Overclocking Site Solves Mystery Behind Higher Ivy Bridge Temperatures

35

Comments

+ Add a Comment
avatar

PsychNerd

Well there are several issues that have to do with heat in this instance. Yes the process size has shrunk once again, but they are adding more transistors, they are building in layers, and using inferior transference materials. 1 - More transistors does equal more heat period. You just can't help it. 2 - In the past dies were 2D, now that they are structuring dies in 3D suddenly you go from an x/y axis to a x/y/z axises. Heat can not escape the extra dimension as easily they need something between the layers to help dissipate the heat that is trapped. Since carbon and in particular diamonds have an extremely low thermal resistance I will not be surprised to find the use of something like fullerene tubes to help facilitate the release of the extra thermals. 3 - Materials science will be our friends here once again. This will end up having to be part of the manufacturing process, and I will not be surprised if there are not many failed attempts by corporations and the end solution come from university research. I will also not be surprised if the research comes from out of the US since the best research lately especially in materials, computer sciences and mathematical engineering has been coming from out of the country in greater ratios than has been in the recent past.

avatar

JohnP

Two of my i7-3770Ks came in today (one more on its way). The Thermal Spec on the box is 95W. Some folks were saying it would be 77W. I don't think the heat issue will be solved by just a better heat spreader...
EDIT: I found out that the chip actually says 77W in Aida. The reason why 95W is posted on the box is so mobo manufacturers will not make a board that can run Ivy Bridge BUT NOT Sandy Bridge. 95W will be enough to handle either chip.
Want to see an Ivy Bridge love-in?
http://www.overclock.net/t/1250090/official-ivy-bridge-owners-thread

and you thought I was bad...

avatar

Peanut Fox

If you read the information about the chip where it says it uses TIM the wording is exactly the same as what they have for Sandy bridge. I wonder if Sandybridge uses TIM like Ivy Bridge does.

Custom liquid cooling system's making a comeback? I'm interested to see what my setup can do when I get this chip underwater and start applying voltage. I'm hoping to hit 5.0 or 5.2 and then backing the clocks down to a more reasonable 4.5 or 4.8.

I know a lot of reviewers are using the H80 and H100 coolers, but their performance still only hovers around high end air cooling solutions. Ivy Bridge seems to have more head room if you can keep the chips cool enough.

avatar

JohnP

Good story Paul, but the information is changing fast. For instance, someone "delidded" an IB (I didn't know you could do that!) and put a massive heatsink directly on the chip. It didn't help, 4.8K still got up into the 100 degree C range.
http://www.overclock.net/t/1249419/pcevaluation-intel-i7-3770k-temperature-measured-without-ihs

There is a forum with a lot of smart people talking about this topic:
http://forum.avsim.net/topic/372250-more-on-the-hot-ivy-bridge/
One of the best posts states that the density is just too high with the voltage leakage to have any successful thermal solution for high o/c. Another is that the folks who put the cooler directly on the chip did it wrong.

Perhaps Intel found out that it didn't make a hot damn (heh) lot of difference which way they lidded the chip. This will be a hot topic for a while.

Meantime, I will be sticking to 4.4GHz o/c as I find that my computers operating system always becomes unstable long before I run into thermal issues.

avatar

Danthrax66

A lot of mounts now have a "bottom out" point that they won't get tighter than. So if you remove the lid there is less force on the chip which could be a reason for the temps still being high. That's my thoughts on the matter though.

avatar

JohnP

Exactly right Danth. One poster said that they needed to "shim" the cooler (so that it is firmly seated on the chip I imagine) and lap either the cooler or the CPU chip. I will be damned if I am going to try this at home!

avatar

Thral1

I wonder if Intel did this to limit overclocking ability. It may sound crazy, but they've already locked down overclocking on most of their CPUs.

I was thinking about getting the the 3570k, but if I can overclock my SB more then would I really be upgrading? I really hope that it's just a test sample thing or that they revise it for the next stepping.

avatar

JohnP

The consensus from all sites is a firm NO to upgrading from SB to IB. There is only a 3-10% speed advantage and SB o/cs better. Like most folks say, go for a graphics card upgrade first if you are a game player.

avatar

pastorbob

I agree John. That is exactly what I did (upgraded the graphics card) and my current SB system will meet my needs for a long time to come.

avatar

Peanut Fox

If you have a SB chip I see no reason to upgrade to IB. Unless you find something in a new motherboard and the z77 chipset, I'd spend the money on something else.

avatar

RUSENSITIVESWEETNESS

All the reason I need to skip this platform.

Try again, Intel.

avatar

loozer

Maybe this is an actual instance where it will be a good idea to remove the heat spreader.

avatar

TheKitty

I planned on an H100 anyway. We'll see in the next few days.

avatar

Geob1kenobi

Intel believes that if consumers can afford an Ivy Bridge build, then they can afford an H 100 or water cooling too. Nitrogen is an option!

avatar

sumedh

exactly...people who buy these chips are obviously overclockers... and they do know a lot about liquid cooling and stuff.. so i don't think temperature is much of an issue...

