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

The discussion is on TIM vs solder. All the people did in your's was change the TIM which will obviously not help much, if at all.

Aye. Changing TIM between best and worst will give a few degrees benefit at most. I'm beginning to wonder if Intel didn't use solder due to heat concerns. It's possible with the smaller die they couldn't find a way to apply the solder without the heat of the process becoming an issue. :confused:
 
If you are going to remove the IHS, why not solder it back on instead of just replacing the thermal paste?

1) Take off IHS
2) Heat IHS with a propane torch
3) Apply solder to IHS while keeping it hot with the propane torch
4) Remove heat of the propane torch and quickly align the Ivy Bridge CPU on the IHS
5) Solder will cool fusing the IHS and the die together

Might want to create a jig for aligning the IHS and the CPU, but it should be doable. Who wants to be the first to try? :D

Just don't screw it up. I can see the cockeyed IHS now.
 
Seriously now, it would be far simpler and far less risky (I said LESS risky) to just take the IHS off and jimmy an aftermarket cooler onto the die itself.

Honestly, if the chips look like the pic we have seen it should be fairly easy to get a Coolermaster or similar unit onto these things without breaking the die. We did it for years with the old AMD Athlons...

Why go to all that trouble? You think your warranty would be maintained and Intel wouldn't notice if you sent the chip in? Yeah, right.

In some ways, it would be exciting if it was that easy to take the top off. You could get DRAMATICALLY better temps by bolting straight to the die. A real enthusiast maneuver for once.
 
Might want to create a jig for aligning the IHS and the CPU, but it should be doable. Who wants to be the first to try? :D
Any temperature high enough to get the solder liquid on the die (not sure it would even stick) will likely heat through it to the tiny solder balls used to mount the CPU die on the LGA packaging. What you're suggesting is a terribad idea as it would likely destroy the CPU.
 
Seriously now, it would be far simpler and far less risky (I said LESS risky) to just take the IHS off and jimmy an aftermarket cooler onto the die itself.

Honestly, if the chips look like the pic we have seen it should be fairly easy to get a Coolermaster or similar unit onto these things without breaking the die. We did it for years with the old AMD Athlons...

Why go to all that trouble? You think your warranty would be maintained and Intel wouldn't notice if you sent the chip in? Yeah, right.

In some ways, it would be exciting if it was that easy to take the top off. You could get DRAMATICALLY better temps by bolting straight to the die. A real enthusiast maneuver for once.

One problem is that the retention mechanism works on the IHS, so removing that would make it impossible to clamp the chip into the socket - so you'd have to hope the cooler was able to do that effectively.

Different situation, of course, but I don't remember guys getting dramatically different temps when they were taking the lids off Athlons - something on the order of 5-7C if I remember right.
 
Seriously now, it would be far simpler and far less risky (I said LESS risky) to just take the IHS off and jimmy an aftermarket cooler onto the die itself.

Honestly, if the chips look like the pic we have seen it should be fairly easy to get a Coolermaster or similar unit onto these things without breaking the die. We did it for years with the old AMD Athlons... .

You have to remember with today's Intel chips, we have a load plate that grabs the IHS to hold the CPU in place. Much different than the old AMD Athlons.

On Intel, if you remove the IHS, the load plate would have to clamp down directly onto the cpu pcb and may not hold the cpu in place correctly or worse crack the pcb.

Edit: ^ Forceman beat me to it
 
Any temperature high enough to get the solder liquid on the die (not sure it would even stick) will likely heat through it to the tiny solder balls used to mount the CPU die on the LGA packaging. What you're suggesting is a terribad idea as it would likely destroy the CPU.
That is why I said melt the solder on the IHS and not on the die itself. The trick is to have the IHS and solder cool down fast enough to not melt the solder bumps. It should work with the right amount of heat. Not that I would ever try it (unless I had $300 I didn't care much about).
 
Looks like TechReport got word back from Intel on the issue. Seems we might just have to deal with it.

Curious, we asked Intel about the interface between the Ivy Bridge die and the heat spreader. Intel has confirmed to TR that Ivy uses a "different package thermal technology" than Sandy Bridge. The firm stopped short of answering our questions about why the change was made and how the thermal transfer properties of the two materials compare. However, Intel claims the combination of the new interface material and Ivy's higher thermal density is responsible for the higher temperatures users are observing with overclocked CPUs

http://techreport.com/discussions.x/22859
 
Seriously now, it would be far simpler and far less risky (I said LESS risky) to just take the IHS off and jimmy an aftermarket cooler onto the die itself.

