Ivy Bridge heat problem is caused by Intel TIM choice

Japanese website Impress Personal Computer Watch has proven what many have suspected ...

More...

 

yeah too bad it voids warranty

 

Can anyone solder the the CPU die to the internal heat spreader (IHS) with fluxless solder?

 

Wow, talk about shooting yourself in the foot with the enthusiast market.

 

Oops looks like someone at Intel got the thermal paste mixed up with the peanut butter....

 

Imagine a 20c degrees increase for Ivy notebooks. Don't want that on my lap. I'm in the market for a new notebook, some told me to wait for IvyB, but i guess i might as well go for Sandy.

 

Hopefully, this will mean a new revision?

 

Morons...how could they do this? its almost as bad as nvidia using crappy VRMs. But whats with those voltages? Their ridiculously high for a 22nm tech. Did the jap site get a bad chip? or they upped the voltage intentionally for testing only?

 

The voltage needed is the voltage needed.

 

Really annoyed that intel did this, bunch of back to fronts.

 

Yup, and i was about to buy one of these thanks to that flaw im not gettting IB.
Maybe haswell will be better.

 

People, keep in mind that temperature issues are only there (maybe, I have not tried to go extreme) if you OC the chip to hell. With moderate OC it is cooler than SB, at least in my case. In my opinion people are exaggerating this "issue".

If those performance drops would apply for all cases, my chip should run at ~35c under full load sometimes hitting 40c. That in 4,3 GHz.

 

WHats you full load temp at 4.3Ghz ?

Because 4.5Ghz would be least i want from IB, on AIR (Stable and not hot).
Since my i5 750 is rock solid 2 years now at 4.0Ghz.

 

How hard is it to remove the ihs?
And is it possible to directly cool the dice without the ihs?

 

Not really my problem or anyone's. If i could get proper thermal paste with proper temps, even if that would cost me $1 more, i would gladly pay for it. Intel should have known and not skimping on good materials.
This is not about hardware failures/errata or revision problems, this is about reducing costs at the expense of quality.

Perhaps you have a good revision? If so, you're lucky. Or perhaps they were bad batches? Who knows.

I think it will be "fixed" later on, perhaps a new revision. Until then, better to skip IB and to go for SB or Haswell.

And TBH, this is G3D! Peeps here love to take full potential of their hardware. So yeah, OC'ing is one of them and people here don't like this. Higher temps from 10c to 20c is just way too much. Could quite make a difference, plus it could keep you warm in the winters and serve as a grilling stove in the summers. Multifunctional i would say! :banana:

 

That's nasty. My W3680 on the old 32 nm silicon runs at 50 full load stock clocks on the stock cooler and thats with HT on, and this is a 130W chip. What the hell are intel playing at?

 

+1 Im in exactly the same boat.

Either going to be Haswell or AMD refresh if they can sort out the kinks(power, IPC etc).

 

Bought a 3570k, will be mad if i see a rev b in 6 months with good paste/solder.

 

Mustnt forget that the CPUs heat comes out of a smaller die so there are at least 2 issues at play when voltages are raised.
Not all of the high temp is due to the TIM although it looks like it may contribute substantially.

But...
the testers may have put higher pressure on the die/cooler contact, improving thermal transfer, making it look like it is just the TIM causing the improvement.