Is my overclock stable?

maki

I've just overclocked my Q6000 (G0) to 3Ghz and after a boring day full of stress testing I think that it might be stable but would just like a second opinion.

Overclock details:
Upped FSB from 270Mhz to 333Mhz
DDR2 800 with 1:1 divider -> 667Mhz
CPU Vcore set to 1.27 in BIOS (VID = 1.2625)
Haven't messed with anything else.
Temps are about 32 idle and maxing out at 58 with IBT.

I've successfully ran 1 hour of OCCT, 20 passes on Intel Burn Test, and 7.5 hours Large FFT's on Prime95 with no errors.

Would this be considered stable or should I also be doing small FFT's, blend, a few passes on memtest, or anything else?

Also, since the RAM is technically underclocked, should I even need to test memory stability?

Solitaire

maki wrote: »

Would this be considered stable or should I also be doing small FFT's, blend, a few passes on memtest, or anything else?

Small FFTs for a few hours should seperate the men from the boys

Also, since the RAM is technically underclocked, should I even need to test memory stability?

Yup, just in case its a tad unstable because the MCH/chipset is too stressed with the increased FSB and needs a small voltage hike itself. Shouldn't happen too often below 900MHz even at 333MHz bus speed unless you have all four DIMM slots filled though.

IBT for 50 passes negates the need for small ffts in prime so you only need to run large ffts for about 12 hours and you can be fairly confident of stability. Of course you can never claim 100% stability, if you run prime indefinitely it will error eventually but unless you are folding on the machine the above is safe enough. And defo run memtest as solitaire pointed out as mch stability is often an issue. If you just set vcore you need to make sure that cpu term and cpu pll are within cpu spec. Auto on these often jumps alot if you increment vcore even a little.

maki

Thanks for the advice guys.

I ran memtest86+ and got an error within 2 minutes at stock DRAM voltage of 1.8v. I upped it to 1.9v and got an error within 5 minutes, and got an error at 2.0v after about 20 minutes.
I haven't tried changing the MCH voltage yet; is it safe to do so?

One strange thing though.
In the Gigabyte BIOS there's a Performance Enhance option for RAM. Its default value is Turbo but I had it set to Standard during all tests as it said it would improve overclocking stability. I changed it back to Turbo and reset my DRAM voltage back to 1.8v and interestingly I didn't get an error in memtest until almost 2 hours in.
An error's an error but it seems a bit odd that it lasted significantly longer on that mode.

I've a suspicion that dodgy RAM might be an issue here so I'm going to run memtest overnight with everything back to stock values.

Other than that though what could be the issue? To be honest I never thought I'd have to go messing with RAM voltages on a fairly common overclock. Maybe my mobo just hates me...

Gigabyte EP43-DS3L
2x2GB Corsair CM2X2048-6400C5 @ 5-5-5-18

Solitaire

Check your timings. They might have defaulted to much tighter values if they detect you're running them at 667MHz rather than 800MHz. You might need to take any CAS Latency settings off Auto and set them yourself in this case.

PogMoThoin

When pushing higher fsb You usually need to add a little to the mch to push it through, don't leave it on auto

maki

Ok, after resetting everything back to default values I ran memtest overnight. I ended up with 12 errors over 7 passes.

Looks like broken RAM was the culprit after all. Strange though that in over a year I've never once had a BSOD or any sort of critical errors.

Oh well, guess an RMA is in order :mad: