Ivy Bridge Temperatures Could Be Linked To TIM Inside Integrated Heatspreader: Report

[kiwi]

This is fine and all but customers SHOULD NOT have to disassemble a CPU package, replace the TIM, and reassemble the package just to use their K-branded overclocking CPUs at safe temps.

INTEL NEEDS TO FIX THIS IMMEDIATELY!

Right, you should not raise voltage either They do not have to do anything unless it is overheating at stock settings and many customers complain about it

[Trox]

Why releasing K series CPUs then? I think is more a marketing strat than an engeneering problem.

Cheers

[Mats]

Why releasing K series CPUs then?

To boost Z77 sales??

Seriously, mobile market is more important than ever, they don't have any lids but that doesn't matter because they won't be overclocked anyway.

Intel has been limiting overclocking step by step for years.
No budget CPU overclocking, limited maximum multiplier on "unlocked" CPU's, only possible to overclock with Z(P) motherboards. . .

With little competition, declining desktop market, and a competing high end LGA2011 platform, things aren't going to change for the better anytime soon for 1155/1150, if ever.

[Unbornchild]

Right, you should not raise voltage either They do not have to do anything unless it is overheating at stock settings and many customers complain about it

Wrong. Intel is not against (responsible) overclocking. If they'd start to be, their horny CPUs and the high end gamers boards (like the Asus ROG series) would stop being popular amongst overclockers and high demanding gamers with money.
Just before launch of the 2011 socket i built this system i have with a 2700K. I'm glad it's soldered! 4.8G on air no prob.

[kiwi]

Being not against overclocking is one thing but they never guarantee that you will achieve any overclock. And in this case you can still get pretty decent results - customers just often expect way too much

[Kallenator]

Being not against overclocking is one thing but they never guarantee that you will achieve any overclock. And in this case you can still get pretty decent results - customers just often expect way too much

Very true kiwi! However Intel has severely tightened the overclock-ability of theirs CPU's over the last few years, so when the models that are aimed for oc by having an unlocked multiplier starts struggling because of what seems to be shortcuts in the production, then people will get annoyed. We do pay extra for it after all

[UrbanSmooth]

http://en.wikipedia.org/wiki/Haswell...rchitecture%29

[ReKcOlNu]

Very true kiwi! However Intel has severely tightened the overclock-ability of theirs CPU's over the last few years, so when the models that are aimed for oc by having an unlocked multiplier starts struggling because of what seems to be shortcuts in the production, then people will get annoyed. We do pay extra for it after all

True but for the performance you get the prices are pretty low IMO, it wasn't that long ago new CPUs from AMD/Intel costs much more then prices we see now and only the very expensive hi-end chips had unlocked multipliers.

My HTPC has an AMD64 X2-4200+, the board/CPU are now 6.5 years old, the (multiplier locked) CPU cost $400 and that was just after the price dropped from $549!

If I was Intel I would of added another $50 on top of the price of "K" models.

[Conquistador SW]

If I was Intel I would of added another $50 on top of the price of "K" models.

They do in Europe, or at least in the Netherlands

[onewingedangel]

The choice of TIM is likely a choice to limit overclocking and enure Sandy Bridge-E keeps a niche on the desktop.

It doesn't affect power consumption, temperatures are within acceptable limits at stock, and notebooks where it's more critical don't have an IHS.

Any overclocking headroom is a bonus - your buying a product that must work at stock, and Ivy Bridge does.

[BeepBeep2]

The choice of TIM is likely a choice to limit overclocking and enure Sandy Bridge-E keeps a niche on the desktop.

It doesn't affect power consumption, temperatures are within acceptable limits at stock, and notebooks where it's more critical don't have an IHS.

Any overclocking headroom is a bonus - your buying a product that must work at stock, and Ivy Bridge does.

But the K models are made for overclockers.

Of course, you find K models in OEM machines too, but part of that is because on SB, 2600K had better integrated GPU than 2600.

[onewingedangel]

But the K models are made for overclockers.

The unlocked multiplier is there to give more flexibility when overclocking, but you're still not guaranteed anything over stock turbo speeds.

[mav2000]

Did anyone read what Asus had to say? Basically that core temp and real temp are not reading the temps right?

[nullface]

Did anyone read what Asus had to say? Basically that core temp and real temp are not reading the temps right?

Wait what?

[mav2000]

Yup it's there I think at hard forums if I am right.

[RPGWiZaRD]

Well I wouldn't be suprised, in fact I also suggested something like that earlier as to me it doesn't make sense how the temps suddenly start climbing drasticly at around 4.7~4.8GHz while the power consumption doesn't quite rise as quickly, even at those speeds it's still not using as much power as Sandy at similar speeds in terms of power consumption, for 20~25C differences between Sandy and Ivy at that point it doesn't seem realistic. Neither does it when it keeps showing 100C load temp but u can still keep it priming.

