decreased into range intel typically uses paste...to save millions, then intel used paste. overclocking specs are not considered.

The theory of unable to use solder on those will get borked when IVY E and then Haswell E comes out, those at 130W are going to have solder to meet specs, and fall in line with intels historical use of solder vs paste.

Intel using tim on 84W TDP processors is the norm, not the exception.

What changed is TDP driven down by efficiency chasing for mobile applications.

[Ren]

Basically, you'd see much higher prices on these CPUs if they were to solder them at all.

I think if you want performance from Intel, you should just go to the enthusiast grade processors. Otherwise, you're stuck paying $180+ for just a "really good consumer CPU" which really has always been the case but this whole DC thing is reminding people of that truth.

[warmmilk]

http://www.overclock.net/t/1398586/a-possible-reason-why-intel-was-forced-to-use-tim-instead-of-solder-with-ib-haswell

 

I heard one intel employee say this too but cannot find the source anymore because it was ages ago. =(

 

so technically its not cost, but in reality its still cost...  soldering at smaller die sizes meant more dead chips, so more cost to intel

 

 

Maybe they should just solder the K variants. 

 

see post 14 in this thread