If anyone wonders if the IHS and bad TIM are why IB can't be cooled. (hint, yes)

[Ferzerp]

I give you this.

Yes, I shut if off right after the capture. BIOS was set to 1.35. 1.392V freaked me out. It doesn't likely need that much voltage, but I was really just seeing if it would hit 105C at those settings. I really have a lemon OC'er. 4.4 is all it really likes.


edit: to clarify, this is with the intel TIM replaced, but the IHS left on at a stupid high voltage.

 

[r3pshow]

Will not bad ...even the temps on full load max 86c ...damn you Intel for using cheap TIM >_<...!!

just a Quest : Where the heck are the other 4 threads ? have you disabled them?

 

[Ferzerp]

I turned off HT, and all the power saving features to bump it to the highest clock I could get it. It wouldn't do 4.8 at those settings either.

 

[Ben90]

Lol, My 920 would love to see those load temps.

 

[magomago]

clearly only getting to 4.4Ghz means you only have a 'lemon' overclocker....

*sarcasm*

 

[TakeNoPrisoners]

clearly only getting to 4.4Ghz means you only have a 'lemon' overclocker....

*sarcasm*

Well he does have a custom loop so it really does mean a lemon.

 

[Ferzerp]

Well he does have a custom loop so it really does mean a lemon.

I only have a corsair H100.

I might be able to do 4.5 after all the cooling modifications I've done, but I'll check after I finish. Right now I am keeping the settings the same so that I have a proper comparison between the little things I am doing.

 

[TakeNoPrisoners]

I only have a corsair H100.

I might be able to do 4.5 after all the cooling modifications I've done, but I'll check after I finish. Right now I am keeping the settings the same so that I have a proper comparison between the little things I am doing.

A $100 water cooler can still work well.

 

[rgallant]

just saying again
can we know if ib 85c is really 85c or is it just a number pulled from the software's butt?


"RealTemp 3.70 should be reading Ivy Bridge CPUs correctly but I don't have an Ivy CPU to confirm this and most of the people that do have Ivy CPUs are not talking.

From what I understand, the temperature data is still in the same register as previous Core 2 and Core i generations and the TJ Target or TJ Max register is also still in the same location as before so RealTemp and other monitoring applications should work as is.

There is a rumor that Ivy ES processors have TJ Target set to 91. Here is the formula that all software uses.

Reported Temperature = TJ Target - Digital Sensor Reading

All software assumes that the TJ Target value written to each CPU core is the same as actual TJ Max but the two values might not be the same. Intel has previously stated that actual TJ Max might be higher. Actual TJ Max might be 91, 98 or 100, 105 or some number in between. It might be consistent for all cores or it might be completely different from core to core. No one knows and the few people inside Intel that do know are not talking.

If there is a pile of unknown error in the TJ Target / TJ Max value and the sensors are far from 100% accurate from idle to full load then reported temperatures are more like random numbers.

---NOTE :::::::>>>>>I have read some forums where all sorts of conclusions are being drawn on how hot Ivy Bridge runs. That's nonsense. These conclusions are based on temperature data that can not be relied on for any of Intel's recent CPUs. Temperature data from Intel CPUs should not be compared to other CPUs in the same CPU family let alone comparisons to CPUs from different families.

RealTemp is a good program and the sensors are useful enough so that if you swap a heatsink you can see if things are better or worse but beyond that, Intel's core temperature sensors are not accurate enough or documented well enough to provide users with 100% accurate core temperatures. "


http://www.techpowerup.com/forums/showthread.php?t=64185&page=40

post 993
-Temperature data from Intel CPUs should not be compared to other CPUs - I think sum's it up petty good.

-different tmax. per core could expand the 1-4 core temp. variance maybe

 

[Avalon]

Were you hitting 105C before replacing the TIM?

 

[Ferzerp]

Somewhere in the low 1.3xx range, yes, Avalon.

 

[Avalon]

So I read in your other thread that the die was not touching the IHS, it was just touching the stock TIM. If I sand down the mounting contact of the IHS, it will sit lower and not as much TIM will be needed, correct?

I am definitely interested in trying this.

 

[Ferzerp]

That's correct. The stock contact is really, really poor. I don't know how my contact is now, I didn't check it well after the last time I pulled it apart (removed the double sided tape I'd used, and cleaned off the last of Intel's adhesive rubber stuff).

I need to get some coarser sandpaper because trying to lower it with 600 sandpaper just takes too long, but that was the coarsest I wanted for lapping.

 

[TidusZ]

what process did you use to actually remove the IHS?

 

[Ferzerp]

A razor blade and some patience. You can probably just cut it off, but I was being very careful.

 

[nyker96]

intel really dropped the ball on this one, why use such cheap TIM on a 200-300 cpu? I wonder if it's just oversight or as someone suggested a real intentional mistake to boost 2011 cpu sales.

 

[Rifter]

This makes me happy im skipping ivy and waiting for haswell.

 

[Don Karnage]

I might Delid tomorrow

 

[bryanW1995]

This makes me happy im skipping ivy and waiting for haswell.

Reports like this make me so sad that I'm skipping ivy and waiting for haswell.

 

[tigersty1e]

this makes me sad because i'm probably moving to IB very soon for the power savings alone.

 

[Rifter]

Reports like this make me so sad that I'm skipping ivy and waiting for haswell.

really? were you looking forward to delidding a CPU just for the fun of it?

 

[MisterMac]

One thing makes me wonder, if it's not the TIM so much as, the piss poor ihs to core contact?


Ferzerp, the space between your ihs and core now, is less than stock correct?
TIM is maybe partly responsible, but the massive gap (in terms) could also be the major sinner here?
(which would look like a oversight rather than on purpose move imho).

 

[Rifter]

One thing makes me wonder, if it's not the TIM so much as, the piss poor ihs to core contact?


Ferzerp, the space between your ihs and core now, is less than stock correct?
TIM is maybe partly responsible, but the massive gap (in terms) could also be the major sinner here?
(which would look like a oversight rather than on purpose move imho).

I would tend to agree that a large gap would have a much larger effect on cooling than cheap TIM.

If so could this be solved simply by sanding down the bottom of the IHS to make the gap smaller?

 

[Idontcare]

I would tend to agree that a large gap would have a much larger effect on cooling than cheap TIM.

If so could this be solved simply by sanding down the bottom of the IHS to make the gap smaller?

Good question. I can foresee investigating this a bit.

Reminds me of what I found when I delidded my GTX460. The interface between the GPU silicon and the underside of the IHS was just terrible.

Thread: Delidded my GTX460...[update 9/18] results are in!

 

[Silverforce11]

Reports like this make me so sad that I'm skipping ivy and waiting for haswell.

You already have SB and a good OC, why would you even be tempted with IvB?

That applies to every other SB owner who "upgraded"* to IvB, are you all mad?

* By "upgraded", I really mean down-grade, as SB at >4.5ghz is easier/cheaper to achieve and it runs cool and the 3-5% IPC difference is meh when IvB can't clock as high as often.