[GravityKnight]

My chip sucks... it has always taken a ton of voltage to get a good OC.....

System Specs
3570k
gigabyte z77x-ud3h
gigabyte gtx770 oc'd
8gb trident x 2400mhz 9 11 11 31 1t 1.68v
2 128gb samsung 840 pros
corsair vengeance case
corsair hx750 psu
etc. etc.

Previously I had a double fan h60 running on my cpu delided w GC extreme TIM. Think I over did it on TIM, but either way it was pretty hot. Settled on 4.6ghz at 1.39v
and temps would get up around 80c on prime95 after a long run, and around 90c on burn test after some time. But it was stable there... at 4.7, even with more voltage it might pass prime and burn test (until I would shut them down due to heat) but ever once and a while I would get an error in WoW or something other kind of instability issue.

I decided to go bare die and see what would happen. I bought an h100, it's running in push pull, my case has 12 120mm fans and inside temps are about 68F right now. I have access to a machine shop, so I put the pegs in a lathe and took off approx (got it within about 3 thousandths) of what the IHS is (mine measured out to 2.58mm). Put everything together (just left the stock TIM on the h100i) and I am pretty impressed with the results.

For comparison sake

Intel burn test before at 1.42v would peak 91c at about test 8 or 9 and I would shut it down.
Now I ran burn test at 1.45v and it peaked at 88c on the hottest core after 15 tests run.
Both these at 4.7ghz.. I am of course going to run it longer, but I am easing into things....

Prime (max heat) before at 1.42v after 30min of running would peak at 80-81c.
I just ran Prime at 1.452v for 30+ min and saw a max of 74c and it has been holding under that without moving for the last 20 min. Of course I will run it for hours after I feel comfortable walking away from it and see what happens.

These are just prelim numbers.. but it's obvious there is a large drop in temps (seeing a larger variance across cores now which is odd, didn't expect that.. maybe 12-14c).... maybe I can sneak up to 1.46 or 1.47 and 4.8ghz before it's over, but I don't know this chip hasn't been very OC friendly.

Attaching a couple pictures of the process.... this was a fun project!

 

[GravityKnight]

Update: I was able to get 4.8GHz at what appears to be stable.. but the amount of voltage required... barf... 1.50v under load.

Interestingly enough, World of Warcraft is a pretty good stability test. I was able to get 4.8 stable under prime95 for a couple hours of running with less vcore, and intel burn test as well... but WoW would crash over and over. Kept tweaking settings with no luck until I raised vcore up a little more. It appears to be stable now... but we'll see.

Temp wise prime 95 27.7 peaked at 88c (though it's holding about 85c on the hottest core) Not bad considering the insane voltage. Wish my chip was better so I could push higher clocks with this voltage, but I'm too afraid to go above that ~ from what I've read 1.5 is really the max I should go, even though temps are ok.

 

[GravityKnight]

Another update:

I pulled everything apart. I was nervous, as I used electrical tape to hold down the cpu and project from it shorting out anything against the water block (read this can happen with what's exposed, not sure but didn't want to take chances).

I was wondering if the tape had melted. It had not.... but I decided to cut out a paper square to go around the die to protect it as paper doesn't ignite from heat until 451F. While in there, I had to get rid of the h100i stock TIM and replaced it with my personal favorite (up to this point) GC Extreme.

Reseated, fired it back up... I'm 1.5hrs into Prime95 27.7 on small at 1.5v core (4.8ghz) with a max of 79C and holding around 77-78.... I can't believe how cool this thing runs with this kind of voltage. With a better chip I would imagine this could be a 5Ghz setup no problem... makes me want to grab a 3770k sometime....