Does Ivy still use cheap Thermal Paste between the Chip?

Well I don't know the plans for our next generation of processors, I can say that there are no plans to change on our current 3rd generation Intel® Core™ processors to anything but the TIM that we have been running.
 
Just de-lid and put a block straight to the die. Intel screwed the pooch on the IB's.
 
I've got a 2600K @ 4.6 24/7 running nice n cool. I really want IB/Z77's new features, but I'm not sure if a 3770K @ 4.6 will be quite as easy or nice n cool. Anybody have advice besides delidding?
Is 4.6 an easily attainable goal for the 3770K? Cooler's a 2-fan TRUE.
 
Ivy Bridge can be tough to cool. No doubt about it. However at 4.4GHz it's just as fast as a Sandy Bridge CPU would be at 4.6GHz. At 4.4GHz or lower they are easy enough to keep cool.
 
Been thinking about taking the lid off mine. I have a 9x140mm rad and at [email protected] i still hits 70c with prime or intel burn. The water temp doesn't change so I know its the crapy paste on the chip not transfering the heat. I figure with no lid I might be able to hit 5ghz. But if you just want to hit 4.4-4.5ghz you can usally do it on 1.2v and keep it cool on air. Mine at 4.5ghz wouldnt break 58c at full load.
 
It's not the paste, it's the large gap between CPU die and the heat spreader that the TIM has to cover. Idontcare, a very knowledgeable guy over at Anandtech forums found this out experimentally. Delidding automatically reduces that gap because you remove the adhesive that holds the IHS in place.
 
It's not the paste, it's the large gap between CPU die and the heat spreader that the TIM has to cover. Idontcare, a very knowledgeable guy over at Anandtech forums found this out experimentally. Delidding automatically reduces that gap because you remove the adhesive that holds the IHS in place.

Wow so it wasn't the paste after all.
 
Why on earth did Intel do this. IB could have been pretty beastly. A much more acceptable upgrade path from SB. It's a side grade at best.
 
Why on earth did Intel do this. IB could have been pretty beastly. A much more acceptable upgrade path from SB. It's a side grade at best.

I would argue that this was an intentional step to limit the overclockability and to make the lga2011 platform a upgrade over lga1155 for some. Although it may just save them a little money in production costs and they are looking at the bottom line..
 
I would argue that this was an intentional step to limit the overclockability and to make the lga2011 platform a upgrade over lga1155 for some. Although it may just save them a little money in production costs and they are looking at the bottom line..

I saves them truck tons of money when you do the math and realize how many of these chips they sell.
 
I don't know about you guys but I de-lidded mine, replaced the cheap TIM with Coollaboratory Liquid Ultra on both the die and IHS and got 4.8GHz @1.36V. My hottest core got up to 74C. I'd say the TIM is the culprit.

SwiftechMCP65548GHz136V.png
 
Did you measure the thickness of the whole CPU before and after delidding? If you replaced the TIM again with the original one, I wouldn't be surprised if the result was the same.
 
Problem is more related to poor contact between the IHS and DIE.
The TIM looks cheap as well.
 
I saves them truck tons of money when you do the math and realize how many of these chips they sell.

It cant possibly cost more than a couple of dollars of which they could easily pass on to the customer.

Hell, I would gladly pay $15 more for a soldered on IHS Sandy Bridge style.

Fuck, an even better idea would be to spend the extra money on the IHS Solder and leave out the stock HSF assembly. Anyone that buys a K series processor and uses a stock CPU cooler is a retard anyways.
 
Agreed. Gonna be a long wait for Haswell. I'm not down with modding a $320 chip and not being able to sell it when I upgrade. Highly doubt anybody would buy a delidded IB. I already take it in the shorts if I lap a CPU and sell it.
 
It cant possibly cost more than a couple of dollars of which they could easily pass on to the customer.

Hell, I would gladly pay $15 more for a soldered on IHS Sandy Bridge style.

Fuck, an even better idea would be to spend the extra money on the IHS Solder and leave out the stock HSF assembly. Anyone that buys a K series processor and uses a stock CPU cooler is a retard anyways.
Excellent idea.
Intel could easily save a shit ton of money and win over enthusiasts again if they did exactly that, they could also leave the non-K processors with the shitty paste also.

In other words:
*Non-K processors=TIM between Die and Heatspreader, comes with HSF. Saves shit ton of money not using solder.
*K series unlocked processors=Soldered Heatspreader to Die. Does not come with HSF just like SB-E chips since anyone who O/C's at all doesn't bother with the stock HSF. Saves a shit ton of money from not bundling a HSF and on reduced shipping costs (less weight) while keeping enthusiasts happy. Throw disclaimer on box and manual stating no HSF included with chip, then offer to upsell a few half decent cooling solutions through their website for the clueless.

Everyone wins, especially Intel. I hope an Intel representative passes along this idea.
 
I've got a 2600K @ 4.6 24/7 running nice n cool. I really want IB/Z77's new features, but I'm not sure if a 3770K @ 4.6 will be quite as easy or nice n cool. Anybody have advice besides delidding?
Is 4.6 an easily attainable goal for the 3770K? Cooler's a 2-fan TRUE.

You should be fine. Mine was a 3570k but I was able to hit 4.8 without delidding. After I de-lidded I went from being temperature limited to vcore limited. A true should do just fine as the heat is actually trapped between the die and the heatsink and the bad intel TIM makes the thermal transfer slower.
 
It cant possibly cost more than a couple of dollars of which they could easily pass on to the customer.

Hell, I would gladly pay $15 more for a soldered on IHS Sandy Bridge style.

Fuck, an even better idea would be to spend the extra money on the IHS Solder and leave out the stock HSF assembly. Anyone that buys a K series processor and uses a stock CPU cooler is a retard anyways.

