Maximum PC

Looks good but you really need to clean those cables up a bit

Hehe .. I will, once the build is complete.

3.6GHz stable at 49C with a 0.5v volt increase to 1.2v

Now running at 4.0 with another 0.5 voltage increase to a total of 1.25v

I had a couple of blue screens as I booted into 4.0 .. one during the boot process (so I pushed the volts up 0.25v) and the second was shortly after the desktop loaded (so I pushed the voltage up again).

So far, it is running stably at 4.0 (20x200) with temps around 54C ... I'm going to put the side cover back on and let it run for a while to stabilise the temps.

I bet you won't get too much more out of it. It seems like the i5 and i7 max's out around 4.3GHz for most people on full water cooling systems. So you're doing AWESOME! 54°C is your loaded temp at 4.0GHz? What's your ambient temp?

I'm still working to get it fully stable before reporting final temps. Prime95 is going for about 10 minutes before cores start to throw errors ... only one is capable of continuing past that.

Temperature in my apt. is 22C and the motherboard is still reporting 30C. This cooler rocks.

I should note that I am close to finding a stable configuration, it is far from stock and currently temps are running in the high 60s at 4Ghz. Among other tweaks, I've pushed the CPU voltage to 1.4v, which Asus claims is the minimum necessary to run a 200 BCLK.

Update: using the following settings, I was able to hold 4.0Ghz with a 200BCLK for 2 hours using Prime95 ... with a temp that maxed out at 72C. This is really close to my max, so I'm not excited about that .. but I'm smarter than the average bear.

QPI: 6407
CPU Diff Amp +900mV
CPU Voltage: 1.40v
Integrate Memory Controller Voltage: 1.38125V
PLL: 1.9V
PCH: 1.15V

I'm not happy with that temp but it is true that these boards really do need extra juice for 200BCLK. Thankfully, there is an alternative. Although the i5 750 is supposed to be locked with a 20x multiplier, the fact that it has the Turbo feature means that it isn't really locked. Many enthusiast motherboards give access to the 21 and 22 multipliers used by the Turbo. Mine allows access to the 21. So I'm dialing down the BCLK and up the multiplier to see if I can get a better overclock.

I'm now trying to find stable and cooler settings at 4.0GHz using 21x190. Stay tuned ..

Good luck. I think it's rather curious that my i7-920 was hitting the same temps and 4.0GHz overclock with 20x200 when I started backing off it. I wasn't willing to go any higher with the voltage. The memory controllers in these babies are well known already for burning up without extreme cooling above 1.425V.

I've had my main rig taken apart for two days now. I've sanded the H50 base plate to a mirror finish although, I won't do that to this $288 CPU. I can't wait to see the temps I get with the H50 lapped. That thing, while it works wonderfully for what it is and its price, doesn't have a very smooth surface to it.

Yogi bear is awsome


now I need an H50, but I doubt it will help my i7 OC. it is at 55 oC on the stock cooler at 2.95 ghz. I can't push it any higher, because it will restart the system within minutes of turning on -bigadv WU's





Ted

Thanks for the feedback guys, as always.

Tedster .. an H50 will definitely help your 920 .. just ask Quake.

Ok, so I got a much better result with 21x190. Prime95 threw no faults, so I used F@H again for consistent temp reporting ... I got 64C under full load at 4GHz with the higher multiplier and lower BCLK.

I used the following settings:

CPU voltage: 1.325V (down from 1.4V)
Int. Memory Controller voltage: 1.325V (down from 1.38V)
PLL: 1.85V (down from 1.9V)
PCH: 1.1125V (down from 1.15V)

As I post this, I'm running Prime95 again to test a new config at 21x195 (4.1 GHz)

Brother, for the price of a high end aircooler, you could have this baby rockin your box man. Seriously, in 18 years of playing with computers and many full water loops during those years, this all in one H50 has probably been the biggest, most pleasant surprise I've ever had in overclocking. I've even used other all in one kits and none of them have been this easy to install and maintain and definitely didn't give this level of performance. They have a winner, that's for sure!