Honestly, if the chips look like the pic we have seen it should be fairly easy to get a Coolermaster or similar unit onto these things without breaking the die. We did it for years with the old AMD Athlons...

Why go to all that trouble? You think your warranty would be maintained and Intel wouldn't notice if you sent the chip in? Yeah, right.

In some ways, it would be exciting if it was that easy to take the top off. You could get DRAMATICALLY better temps by bolting straight to the die. A real enthusiast maneuver for once.

So I can remove the IHS and use some Noctua NT-H1 applied to the die and then remount my H100, sounds simple.
 
Looks like TechReport got word back from Intel on the issue. Seems we might just have to deal with it.
Wow. Looks like extreme overclockers may be removing/replacing the IHS after all. At least that will solve one of the two problems.
 
Let's be honest, upgrading an E8500 C2D, I’ve never had a quad-core, never had USB 3.0, am I going to be missing out on anything if I just get a 2700K and call it a day? This is getting perplexingly out of proportion.
 
Let's be honest, upgrading an E8500 C2D, I’ve never had a quad-core, never had USB 3.0, am I going to be missing out on anything if I just get a 2700K and call it a day? This is getting perplexingly out of proportion.

Depends on what you're doing. For 99% of users, no.
 
Wow. Looks like extreme overclockers may be removing/replacing the IHS after all. At least that will solve one of the two problems.

Maybe a few, but I doubt most will see it as a worthwhile endeavor, even on the more extreme end.
 
Let's be honest, upgrading an E8500 C2D, I’ve never had a quad-core, never had USB 3.0, am I going to be missing out on anything if I just get a 2700K and call it a day? This is getting perplexingly out of proportion.

The 2700k is still a quad core with hyperthreading.

Most boards have had had third party USB 3.0 controllers since SB's launch. The only platform feature you are missing out on is PCIe 3.0 with SB vs. IB.
 
Let's be honest, upgrading an E8500 C2D, I’ve never had a quad-core, never had USB 3.0, am I going to be missing out on anything if I just get a 2700K and call it a day? This is getting perplexingly out of proportion.

To be honest, you probably aren't going to be missing out on anything by just getting a 2500K instead of a 2700K.
 
To be honest, you probably aren't going to be missing out on anything by just getting a 2500K instead of a 2700K.


Well hyperthreading but thats debatable for use. I have tested with 12 gigs of ram with hyper threading enabled and disabled and basic multitasking in windows is def better with hyper. But it's not always 100% necessary and for gaming, it can even be a detractor.
 
To be honest, you probably aren't going to be missing out on anything by just getting a 2500K instead of a 2700K.

Yeah, this is starting to get ludicrous. I'm in the same boat as MeanBruce, and I'm no stranger to overclocking. If I can get similar thermal results at similar performance levels from a 2700K or even a 2500K as I would a 3770K, I'm just gonna get the Sandy Bridge chip. I'm not even running PCIx 3.0 cards in the new build (Crossfired Radeon HD 6850s). What's the point of blowing top dollar on something new if I could have the same performance (roughly) at the same thermal level?I'm going to be doing a lot of gaming, a lot of music-making, and maybe a bit of 3D work. My C2D system lasted me three years before this new build, how long would a SB system last? Well into Haswell, I'm assuming.

edit: Okay, here. If I overclocked a 2500K, a 2600K, and a 2700K up to 4.0GHz on the same motherboard, what would the differences in performance be?
 
Last edited:
I'm going with Ivy Bridge. They run hotter but I don't think I'd take it past 4GHz so I'm not overly concerned. I'm upgrading from a Core 2 Duo and the 2500k is still $210-230 around here so unless the 3570k is way overpriced I'm gonna go with that.
 
Yeah, this is starting to get ludicrous. I'm in the same boat as MeanBruce, and I'm no stranger to overclocking. If I can get similar thermal results at similar performance levels from a 2700K or even a 2500K as I would a 3770K, I'm just gonna get the Sandy Bridge chip. I'm not even running PCIx 3.0 cards in the new build (Crossfired Radeon HD 6850s). What's the point of blowing top dollar on something new if I could have the same performance (roughly) at the same thermal level?I'm going to be doing a lot of gaming, a lot of music-making, and maybe a bit of 3D work. My C2D system lasted me three years before this new build, how long would a SB system last? Well into Haswell, I'm assuming.

edit: Okay, here. If I overclocked a 2500K, a 2600K, and a 2700K up to 4.0GHz on the same motherboard, what would the differences in performance be?

