Originally Posted by El Media Vida
Why? I mean, is it worthy to do it 3570K?, i want do OC at 4.5 or more, obviously with water cooling.
I haven't had a single 3570K (I've had about 5 of them) that couldn't hit 4.5GHz as it is... and many of them were on air. So on that note - unless you get one that requires a ton of voltage to hit 4.5GHz you don't have
to delid. On the other hand, more than half of mine were heat limited at that value... and even the best one at 4.7GHz on water (peak temps ~92C). So as far as temps are concerned I can't see any Ivy Bridge CPU not
benefiting from delidding. As to whether that helps with the OC or not is dependent on the chip itself. None of mine would have ever gotten higher than 4.7GHz without requiring more than 1.52v Vcore so I only delidded one of them and the rest I've either sold or put in regular builds at near stock speeds.
However, if you look in the 5GHz thread... there are obviously some very good 3570K's out there - just not any that were ever in my possession.
By the same token, I've never had a 3770K that wouldn't do at least 4.8GHz - including the disaster in my main rig which couldn't do 5GHz even with 1.56v - but runs 4.7GHz @ 1.42v without going over 70C after the delid (on max IBT runs even).
My new one will do 4.7 on air
with only 1.28v - lidded - so I can wait to uncap it and get it under water.
Originally Posted by FtW 420
It is still worthwhile for a 3570k, if you are hitting a temperature wall where you don't want to push the voltage any higher for a higher overclock, delidding can drop the temps & open up a bit more headroom.
3770k vs. 3570k is a personal choice for people, for hardcore gamers who play games that never use more than 4 ores & threads, 3570k being cheaper in price made it an easy choice. Guys who do use multithread will pay a bit more for the extra threads that speed things up.
The 3770k are generally binned higher than 3570k too, for guys looking to get 5Ghz stable or have better chances at running 2666 or 2800mhz memory, it was easier to find a 3770k that could do it. Luck is always part of it, but there are just more high clocking 3770k out there than 3570k.
This completely. In my experience, the same clocks will require up to .15v more on a 3570K SKU than they do on a 3770K - but I've purchased them at varying times so that's really not definitive in any way. One thing I will say is my very latest 3770Ks were all very close in peak temps across the cores - so I'm wondering if Intel didn't tweak their TIM & glue assembly process slightly after reading this thread.
Originally Posted by deepor
It was soldered looking at your own picture. That's metal, isn't it? But I'm only seeing the picture. You're the one who has seen it in person.
Yep that's solder alright. I guess the biggest fear I would have with SB is that has to put more shearing force on the die than the TIM does... I'd want to see a longer timeframe to be sure there were issues with die cracking on that like the two Haswell delids - although I think it's pretty safe to assume those were either loading pressures or temp related. Still breaking solder is a bigger deal.
Originally Posted by skyn3t
the tim on the sandy is crappy too. The Gelid Extreme GC droped almost 30c on that chip it idles at 32c and mx temp is 55c. I don't have it in hands now but IIRC the top temp at 4.7 was 78C to 83c
I will going to search the old temp pic's and post it here
edited my mistake the temp was not that high
here the old temp 4.7 @ 1.368v top temp was 64c
so after delid it i gained -0.032 actual voltage at 4.9 is 1.336v and 200Mhz and droped 9c on top so it is a huge improvement.
How did you clean the solder off the die and IHS? Not like wiping it with alcohol is going to do much (or at least I would think so). Did you heat and wick or ??? Good job on just plowing ahead on that... I assume you figured it would result in a dead chip initially?