( people who are not overclockers and are happy with the default clock speed, a rise in temp wont bother them.. their simple air cooler would be more than sufficient... But some smart-ass people tend to forget this and start blaming and criticizing like fools.. )

avatar

aarcane

Is it possible to remove the IHS, apply AS5 or similar to each of the thermal components, and clamp your water block or heat pipe assembly directly to the bare proc like with older core 2 duo and similar procs?
Is it possible to get out a soldering iron and replace the thermal paste manually?

Perhaps more importantly, Does it help, and if so, when will MaxPC post a how-to guide to "Enhanced overclocking with IHS upgrades", complete with warranty void disclaimer?

avatar

Veni Vidi Vici

Was wondering that myself. I seriously doubt it though. Even if it's possible, I'd bet it's VERY risky.

avatar

Danthrax66

People have been doing this one Intel chips for a long time before they started using solder on them. And I have had many graphics cards where there was no IHS either, so you can definitely do it you just need to be careful when removing the IHS as you might break something else.

avatar

bling581

Most "average" people wouldn't have a clue about this or even care. However temperature matters to OC'ers. I'm just curious how the paste impacted the cost of the processor. If there was no cost difference or it was very minor then they shouldn't be "cutting corners".

avatar

vrmlbasic

This could be bad news for poster JohnP. I kid, I kid ;)

avatar

Brad Chacos

Good thing we have above-average readers!

avatar

Baer

I just decided to keep my i7 920 for a while longer.

avatar

tiger8sugar

Same here.

avatar

cmasupra

Same. I was considering upgrading to Ivy Bridge E when it comes out, but the way Ivy Bridge is acting makes me think IB-E will be a failure as well. Hopefully whatever comes after IB-E will be good enough to justify the upgrade.

avatar

bartyh5

It seemingly isn't the IHS though. At Overclock.net, someone explored the chip without the IHS on and it seems as though it isn't the TIM. It just has to do with the density of the 22nm cores.

http://www.overclock.net/t/1249419/pcevaluation-intel-i7-3770k-temperature-measured-without-ihs

avatar

Strangelove_424

The thermal density issue is a big factor. I wonder what this spells for the future of enthusiast/workstation class desktop chips. I imagine once they go to 22nm tech, and below, that there is no going back to 32nm technology unless it’s for server class chips. It would be sad to see the mobile CPU market destroy the enthusiast desktop CPU market with compromised manufacturing techniques and impossibly small dies. I can easily imagine being forced to buy expensive, server-class chips to do power computing, and looking upon Sandy Bridge as the last of the “good ‘ole days”. I hope it doesn't go that way though.

avatar

Danthrax66

I doubt it, it's likely just an issue because this is the first run it will get better. The first 32nm cpu's ran hot as well but no one remembers them because it was an extreme edition and another high end 1366 chip that no one needed. Sandy Bridge was a second gen 32nm fab and it performed a lot better. Besides Ivy is just a small upgrade and a test at the die shrink the next gen will be the replacement for Sandy Bridge.

avatar

Peanut Fox

Maybe they need a nice big chunky IHS like socket 2011 or 1366. Spread that head load out some more.

avatar

USraging

hmmmm... what should i use for heat dissipation? oh i don't know George how about we just squirt a tube of grease in there? do you think anyone will notice? naaa by the way when your done with that make sure you super glue the mem controller in place...

avatar

Veni Vidi Vici

I don't know why you'd say Intel hit a homerun on this chip. It runs hotter and barely gives any performance boost, yet costs quite a bit more than their respective SB chips.

This release is a total letdown IMO.

avatar

barto

Well, I'll admit that I am disappointed. I plan on upgrading during the summer and I was hoping the Ivy would be better than the SB. As for CPU performance, it isn't. As for graphics, it is, sort-of. Remember, Ivy is a Tick in Intel's Tick-Tock design. This means that Ticks are new tech while Tocks are improvement of the new tech. That is why the Sandy was/is so successful.

avatar

dstevens

And this is how Intel will join the bandwagon of companies that limit what you can do with the hardware that you purchase from them for whatever reason, in this case I'd say to prevent longevity of the CPU desireability by limiting the overclock potential intentionally thereby ensuring another sale sooner

avatar

alexw1234

I always put my own thermal paste on anyways.

avatar

DDRDiesel

You're not understanding what the overclockers are talking about, here. The thermal paste they found was between the IHS (Integrated Heat Spreader) and the die itself. Normally, chip makers user a type of solder that creates a more thermal-friendly barrier to efficiently dissipate heat. Using thermal paste may be a less-expensive option, but can raise temps by quite a bit, as seen in Ivy Bridge

avatar

d3v

Not underneath the IHS, you don't.

Log in to MaximumPC directly or log in using Facebook

Forgot your username or password?
Click here for help.

Login with Facebook
Log in using Facebook to share comments and articles easily with your Facebook feed.