Im glad i din't wait for Kaby Lake. Thermalpaste problems again, like with Haswell

[pansertjald]

The New Kaby Lake CPU get's REALLY warm, so it has to be delidding to gain normal temps

delidded core i7-7700k runs 26 degrees c cooler

[Fascinate]

I made this thread already, but if you think about it this wont affect that many people. All these chips have that dreaded "thermal wall" (except the amazing sandy bridge) most people will just set volts below this and be happy with a 4.7 or 4.8ghz overclock and much better temps.

At least, i think thats whats going on with kaby.

[Deleted]

I made this thread already, but if you think about it this wont affect that many people. All these chips have that dreaded "thermal wall" (except the amazing sandy bridge) most people will just set volts below this and be happy with a 4.7 or 4.8ghz overclock and much better temps.

At least, i think thats whats going on with kaby.

Your average consumer who does overclock, most of these people do not touch voltages even when buying these kinds of hardware.

Every time I've done an overclock and used auto volts as a starting point, both AMD and Intel always set an aggressive voltage for auto, or rather the board manufacturers.

[Fascinate]

I just have to wonder why intel got away from the method sandy bridge had between the die and cover, it couldn't have cost that much more right? My CPU's temps scale like perfectly as i increase volts, and even with a stupid 1.5v it never goes over like 75c and thats with a cheap air cooler.

[pansertjald]

I just have to wonder why intel got away from the method sandy bridge had between the die and cover, it couldn't have cost that much more right? My CPU's temps scale like perfectly as i increase volts, and even with a stupid 1.5v it never goes over like 75c and thats with a cheap air cooler.

Whit no competition from AMD, they can get away with it, even if they only save a peny on it. I REALLY hope that AMD ZEN will give them some serious competition, so Intel will stop doing things like this. As it looks now, you wont get anything out of upgrading to Kaby Lake from Sky Lake at all. The whole seeling point of Kaby Lake was that it could OC higher then Sky Lake, but with those temps, i don't see it

[Remilia]

On the business side.
Solder does cost way more than crappy TIM, relatively speaking at least. To them it's just higher margin, even if only by $1, which if you multiply by a million is a lot of money. All numbers pulled out of ass, so it may be less dramatic.

On the consumer side.
Fuck em. Plain and simple. lol.
Even AMD's rather low cost APUs use solder between the IHS.
https://www.techpowerup.com/228255/a...-apu-de-lidded
Granted he destroyed the APU trying to delid it due to heating it up to insane temps, but hey you don't need to worry about poor thermal conductivity where even an average TIM will grant a 10C difference.

[Evildeffy]

I just have to wonder why intel got away from the method sandy bridge had between the die and cover, it couldn't have cost that much more right? My CPU's temps scale like perfectly as i increase volts, and even with a stupid 1.5v it never goes over like 75c and thats with a cheap air cooler.

Their official reaction was "We do not have the capability to solder the new Ivy Bridge cores like Sandy Bridge, it is not compatible".
Shortly after Ivy Bridge-E was released and that was soldered and Intel was completely silent on the matter.

Intel then was silent for a long ass time, allowing for theories to pop up which are very likely:

1. Intel had no competition in the field thus have no reason to spend more.
2. Price, a penny saved on millions of CPUs is still a penny saved every time
3. Intel did it to differentiate between LGA115X and LGA2011.
4. Intel did it because it would make their CPUs too good and prevent people from upgrading, why upgrade when you can overclock heavily possibly making an older architecture better than the new one? (Sandy Bridge overclockability and temps vs. Ivy Bridge and Ivy Bridge vs. Haswell (4670K and 4770K))

There are plenty more theories, these are the most acceptable ones.

Later on however a study showed up (this was 13 months ago) that according to Intel itself caused solder to crack on small dies.

http://www.overclock.net/t/1568897/a...#post_24286980
This is the forum location of the guy who found the study.

http://iweb.tms.org/PbF/JOM-0606-67.pdf
The actual study itself here detailed.

The funny thing about this study is that the actual die size on Ivy Bridge and Haswell are larger than Sandy Bridge due to their iGPU implementation.

