So did I get it right? I hope so.
The problem with 22 nm dice seems to be micro-fractures that occur due to thermal cycling (eventually leading the chip failure). There's not doubt that Intel knows exactly what the limits are.
But how do microfractures relate to an excessively large gap between the die and IHS? Surely the IHS itself couldnt cause a microfracture. If there was no gap it would be sitting perfectly flush against the die, there wouldnt be any uneven stresses. It makes no sense to have a gap for any reason other than for limiting performance via forced throttling.
I'd say largely right, though it might be more more accurate to say that mainstream chips will not be soldered in spite of enthusiasts, not because of them. The days of OCed mainstream chips nipping at the heels of Intel's E series are coming to a close.
What do you think happens to metal when you heat it and cool it?
Not to mention the pressure. IHS directly on the die would give quite a short lifespan. Maybe no lifespan at all if you use a large cooler. or simply pressed abit too hard at the installation. (Stock cooler with pushpins comes to mind.)
Has anyone in the aftermarket been able to metallize the surface of an IB chip? This would be required to properly solder a chip after delidding.So does a Ivy chip that was delidded and had better solder added OC as much as before? That is to say, are the OCs on these chips thermally limited?
Has anyone in the aftermarket been able to metallize the surface of an IB chip? This would be required to properly solder a chip after delidding.
I'm referring specifically to people like IDC who replaced the TIM and say a 20C temp drop. To me, dropping your OC from 90 to 70 should leave more room to OC, if you were thermally limited before. Am I wrong in my understanding? It's not as good as TIM, but it's an indicator?
I'm referring specifically to people like IDC who replaced the TIM and say a 20C temp drop. To me, dropping your OC from 90 to 70 should leave more room to OC, if you were thermally limited before. Am I wrong in my understanding? It's not as good as TIM, but it's an indicator?
It was mainly due to the gap. The TIM Intel uses is somewhat equal to NT-H1 if I recall right.
Getting rid of the TIM mainly improves thermals, but it can allow for a bit more OC.
So I again pose my question: hypothetically, would switching from TIM to solder improve the OC headroom on IB or Haswell CPUs? Has anyone closed the gap up and put fresh TIM on to see what result they get? Does it allow them to OC beyond what the limit was prior?
I ask because I've read that IB overclocks were NOT limited thermally...meaning that sitting here complaining about TIM being used is something of a moot point, right?
Have you read Idontcare's excellent 3770K delidding thread?
I think it answers your questions, as well as being educational in other ways.
From an "extra OC'ing headroom" standpoint, popping off the IHS and replacing the Intel stock TIM under the IHS with some NT-H1 and then putting the IHS back onto the CPU results in gaining access to an extra 200 MHz clockspeed without running into the TJmax brick wall, provided you are willing to stomach the elevated power consumption and required voltage to hit 4.9GHz.
Didn;t realize he covered that specifically in his thread (read it months ago,so you'll have to forgive me )
So he's essentially saying that all the extra work he did lowered his temps a considerable amount...without giving him that much extra OC headroom. I just don't see the basis here for the crying people are doing saying that intel is trying to force enthusiasts to the 2011 platform (I mean, isn't the lack of PCIe lanes more of a force to move up to LGA2011?)
Not everyone cries about it, but running a chip at 90°C can be a bit nerve wracking. K series chips are supposedly designed for overclocking, and it didn't seem too much to ask to provide a better thermal solution since they are more expensive. But it's subjective, of course. Make of it what you will.
I've pointed this out before, the expansion coefficients of the various materials in their various configurations and thicknesses must play a part in any decision to set a particular die to IHS distance. Once the IHS is bonded to the PCB, the amount of pressure required for an expanding die and IHS to move apart from each other even a very small distance becomes very great.
At least with direct to die HSF mounting, there is usually some forgiveness in the bracketry and mobo PCB that will allow for expansion.
Not everyone cries about it, but running a chip at 90°C can be a bit nerve wracking. K series chips are supposedly designed for overclocking, and it didn't seem too much to ask to provide a better thermal solution since they are more expensive. But it's subjective, of course. Make of it what you will.
Yep, what he said. Most chips were passively cooled back in the 486 - Pentium days; you can't really do an apples to apples comparison in that respect. You didn't need a huge heatsink fan, all you needed (for a 486 or Pentium I) was something like this:
Obviously cooling solutions are far different now than they were back then. The chips were tailored to the required cooling solution, and chips had far less eloquent coolers then and output far less heat.
Actually, I would completely love it if current chips only required such a small cooler. And still overclocked well. Hahah.
I think you forgot how many thunderbird cores got chipped or crushed. RMA rates would go from sub 0.01% today to what 2-3% if not more?
I bought the Intel warranty for my Ivy Bridge 3770k. I purchased the 3770k from MC for $229 plus tax so the $25 I put into the Warranty seemed like a safe move.Intel offers a overclocking warranty, whereas AMD does not. 90C is perfectly within specs. If you have any concerns though, alleviate them via the OC protection plan and move on.
I can't really relate here though, my chip games at 50c at 1.155v, 58C at 1.25v.. Of course if I dump AVX on it that shoots up 20C, but realistically temps aren't limiting me.
I think you forgot how many thunderbird cores got chipped or crushed. RMA rates would go from sub 0.01% today to what 2-3% if not more?
wouldn't something like this (used in many GPUs) be cheaper than covering the entire thing?