Except then Intel needs to set up two differnet processes and two chip packaging lines, which can't be cheap. And face it, as far as Intel is concerned the TIM works just fine, so why spend money fixing a problem they don't care about? I really doubt they are losing enough sales to make up for the fixed cost of switching part of their product line back to solder.

They really do need to ditch the stock cooler with K chips though, no matter what else they do.
 
I would argue that this was an intentional step to limit the overclockability and to make the lga2011 platform a upgrade over lga1155 for some. Although it may just save them a little money in production costs and they are looking at the bottom line..

There is a heat wall with Sandy Bridge-E CPUs as well. They take a lot of volts to get over the 4.4GHz-4.5GHz mark as well. There are slightly different reasons for the heat wall but it amounts to about the same thing.
 
I delidded my 3770K and used Cool Laboratory Liquid Pro under the heat spreader and on the water block and got nearly 40C cooler temps. 5Ghz @ 1.36 and only 65C.
 
I delidded my 3770K and used Cool Laboratory Liquid Pro under the heat spreader and on the water block and got nearly 40C cooler temps. 5Ghz @ 1.36 and only 65C.

That's a nice improvement.
 
Any DIY how to avoid putting the lid back on? It seems some rubber washers will help achieve the clamping that is needed to keep the CPU from moving. Or is the die sitting to low for a heat sink to make contact with it?
 
Did you measure the thickness of the whole CPU before and after delidding? If you replaced the TIM again with the original one, I wouldn't be surprised if the result was the same.

Didn't need to measure thickness. The TIM coumpound is what matters. I tried 3 different TIMs. Arctic Silver Ceramique, Arctice Silver 5, and Coollaboratory Liquid Ultra. Both Arctic Silvers were first applied with the line method and spread method, FAIL. I got 95C even at 4.6GHz with water cooling. I then used Liquid Ultra, spread with a brush once, and that's what gave me my temps. I'm no engineer but I believe Liquid Ultra is the best thing next to solder due to its liquid metal compound form.
 
Didn't need to measure thickness. The TIM coumpound is what matters. I tried 3 different TIMs. Arctic Silver Ceramique, Arctice Silver 5, and Coollaboratory Liquid Ultra. Both Arctic Silvers were first applied with the line method and spread method, FAIL. I got 95C even at 4.6GHz with water cooling. I then used Liquid Ultra, spread with a brush once, and that's what gave me my temps. I'm no engineer but I believe Liquid Ultra is the best thing next to solder due to its liquid metal compound form.

I'll have to start giving that a try. I'll need more thermal compound soon. I burn through lots of it as you can imagine.
 
I burn through lots of it as you can imagine.

Same. Every time I see Mwave with their 5x 1.5g MX-2 deals I quickly buy it.

BTW, using MX-2 under the IHS dropped my temps by ~10°C with my 3770K @ 4.5GHz. Going to order Coollaboratory Liquid Pro in a few days. :D
 
I delidded my 3770K and used Cool Laboratory Liquid Pro under the heat spreader and on the water block and got nearly 40C cooler temps. 5Ghz @ 1.36 and only 65C.


At 1.36v 5GHZ my 3570k is right around 65C as well. Have to run the memory 2133mhz CL9 for those temps. I prefer 5GHZ with memory at 2400mhz CL9 with temps in 70s.
 
Last edited:
Really making me think about deliding mine.. Any one took the lid off and lapped the lid? I mean heck the warrenty is already gone.
 
Last edited:
I saves them truck tons of money when you do the math and realize how many of these chips they sell.

Not if enough people are choosing to bypass IB because of it. Companies don't make more money when they skimp necessarily. Word gets out and some who would have purchased decide not to. That may not happen here - Dell, et al. are probably buying enough regardless but as an individual consumer, why pay more for a very minimal upgrade that is much more difficult to keep cool?

Plus it makes Intel look cheap. Which they are, in this case. They had better be sure that TIM won't dry out in a few years or they are going have some pissed off customers.
 
I can't imagine that the enthusiast/OCing crowd is big enough, relative to the normal market, to get Intel to change the way they build CPU's for. Maybe it's not a conspiracy?
 
That's the thing though - the Ks are enthusiast chips, and they are marketed as such - "unlocked" etc. specifically FOR OC'ing.
They need to put their money where their mouth is. This was a bad call at intel.
 
Yes..delid and use gelid pro or coolab liquid ultra.I used a utility knife blade..kinda worked it in nice n slow without too much pressure..keeping a slight up-angle towards the metal and away from the pcb.
 
Not if enough people are choosing to bypass IB because of it. Companies don't make more money when they skimp necessarily. Word gets out and some who would have purchased decide not to. That may not happen here - Dell, et al. are probably buying enough regardless but as an individual consumer, why pay more for a very minimal upgrade that is much more difficult to keep cool?

Plus it makes Intel look cheap. Which they are, in this case. They had better be sure that TIM won't dry out in a few years or they are going have some pissed off customers.

I can't imagine that the enthusiast/OCing crowd is big enough, relative to the normal market, to get Intel to change the way they build CPU's for. Maybe it's not a conspiracy?

The chips run cool enough at stock speeds. Even cooler than Sandy Bridge did. They overclock well and still reach within 200MHz of Sandy Bridge and given the IPC improvement the performance is about the same. And few people are pushing their chips that hard anyway. And it's not like Intel has any competition in this area from AMD. Even though you can overclock the FX series CPUs this high and higher, they still don't out perform their Intel counterparts at anything really. They just close the gap on the stuff they are good at.

So why would Intel change the TIM now when it's not causing an issue?
 
Back
Top