It was a I have nothing better to do unplanned moment, I replaced the IHS, no glue used or scientific temp comparisons, XTU always shows about 10 degree higher temp momentarily than AIDA, dropped the temp 5 to 10 degree's for same settings in that benchmark. Devils Canyon is just around the corner and I due hope we are all not disappointed, I imagine there will still be golden chips, and maybe there will be a greater chance of getting a decent clocker, but not holding my breath

I've delidded 3 4770K. 2 out of 3 ain't bad. One worked mostly, but the IMC went away. Only one channel worked and then only at SPD timing. The other two were great until I did too much to them.

I think of several steps of cooling as a step up in power I can use to overclock. Since the temptation is to run benchmarks very hot for a short time, I use a temperature in the low 90s for reference: (stress test is OCCT small data set for even, steady distribution of power over the four cores)

not delidded, MX-4 TIM, Hyper 212 Evo air cooler: 117 watts at 90.2C
not delidded, MX-4 TIM, custom water cooling: 137 watts at 90.9C -- 20 more watts to play with
not delidded, Liquid Metal TIM, custom water cooling: 144 watts at 92C -- not worth the expense or risk
delidded, Liquid metal TIM under and over the IHS, custom water cooling: 186 watts at 84C (stopped OC at 1.56 Vcore)

Guessing that power could go over 200 watts within the 100C throttling limit, delidding gained another 60 or so watts I could spend on overclocking -- a big gain that makes delidding and electrically conductive TIM risks worth it.

Jeff

Guessing that power could go over 200 watts within the 100C throttling limit, delidding gained another 60 or so watts I could spend on overclocking -- a big gain that makes delidding and electrically conductive TIM risks worth it.

I too used Coolaboratory Liquid Pro over and under after delidding and lapping with pleasantly surprising results. Still, I've been afraid to push my 4770k since reading that it was a bad idea to go over 1.4 Vcore regardless of temperature because it would still seriously degrade the life of the CPU. Is this understanding faulty? Temperature wise I have plenty of headroom. I can easily stay under 50C with 1.34v @ 4.6 Ghz and can post and run general internet activities with 1.4v @ 4.7Ghz, but can't pass RB ( and temps sneak closer to 60C under heavy load before BSOD). Is temperature the primary limiting factor and I'm just being a big chicken or am I rightly concerned to keep Vcore limited to 1.4v? Is there an authoritative and definitive answer on this or is it mostly conjecture and speculation? On the other hand, if I do push a little too hard and things just happen to go south on me, it could provide a good excuse to try out a Haswell refresh that's compatible with the 8-series chipset (I haven't looked far enough into this.)

BTW, I have to say that the method I used to delid was calming and worked like a charm. I positioned my razorblade in a vice with just enough of it exposed ( about 3/8" or 25mm) so that no matter if I slipped or pushed too hard, it would be stopped by the jaws of the vice before the sharp edge of the blade reached the brains, and there were no worries except for angle. If I ever do it again, I will do it exactly the same way. Easy-peasy.

I am no expert on this issue but my best guess is yes it will degrade and possibly kill your CPU, slowly increase voltage until you reach stability, if it's to high for your comfort than stop
From the reviews I have seen the refresh model won't net you any gain over the K processor except the 100mhz bump at default, it's all about Devils Canyon

I am no expert on this issue but my best guess is yes it will degrade and possibly kill your CPU, slowly increase voltage until you reach stability, if it's to high for your comfort than stop

Yeah, that's sort of what I did. Since I already had 4.6 dialed in on manual, I just bumped up the voltage to 1.4 and the clock to 47. It only became unstable with a 100% load when RB went into H.264 encoding. I gave up after that even though my temps were fine. It left me feeling...a little unsatisfied, but I'm guessing once bitten you will always be left feeling a little unsatisfied and wanting to wrench out just a little bit more no matter what. So I have that...

Voltages over 1.4 aren't the stuff of 24/7 stability, but . . . I pushed my delidded dog with 1.58Vcore, but only during benchmarks. That got me to 4750MHz +/-. (Now looking for a better one to get higher clocks.)

The CPU didn't show any degradation, but benchmarks like Realbench, XTU, Cinemark, etc. only run for a few minutes. The rest of the time, Vcore was 1.33 for 4.4GHz.

Jeff