Well the 2600k and 2700k are identical except for clockspeed.

The 2500k is a bit trickier. First the minor difference in L2 cache means the 2600k/2700k would have a minor (and I emphasize this) advantage depending on applications in performance.

The larger difference of course is hyper threading. In some situations hyper threading is actually detrimental (or causes other issues) to performance, of course you can always disable it should this issue crop up. A notable example of this would be BF3 having issues with HT at launch. For gaming in general at this point you will not see an significant performance difference between HT and no HT. Now for your productivity apps these are more likely to take advantage of HT such as with 3D rendering. I believe most 3D rendering tests show ~10-20% improvement with the 2600k over the 2500k.
 
edit: Okay, here. If I overclocked a 2500K, a 2600K, and a 2700K up to 4.0GHz on the same motherboard, what would the differences in performance be?

What limitedaccess said, with the addition that the difference is almost certainly less than the $100 price difference, unless you are using it for your job and time is money.
 
Well the 2600k and 2700k are identical except for clockspeed.

Oh, derp. I knew that. Sorry, haven't slept much.

The 2500k is a bit trickier. First the minor difference in L2 cache means the 2600k/2700k would have a minor (and I emphasize this) advantage depending on applications in performance.

Is it anything that would matter in the long run? Are we talking like 3-5%, or .3-.5%?

The larger difference of course is hyper threading. In some situations hyper threading is actually detrimental (or causes other issues) to performance, of course you can always disable it should this issue crop up. A notable example of this would be BF3 having issues with HT at launch. For gaming in general at this point you will not see an significant performance difference between HT and no HT. Now for your productivity apps these are more likely to take advantage of HT such as with 3D rendering. I believe most 3D rendering tests show ~10-20% improvement with the 2600k over the 2500k.

So Office and Poser and Photoshop are going to be a little bit faster with the i& over the i5 then. Question is, is it worth the increase in cost over the i5... hm. God, this was so much easier when Ivy Bridge didn't have these problems. :<
 
The discussion is on TIM vs solder. All the people did in your's was change the TIM which will obviously not help much, if at all. So I don't see how you can say this discussion is not true.

And he didn't use a shim for the socket clamp? If that's why he says there HAS to be an IHS in place, then wtf. :rolleyes:
 
And he didn't use a shim for the socket clamp? If that's why he says there HAS to be an IHS in place, then wtf. :rolleyes:

You'd think the heatsink would provide enough force to keep the chip in place even without the retention bracket - but it kind of sounds like the chip bends when there is no IHS to provide rigidity (depending on how you interpret those comments), so maybe that is more the problem.
 
Well, since I'm not ready to pay $8.60 per 1% situational performance increase, I guess it's i5-2500K time for me. The only question is Z68 or Z77.
 
Well, since I'm not ready to pay $8.60 per 1% situational performance increase, I guess it's i5-2500K time for me. The only question is Z68 or Z77.

I'd say Z77. That way, you are ready for a true IB system later on should you decide to jump on IB CPU. Well, then by, Haswell may be out already and might as well go with it instead of an IB system.
 
Well, since I'm not ready to pay $8.60 per 1% situational performance increase, I guess it's i5-2500K time for me. The only question is Z68 or Z77.

What is the price differential you are looking at?

The difference in chipset features between Z68 Gen 3 and Z77 is native USB 3.0 support for the latter while the former relies on third party controllers. Z77 also has a somewhat wider range of memory support, although this is not likely to impact typical desktop builds.

Outside of that you'd really need to consider the boards individually.
 
My only saving grace in this torrid state of affairs is the motherboard I have waited on for months was not released in Z77-round one, nor will it be next month in round two. For some reason beyond knowledge or knowledge beyond reason, Asus has decided to not release or even announce the M5E only hint that it will be coming &#8220;soon after&#8221;. So I am forced to wait out this Ivy fiasco until the dust settles whether I like it or not, hopefully they won&#8217;t discontinue the 2700K over the next quarter, or the 3770K will undergo its first revision. So I finally get my long awaited cpu and mobo in say September (exaggerating for effect) and then 6months later Haswell debuts, geeeeeeeez!

Something unrelated but very cool, Portal 2 DLC to allow gamers to create infinite maps, and its free May 8th. Cave Johnson here, hahaha.

http://www.legitreviews.com/news/12973/

Portal-2-Puzzle-Editor.jpg
 
What is the price differential you are looking at?