Have fun wrapping your head around that.

[Kagthul]

Seems like it isn't even about the TIM, really, but about the space created by the crappy glue they use to hold the heat spreader on. It creates excess space. Guy de-lidded, removed the glue, and just put the lid back on with the existing TIM and saw immediate improvements. (and the study where they showed the solder cracking on small dies, particularly 3D transistors like Haswell, was repeated by a third party; in some cases it would crack right off the die completely, leaving the CPU with no heat transfer at all).

Honestly, we often allow ourselves to think we represent even a tiny portion of the market and we really dont. Enthusiast parts make up such a tiny part of their business they could literally lose all of us and not care or notice.

For the average user, these chips are fine.

[Sturmbringe]

I don't buy the story in the OP. Smt fishy.

Honestly, we often allow ourselves to think we represent even a tiny portion of the market and we really dont. Enthusiast parts make up such a tiny part of their business they could literally lose all of us and not care or notice.

For the average user, these chips are fine.

Even for an enthusiast they may be fine.

Take me, for instance. 99% of the time my 4770k runs at stock. (I can't even be arsed to change the Single Core Turbo at stock to Multi Core Turbo at stock from the ASUS ROG utility). I only overclock it when playing GTAV or ROME II TW or Elite Frontier or SWTOR, which is not very often.

Because I just can't be arsed to get into the BIOS and do a proper overclock, I OC the CPU from the ASUS ROG utility which only allows overclocking up to 4.1 GHz without running into some serious issues.

So, there you have it.

The bottom line is, CPU's nowadays are plenty fast for gaming even at stock or near stock speeds. Back in the Core 2 era and earlier during the Pentium 4 era, you simply could not game without heavy overclocking. You HAD to overclock else your game simply wouldn't run.

[Evildeffy]

Seems like it isn't even about the TIM, really, but about the space created by the crappy glue they use to hold the heat spreader on. It creates excess space. Guy de-lidded, removed the glue, and just put the lid back on with the existing TIM and saw immediate improvements. (and the study where they showed the solder cracking on small dies, particularly 3D transistors like Haswell, was repeated by a third party; in some cases it would crack right off the die completely, leaving the CPU with no heat transfer at all).

Honestly, we often allow ourselves to think we represent even a tiny portion of the market and we really dont. Enthusiast parts make up such a tiny part of their business they could literally lose all of us and not care or notice.

For the average user, these chips are fine.

For Intel/AMD we as overclockers don't even signify an ant.

That said the study was repeated with the solder that Intel suggested and it could crack off yes but a 3rd party will never have the resources to solder on them like Intel does not are they aware of exactly which one they use.

As much as Intel isn't stupid the dies can be soldered without issue if they alter the composition which allows for that.
AMD has managed to do so even in all their CPUs (regardless of their performance vs. Intel) .. and I highly doubt AMD uses some magic sauce that makes everything doable.

But for the average user it is indeed enough.

Those that want a safe de-lid:

http://rockitcool.myshopify.com/

Have fun.

[Butler to Baby Sloths]

I don't buy the story in the OP. Smt fishy.



Even for an enthusiast they may be fine.

Take me, for instance. 99% of the time my 4770k runs at stock. (I can't even be arsed to change the Single Core Turbo at stock to Multi Core Turbo at stock from the ASUS ROG utility). I only overclock it when playing GTAV or ROME II TW or Elite Frontier or SWTOR, which is not very often.

Because I just can't be arsed to get into the BIOS and do a proper overclock, I OC the CPU from the ASUS ROG utility which only allows overclocking up to 4.1 GHz without running into some serious issues.

So, there you have it.

The bottom line is, CPU's nowadays are plenty fast for gaming even at stock or near stock speeds. Back in the Core 2 era and earlier during the Pentium 4 era, you simply could not game without heavy overclocking. You HAD to overclock else your game simply wouldn't run.

Yeah, my 5820k sometimes doesn't even bother going above 4.2 GHz in WoW. And that's with one core loaded at 90% and the rest idling.