Inktfish

not enough hours based on your logic.
Small FFT and 6 hours.

20 LinX runs. all memory

__________________

BradleyW

I use prime95 small FFT. Simple.

__________________

Official Mascot Of The

Windows 7║Fan Club

alex98uk

I hear you guys referring to "passes" in LinX. Now, not to thread hijack, but i'm using the Linpack tool in OCCT. I assume this uses the same libraries from the Intel Burn Test but instead is measured in time.

It advises a minimum of 6 minutes, but what would you guys suggest for good stability?

__________________
Winamp FLAC > Xonar D2 > Optical Zero DAC > Little Dot MKIII > Sony MDR XB700

British Overclockers Club

Inktfish

If you want to be sure.
Around 3 hours? ;p
I just test my PC for folding atm, if it passes folding then that is what it needs to do.
Seeing that folding is the most difficult task for my PC to do, I will have no problems elsewhere (browsing,gaming).

__________________

alex98uk

3 hours of Linpack!?!?!?!?

__________________
Winamp FLAC > Xonar D2 > Optical Zero DAC > Little Dot MKIII > Sony MDR XB700

British Overclockers Club

chadamir

Perhaps the two are related? 500 is a lot of linpack. Anyone with more ram than you is going to be busy for a long time.

__________________
How to get your i7 to 4.0ghz

Lord Xeb's Unoffical Intel Overclocking Guide for Newbies

Laforet

On a Q6600 each pass with 1GB memory takes about 5min

SO depending on your system 6min could be anything from 1 to 2

Seriously, use LinX, it give you more control and transparency over what is happening.

However LINPACK does have its shortingcoming that it only runs for a relatively short span of time, so it might not pick-up problems in case of high load over time. My personal rule of thumb is 20 passes in any LINPACK and on top of that at least 8 hours of Prime95 Blend. It would be very mistaken to say that 5 PINPACK passes=30min of Prime95 because they are very different tests after all

TwoCables

Not my logic: my experience. I say that because I keep wishing it didn't need to be so relatively extreme compared to everyone else's methods.

But, it seems like I see posts every other week by people saying that their overclock is rock solid stable, but yet it still fails something like a game, or, F@H, or some program, or even just light use of Windows components. It seems like I consistently find out that they only made sure their overclock was stable enough for like 9 hours of the Small FFTs test, and maybe 10 linpack runs, and a few hours of the Blend test.

I had a similar experience: back before I made my overclock stable enough for 12 hours of the Small FFTs test, but before I had ever used the linpack test, I had only tested my overclock for about 8 hours of the Small FFTs test and like 3 hours of the Blend, and just a few passes in Memtest. It was stable, but I was still being frustrated by problems that I just decided were due to a bad installation of Windows XP. But when the day arrived that I was finally feeling up to reformatting, I decided to skip that and get my overclock stable enough for 12 hours just to see what all the fuss is about regarding 12-hour Small FFTs stability. After doing that, all those problems were almost eliminated. And some of them were quirky software problems too.

So, over the next few weeks, I noticed that there were still a couple of quirky little problems. So, I made the overclock stable enough for over 6 hours of the Blend test (I think it was at 7 and a half hours before I decided to use my computer again to reboot it for Memtest), and after that I made it to about the 8th pass in Memtest before I impatiently exited Memtest to use my computer. And of course, 1 pass is when Memtest completes all 8 tests. But once I did that, those quirky little problems disappeared for good. Or, so I thought at least.

One day I found myself reading about linpack testing. Within a week, I went from OCCT, to the IBT to LinX. I liked LinX the best, but the point is: this otherwise rock-solid stable overclock couldn't last longer than about 5 runs. But that didn't surprise me as I was still seeing some quirky behavior in Outlook 2002 that was really getting me down. There was a certain inconsistency with a feature in the window for writing a new e-mail. But once I got the overclock to finally exceed 5 runs, it made it to like 10 or something and failed. But once it got to that 37th run and locked up on the 38th run, I figured that was major success, and so I quit testing. Yeah, I ignored the Read Me's suggestion of 50-100. I wanted to see if 37 was good enough since most people say 10-20 is good enough.

Well, it was certainly good enough. Ever since then, my system has just felt more stable than it ever has. It's probably my imagination, but it seemed just a hair snappier too. But my favorite part was that stupid quirk in the new e-mail window went away. I almost cried. lol I didn't think it would ever go away (it was really pissing me off too even though it was such a minor thing).

So, that experience had a very huge impact on me and now I'll never test an overclock any other way.

__________________

alex98uk

I have LinX, what setting do I use then?

__________________
Winamp FLAC > Xonar D2 > Optical Zero DAC > Little Dot MKIII > Sony MDR XB700

British Overclockers Club

TwoCables

Which version of LinX are you using then?
Depending on the amount of memory and whatever other variables, 3 hours can be pretty wimpy.
LinX's read me says that, for quick testing, values lower than 50 are acceptable. But it recommends 50-100 using the maximum available amount of memory.

And do you know that the Blend test is far less effective at stressing the CPU than the Small FFTs test?

__________________