Now I have no idea if an H50 will allow you to run F@H above 2.95GHz. But I can tell you this much. You can EASILY hit 3.6GHz, which is my everyday settings, with the H50. And my rig was running F@H great until I developed a different problem. But I'll have my i7-920 back Folding @ 3.6GHz before the weekend is over.

Not sure how true this is but I've heard in a few places that the intels prefer odd multipliers to even when ocing. And lapping the h50 dropped my temps a few c and I only went to 600 grit then polished with ic diamond. A second sflex would probably improve it a lot too and try to get fan shrouds as well (or 2 fans on each side).

That makes no sense whatsoever.



Fan shrouds wouldn't help in the slightest. Think about how they work.

In case you hadn't noticed, I'm compiling a work log so that others can see the difference between different types of water cooling and the stock HSF. The more mods I throw into the mix, the less applicable my results will be to the average person.

Stable settings for 4.1 GHz at 66C:

CPU Diff Amp 1000mV
CPU voltage: 1.375v
Int Mem. Cont. voltage: 1.35v
CPU PLL: 1.875v
PCH: 1.375v

Going to start pushing up the BCLK 2 pts. at a time until my thermal max is reached

ok well when you are done with your work log you can try some other stuff but like I said people have been having more success with odd multipliers as in 19 v 20 idk why but it has been happening with i7 920s. And when I said fan shroud I meant fan >shroud>rad>shroud>fan in theory it reduces the dead zone behind the fan motor and I have seen someone do a really thorough test with this on extreme overclockers I think and it was a few c lower.

I had to go back and edit my results. I noticed something that I should have verified earlier: F@H and Prime95 return very different results for temperature. Sorry guys. I should have known that Prime95 pushes the CPU faster and hotter.

Anyway, adjusted temps below:
4.0GHz (200x20): 64C
4.0 GHz (190x21): 60C

Temperature difference is caused by the difference in voltage required to run stably at the two different BCLK frequencies.

At the moment, I'm running at 4.2GHz and 66C ... going to push until the F@H temp hits the mid 70s or Prime95 refuses to run stably.

Well, I can't get it to post with a BCLK above 201 ... my volts are already as high as I care to push them. So it seems that 21x200 for 4.2GHz is the max I'm getting with this cooler.

Interestingly, my temps stayed below 70C ... so it doesn't seem that the cooler is the limiting factor in this OC. The question now becomes: will a full custom water loop provide a better OC? It should provide lower temps and less noise (that Scythe fan is actually kinda loud) .. stay tuned!

I'll run FAH for a couple of hours to record a temp consistent with previous testing. In the meantime, my final settings used for 4.2GHz, stable, running 69C with Prime95:

CPU Freq: 4.2GHz
BCLK: 200
Multiplier: 21
QPI: 6407
CPU Differential Amp: 1000mV
CPU voltage: 1.40V
Int. Memory Controller voltage: 1.30V (I was able to pull this way back)
PLL: 1.9V
PCH: 1.375V

I was ocing the 860 at work today with just the hyper 212+ on it and it wasn't liking 4ghz just kept freezing. I did notice that if I had a fan aimed at the chipset it would go through more LinX runs before crashing; so I'm thinking the mobo is getting too hot it needs some kind of cooling. I'm using the asus p7p55d, same as yours? And I highly recommend LinX over prime95 it pushes it harder in less time and gets the temps 3-4c or more hotter than prime. You can also adjust the ram amount you want to use. http://www.xtremesystems.org/forums/sho ... p?t=201670 link

You can adjust the RAM used by Prime95 as well. I prefer it because, while it may not generate as much heat, it provides a great deal of info regarding errors ... using the logs, I can figure out which parts of the PC are causing the problems.

My chipset isn't getting hotter than 31C .. which is nowhere near problematic. I have two fans drawing air quickly from that area of the case, so chipset heating isn't the issue I'm having.

Are you pushing extra volts through the chipset? If not, you're not going to get over about 160 BCLK stable.

You have inspired me to buy that water cooler and go for 4GHz myself, Jip. Awesome job man! Keep pushing it!

yeah the readings for my mobo temps seem fine but the heatsink near the pci-e slots feels quite warm to the touch. Which voltage setting is the one for the chipset?

Nice! Glad to hear that, Maggard .. good luck and let us know your results.