I can't be the only one having some slight doubts?

[rge]

Here is kevin's (unclewebb/realtemp programmer) response to the nonsensical comment that someone claims was made by someone at ASUS that some temp programs are not updated.
http://hardforum.com/showpost.php?p=...8&postcount=48

The person who claims to have contacted ASUS and reported back, claimed initially that AIDA64 was reading core temps correctly, but Coretemp and Realtemp needed to be updated. Which is interesting since AID64 is reading same temps as Coretemp and Realtemp (within sampling error times, ie 1-2C). That same person initially thought core temps were way off since a huge difference was found between 2 cores while loaded, then realized prime had failed on that core, hence the large difference, which is what set the nonsensical theory in motion, and launched another rumor from people who cant bother to read the whole thread and realize it was a nonsensical user error.

Realtemp, Coretemp, AIDA64 and any other software simply read tjmax directly from cpu register (you can use any MSR editor, go read tech docs find the register and read it yourself), then read distance to tjmax again directly from cpu register, and provide the result. Nothing to screw up. You may see 1-2C difference in temp programs since their sampling times vary. Also HWM and others may read a rolling avg vs realtime of Coretemp and Realtemp, again couple C difference.

Basically to claim realtemp and coretemp are wrong, is to say intel programmed tjmax wrong and/or distance to tjmax is wrong. But then how would AIDA64 be accurate, when it is giving the same temps reading the same information from cpu. While slope error occurs at low end causing low reading core temps on IVY, at high temps, coretemp, AIDA64, and realtemp are as accurate as intels calibration since they are simply reading the same info, albeit sampling at slightly different times.

Ivy is just hot, crappy tim1 + higher thermal density.

[mav2000]

Well he funny thing is my water temps are not rising along with running something like prime, even though the cores are hitting 80 odd degrees. The water temp tends to rise only about 2-3 degrees, which is a bit abnormal. This is the first time I am using intel, but in the case if amd, water temps would rise by 5-6 degrees on full load with the gpu going in bf3. And you could feel the hot air from the rad and the rad would be warm to the touch. With ivy this does not seem to happen.

[rge]

Water temps will rise proportionally to total wattage not core temperature. Higher watts dissipated = higher water temps. Ivy isnt high wattage, it is high thermal density + poor tim1 interface.

My gpu consumes 200W, but core temps on gpu are only 50C while consuming 200W because thermal density is lower. My core i7950 only consumed 150W at 4.2, 1.28vcore, but temps were much higher than 50C. GPU heats water more since higher wattage, but cpu temps much higher since higher thermal density.

Bottom line water temps is not a measure of core temperature, it is proportional only to total wattage.

Take 150W and spread the consumption over a large area, and temps wont be much higher than ambient. condense the area that power is consumed into a small enough area and temps will exceed that of a nuclear reactor (see intel slide on thermal density previously shown). Even though total wattage would be low.

[R101]

Does throttle watch work with Ivy? What does it say when temps get really high?

[amlett]

I've posted a review here with different TIMs used. 25ºC dropped form intel TIM+AS5 to liquid pro+indigo xtreme:

http://www.xtremesystems.org/forums/...P-IX&p=5104115

[cobra_kai]

I've posted a review here with different TIMs used. 25ºC dropped form intel TIM+AS5 to liquid pro+indigo xtreme:

http://www.xtremesystems.org/forums/...P-IX&p=5104115

Wow fantastic results. You are tempting me to pop the top on mine...

[Nickel020]

Well he funny thing is my water temps are not rising along with running something like prime, even though the cores are hitting 80 odd degrees. The water temp tends to rise only about 2-3 degrees, which is a bit abnormal. This is the first time I am using intel, but in the case if amd, water temps would rise by 5-6 degrees on full load with the gpu going in bf3. And you could feel the hot air from the rad and the rad would be warm to the touch. With ivy this does not seem to happen.

You're making an apples to oranges comparison here. Water temp is based on power consumption of the cooled hardware, it's actual temperature is pretty much irrelevant. So water temp is of course low with Ivy, as power consumption is relatively low for a CPU that's at 80C with a good watercooler (relative to other CPUs that hit 80C). To hit 80C with Sandy and the same cooling setup, you need to increase VCore more and have significantly higher power consumption, so your water temp will then be significantly higher. Meaning you can't compare water temp based on CPU temp, only based on power consumption. Another example: If you were to use a really bad watercooler on Sandy, your temp will drastically increase, but water temperature will remain the same, as power consumption is the same.

[iboomalot]

Is there anyway to replace the TIM and reglue the IHS back on and not loose warranty?

[hokiealumnus]

Only if the person at Intel has NO idea what they're looking at.