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

[hokiealumnus]

MX-4 probably would, but not as well as the Liquid Pro. I'd steer clear of AS5 because of its (very minor, probably inconsequential but present nonetheless) capacitance, lest it touch something it shouldn't.

[Mats]

+1

No AS5 without a lid.

[Manicdan]

Why is it that some people seem to see a huge difference in temps after delidding and others see almost none? ...but changing thermal paste nets 10-15c+?

O_o

if they removed the IHS and put the heatsink directly on the chip, its possible the difference could be from bad contact between the chip and IHS, instead of just the paste being used.

a simple way to know this is by reusing the intel paste with the IHS off and see how much changes. but right now it could be 3 things. bad contact, bad paste, or the IHS itself adding extra insulation. but thanks to the way they tested we honestly dont know how much either of these things is really impacting it.

[Mats]

I guess bad contact is less of a problem when using solder, unlike with paste. So if the solder just got replaced with paste without improving contact, that may be the issue. Purely speculating, of course.

[HKPolice]

This makes perfect sense for Intel, they want enthusiasts buying SB-E instead of Ivy because there's more profit for them selling the LGA2011 platform.

[M.Beier]

I have a 3770k coming next week. Really tempted to pop the top on her now. Gonna be water cooled so might even be able to go above 5ghz. Would tim like as5 and mx-4 work?

Dont do it...
you void warranty, and you wont gain much, the issue is the contact surface...

[zalbard]

Why is it that some people seem to see a huge difference in temps after delidding and others see almost none? ...but changing thermal paste nets 10-15c+?

O_o

Some put it back together better than others. Also, it is not the first time liquid metal TIM blows everything out else of the water. It really shines in certain applications.





An interesting part about this article is that they put the heat spreader back on after changing the TIM. So there's no need to disassemble the socket mechanism and do weird stuff with your rig. Very nicely done if you ask me.

Dont do it...
you void warranty, and you wont gain much, the issue is the contact surface...

I suppose you mean surface area.
That's true, but I am not convinced that's the only issue. From what I've seen on various forums so far it appears that IVB temps can widely vary from setup to setup. Which leads me to believe that there is often poor contact between IHS and CPU die (it's not like it wasn't the case in the past, look up Bloomfield, for example, some people had horrible temps, some were fine), and in such cases TIM used does make a big difference.

[RussC]

I can see where removing the IHS and putting the heatsink directly on the die is problematic for most. W/O the porper spacing/spacer or whatever is needed, users trying to do this seems very hit and miss to me. Its very hard to get the heaysink exacly flat on the die with out the IHS or some other PROPERLY engineered spacer for the socket mechanism.

I wouldnt do it myself.

It seems the differneces from cpu to cpu as far as temps go seem to mostly be very high, as oppesed to the temps being on the low side.

RussC

[JaD]

As well as considerable added overclocking headroom.

http://translate.google.com/translat...11_532119.html

[BoredByLife]

I don't understand this move by Intel.

Better thermals result in better lifetime expectancy. At standard conditions it is already a 10°C difference.

[Gorki]

I wonder would Intel revise their Ivys now?

[boxleitnerb]

100 MHz is significant? Oh boy...

[Iconyu]

I wonder would Intel revise their Ivys now?

The big problem with Intel's Tick-Tock is that it makes people afraid to invest in hardware, thus I always thought Intel would do something to make Sandybridge-E look like a good better buy even against Ivy Bridge. I didn't think it would be this transparent however. So soon as I heard that Intel had switched to a TIM under the heat spreader and that it might be the cause of the clocking issues, I thought that it wasn't just a cost cutting move.

Announcing a revision would just damage sales on these chips, so Intel will probably leave it alone and keep doing what they are already doing. The hard core among us know what to do now.

[paulbagz]

+1 Iconyu

I was keen for IB but then with IB-E on the horizon (or possibly not) what do I go for?

Then there's the chatter that Haswell is a definate March 2013.

So hard to just know when to pull the trigger on buying these things.

-PB

[zalbard]

Repost, plus we already have a thread going regarding this issue in this very same section of forums.

I was keen for IB but then with IB-E on the horizon (or possibly not) what do I go for?

Then there's the chatter that Haswell is a definate March 2013.

Neither will be out before 2013 it seems.
And IB is a huge upgrade over your Core 2.

The hard core among us know what to do now.

Just need to find some good glue to put the IHS back on...

[zalbard]

Another person with successful results: PcCI2iminal - 3770k IHS Removals - CPU temp dropped from 79C to 71C

[iboomalot]

100 MHz is significant? Oh boy...

brings back memories getting 100mhz on my 800mhz Slot-A

[[XC] Oj101]

Removed

[Buckeye]

As well as considerable added overclocking headroom.

http://translate.google.com/translat...11_532119.html

Hey Jad, I merged this thread with the other main one that has been going for sometime now.

[Gorki]

Out of curiosity, how good contact would be placing bare chip in socket mbo with IHS removed? I suppose you should be careful with pressure when tightening up cooler/block?

[Mechromancer]

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!

[Andrea deluxe]

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!

agree

[zoob]

Regular locked IVB CPUs will probably never reach temps where using a paste becomes an issue, even using just a stock cooler.

If you want lower temps, go ahead and re-paste your IHS with the potential of damaging your chip in the process -- we've been doing this since A64.
Quit whining and void those warranties. Most of us are going to ditch Ivy as soon as Haswell is released anyway.

Going forward I wouldn't be surprised if Intel started charging an even larger premium to have the IHS soldered on the unlocked series, making it differentiated by not only the extra 100 MHz and lack of VT-d.

[K404]

Has anyone tried with GC-Extreme yet?

[zerotol]

so now we have tests concluding that removing the IHS works in lowering temps and removing the IHS does nothing at all and mounting on a naked core drops temps and mounting on a naked core does nothing.

conclusion: we are right where we started, nothing conclusive can be decided.