PC Pro

by **Wolfman__5** » Fri Mar 27, 2009 7:51 pm

Hi

I'm not sure how long is long enough when stress testing your overclock.

I have overclocked my e6600 to 3.4ghz (with a fsb 380, 9x multipler) and my vcore is 1.525v and the northbridge is overvolted to 1.45v. I know the vcore seems a little high but the core temperatures maxed out at 65 C after 30 mins of orthos (the machine had been running about 3 hours with reboots etc) and is air cooled.

I have stress tested the overclock using orthos which has run for 33mins and showed no errors. Unfortuately I CustomPC benchmarks keep aborting.

I just want to make sure my overclock is stable before kicking back and enjoying it!! lol

Thanks for any help/advice.

by **mega_bite** » Fri Mar 27, 2009 8:51 pm

I'd do 1/2 hour while setting up the overclock.
As you tweak each overclock just run it again.

If you've gone as far as you plan with the overclock I'd go for a 3 hour stress test. This based on the fact that I'm not likely to leave my rig on for in excess of 3 or 4 hours.
If I expected to leave it on for 24hours a day, I'd stress test it for 24 hours.

by **_fac51_** » Sat Mar 28, 2009 9:04 am

I usually do about an hour per small step and then run Orthos overnight for a full stress test, leaving it 8 hours or so.

by **matt___** » Sat Mar 28, 2009 1:46 pm

Personally I've never found 1/2 hour blend stress tests to be enough when using P95, as it was always the small FFTs that my set up failed on! However I'm sure that other set ups would vary as to where instability creeps in.

I have found that when I got an o/c stable enough to last 4 hours, it would nearly always be stable for 8 hours +

once I decided that I couldn't get this setup stable enough for everyday use above 3.5GHz then I ran it for 24hrs and it was as solid as a rock :-)

(needless to say I didn't try the 24hr stress test at every opportunity!)

by **MattZani** » Sat Mar 28, 2009 8:59 pm

Stress it for aslong as you think you will be pushing it for.

When i OC My to be built System, i'll do maybe an 8 Hour Stress Test, as i dont think i will be playing games for much longer than 8 hours straight. I'll just play a Console game or something, and keep a steady eye on it, this is probably more than most people would test it for, but i want to know its a stable clock.

by **Wolfman__5** » Sun Mar 29, 2009 10:25 pm

Thanks for all the feedback, tested for about 3 hours and all seems ok which is about as long as I'm on the machine.

Well one machine done, best start looking at my sons!

Thanks again

by **adam_bagpuss** » Mon Mar 30, 2009 9:00 am

theres also a bit of a debate about how long to stress.

view A

stress for 12-24hrs proves your system is rock solid and that anything you throw at it will be fine. Problem with this is since when are you EVER gunna MAX you CPU at 100% for 12hrs+ !!!!!!!. This is not realistic and even playin a game for 24hrs its never gunna get anywhere near 100% for an length of time. Problem 2 possibility of damaging your components as you are maxing them out for a prolonged period of time.

View B

Stress for 1-6hrs gives a good indication that your system is stable. problem 1 possibility of missing something, could not be 100% stable but will probly be 95% stable.

I stress for around 2hrs normally. Has my i7 running @ 3.95 for months now with no issues at all. I can sometimes play games for upto 12hrs and have found no need to match my stressing with my playing time.

by **matt___** » Mon Mar 30, 2009 3:22 pm

while I see the point about not wanting to 'overstress' components - many stress tests need at least 2-3 hrs to run through the different types of test that would find a fault that may cause your pc to crash right when you didn't want it to...

personally I wanted to prove my o/c was 100 per cent stable - and it is :-)

by **adam_bagpuss** » Mon Mar 30, 2009 3:32 pm

yeah maybe a mix of different stressing types,

like small FFTs, blend,

3Dmark etc etc.

to test lots of different tasks maybe for an hour or so.

by **tomrade** » Tue Mar 31, 2009 1:42 pm

12hours generally will be good but just in case 24hours tends to mean you re completely safe

by **erratum1** » Sat Apr 04, 2009 3:43 pm

Why were the custom pc benchmarks aborting ??

I find them quite good for showing an unstable system, as they use the cpu, chipset and memory.

When i was playing with overclocking, i crashed in the custom pc benchmarks.

And with my graphics card it's lots of anti-aliasing that makes it lock up when it's overclocked too far.

by **bally199** » Sun Apr 05, 2009 11:52 am

Might as well do it properly.

When I overclocked my Compaq (yes, shop-bought and OCd vis software tools) I ran the 2 SMP folding@home clients, 2 instances of the P95 stress test, 2 instances of Orthos with one doing the Blend test (IIRC) and the other doing the small FFTs (iirc again) test, and with it running Crysis @ maximum @ 1600x1200.

Needless to say, that night was rather noisy and warm.

I think I only got about 60*C though on a 1.6ghz Celery Dual Core OCd to 2.4ghz. It has a cooler off a Q6600 on it. It's huge!

by **fakesnake** » Tue Apr 07, 2009 11:28 am

bally199 wrote:Might as well do it properly.

When I overclocked my Compaq (yes, shop-bought and OCd vis software tools) I ran the 2 SMP folding@home clients, 2 instances of the P95 stress test, 2 instances of Orthos with one doing the Blend test (IIRC) and the other doing the small FFTs (iirc again) test, and with it running Crysis @ maximum @ 1600x1200.

Needless to say, that night was rather noisy and warm.

I think I only got about 60*C though on a 1.6ghz Celery Dual Core OCd to 2.4ghz. It has a cooler off a Q6600 on it. It's huge!

what was the frames per hour like on crysis, and yes I am bored

PC Pro

How long stress testing is enough??

Who is online