The Official OC (OverClocking) Thread!

@Sam

pretty sure your right with that answer

@Russ

(it's Done)

got past your 3.8 my friend backing it down now to 3.5 for sure hailed it's own on 3.6 with the 800 ram Corsair with my G.Skill it made 3.9 it's not stable bump the volts up 1.537.. there's a vDroop 4 sure as you can see... it sure did hold at 3.6 at 1.468v stable and maybe if i lower the volts it will to...i'll try 3.7 with this memory in it at 1.468v cause..i tryed it with 800 in it and it did'nt work..i'll yrt with this G.Skill.... Just playing around here

 

Sam,
I'm not so sure about that. Not when it comes to maximum CPU stable speeds. All the records I've seen for air only show a single 512MB DIMM! You're not looking for performance, you are trying to achieve the best stable clock speed your CPU can run, with less voltage!!

Best Regards,
Russ

 

Will,
Try just a single stick of 1066 memory then. You should be able to get it to about 4.0GHz at under 1.5v!

Russ

 

For the record, I tried it. Even at just 3.4Ghz my PC got stock in a reboot loop, I had to put the other stick back to get it to POST.

 

Likewise with my AMD system. It will get stuck in a reboot loop or get bluescreens with a memory buffer overflow, even with apps that don't even touch 2GB. With 2 sticks in dual channel it runs perfectly 24/7. I dunno if it's an intolerance of the system or the dual channel kit of RAM.

 

Estuansis,
Even funnier when you consider that you used to have to boot up the computer with one stick until you set the voltage with certain memory. That supposedly would stop the re-boot cycle on a P5N-E and others. Now you put in two sticks and it breaks the cycle. It's almost comical! The exact opposite of what was done commonly 3 1/2 years ago!

Russ

 

nforce boards don't have reboot loops.... :S
To be fair, If I ever wreck my CMOS settings, rather than go through the tedious CMOS reset process, I just take a stick of RAM out, that resets everything for me, then I shut down and put it back. Difference is, I don't overclock when there's only 1 stick!

 

cmos reset button/pins on the I/O sheild ftw

 

"I pressed this by accident and bricked my board" was enough to make me not want one of those. Besides, you have to be a bit thick to destabilise the CMOS on most boards anyway these days.

 

If mine fails or thinks it fails 3 times in a row it resets itself, tells you its reset itself then gives you the option of loading your saved settings or going to the CMOS page. I think I win ;-)

 

You do realise that is a design feature of all intel chipsets?

 

I doubt he realises it .

I ran my Q9550 today at 516x8.5 (finally got the 500+ fsb I was hoping for) and did 3dmark again and you quessed it just around 18500 points. I raise the cpu clocks the points drop, I just don't get it. Whatever at least I broke my old SuperPI record of 10,828 sec with 10,637 .

 

No no I did not mostly beacuse I've never owned an Intel chipset board I've had VIA (poor me) and AMD!

 

**** me thats impressive!

with a P5Q Pro aswell!

whats your 24/7 clocks?

 

4.2GHz 24/7.

 

That's the sort of overclock you should get out of a P45 board. Running the voltages required to get that stable aren't particularly wise though, so I'm not sure how long that PC will last. As for why the CPU score drops when you raise it, you have to consider two things: a part-stable CPU will actually underperform compared to a fully stable one, as it will lose its ability to multitask properly. Load up an OS with an overclocked CPU that isn't quite running a high enough voltage, and random programs will take longer to open than before, and the CPU can't do as many things at once, this plays a part in 3dmark too. The other factor to consider is that 3dmark is a degenerative process. If you run it over and over again with the same hardware and settings the scores will decrease slightly every time you run it. This is a residual effect that 3dmark seems to leave behind in its install. Only a complete windows reinstall will cure this - yet another reason not to take much notice of 3dmark, your scores will be better or worse than someone else's simply because of how many times you've run the test.

 

The CPU score has only been getting higher but but somehow my Radeon doesn't perform as well when i raise the cpu clocks. I don't really even care just wonderin .

 

That's usually how it manifests itself, the performance of programming, e.g. the graphics driver - it's typically chipset related, from your FSB being too high. I do suspect that with a QX CPU that doesn't need the FSB raised it wouldn't be an issue.