The difference in chipset features between Z68 Gen 3 and Z77 is native USB 3.0 support for the latter while the former relies on third party controllers. Z77 also has a somewhat wider range of memory support, although this is not likely to impact typical desktop builds.

Outside of that you'd really need to consider the boards individually.

Well, the only board type I'm getting a Gigabyte with the Ultra Durable feature set, with that big sheet of copper in the board itself. That's some good stuff. Just have to decide between the Z68 and Z77 motherboards, and for that kind of money, I'm going to need to be convinced not to just go Z77.
 
edit: That being said, if it turns out that retail IBs are full of ceramic paste, I'll personally be taking the spreader off mine, removing all the paste, and filling the bastard with AS5.

Sorry Love this :D ^^^^^^ :D I am curious as to what ivy temps would drop to with the IHS off and good heatsink contact direct to die. Could drop the temps 10 - 19. Decisions, Decisions....Might Have to give it a try Tuesday.
 
MAN. Trying to get a new rig up and running for May 15th (Diablo 3), but there's so much damn drama surrounding this release. I'm leaning toward Z77 and an SB 25/2600k, something I can toss on eBay in six months should Intel bust out some crazy new IB stepping or revision that makes the upgrade worthwhile. I'd consider an LGA 2011 board with a 3820, but that's just overkill ... though I do like the guaranteed upgrade path to Ivy Bridge-E (even more overkill).

Or heck, maybe just this: http://ncix.ca/products/index.php?s...ro-Gen3&manufacture=Bundle Deals&promoid=1058
 
Last edited:
I'd consider an LGA 2011 board with a 3820, but that's just overkill ... though I do like the guaranteed upgrade path to Ivy Bridge-E (even more overkill).

I'm not sure how guaranteed that upgrade path is...not to mention that when you consider people are complaining about IB's longevity being somewhat short-lived before Haswell, the reality is that should IB-E come out it could be an even less permanent solution.
 
I'm not sure how guaranteed that upgrade path is...not to mention that when you consider people are complaining about IB's longevity being somewhat short-lived before Haswell, the reality is that should IB-E come out it could be an even less permanent solution.

Indeed. It's not something I'm actively considering ... it's just there ... in the distance ...

I'm really starting to like the ASUS Z68/2500k bundle, actually. PCI 3.0. IB compatible. I'm selling most of my current parts - I already have a case, H100, SSD, my PSU, incoming 680 - so it would almost be a straight trade up at that price. I just need some memory. Then I twiddle my thumbs until Haswell.
 
Last edited:
I am by no means an extreme OCer. I get nervous adding volts. But the idea of having TIM between the chip and IHS, then TIM again between the IHS and Heat Sink seems rather counter productive to me. The less TIM the better, not more the better. Soldering the IHS to the chip seems like a much better idea since metal transfers heat better than anything else. Tisk tisk Intel. I'm staying with my 2500k ;)

For those commenting on removing the IHS and doing direct core cooling, I don't see why it can't be done. I wouldn't use HDT cooler though. The only benefit to having a IHS that I can tell is that you may get the heat sink more flush to the IHS than directly to the chip core.
 
Last edited:
So Office and Poser and Photoshop are going to be a little bit faster with the i& over the i5 then. Question is, is it worth the increase in cost over the i5... hm. God, this was so much easier when Ivy Bridge didn't have these problems. :<

If you're a photoshop user you're better off buying the newer edition of photoshop which has apparently widened the number of ways it uses gpu acceleration.
 
God, this was so much easier when Ivy Bridge didn't have these problems. :<

I'm sick of hearing this. Ivy Bridge does not have any problems. It just isn't what some users hyped it up/expected it to be. And some users had extremely high, unrealistic expectations.

Additionally, if you're only overclocking to 4 ghz or so, Ivy Bridge is better for you. If you want the larger max overclock number, then Sandy Bridge or even Bulldozer is better.

Ivy Bridge is faster clock for clock, and can almost overclock to the same levels as SB.
 
If you're a photoshop user you're better off buying the newer edition of photoshop which has apparently widened the number of ways it uses gpu acceleration.

So PS doesn't see a huge boost from HT? I do a ton of work on Adobe Creative Suite. If I thought I'd see a noticeable benefit, I'd definitely consider an i7 (be it SB or IB).
 
Back
Top