[zootielolo]

So I am trying to get my system stable at 4.4 Ghz, and I decided to let it run overnight for 200 runs. Anyways, I woke up this morning to find out it had failed at run 21. I find this funny as it seems many people consider 20 runs stable. I upped the voltage one notch to 1.375 (i think) and started the test again; failed this time at run 27. This has led me to the conclusion that 20 runs of LinX at max memory is NOT stable. How many runs do others here use to test for stability for their 24/7 clock?

 

[error10]

About 2,000 runs, give or take.

I don't consider the system stable until it's run at 100% load for a week.

 

[CrazyNikel]

Quote:

Originally Posted by error10 About 2,000 runs, give or take. I don't consider the system stable until it's run at 100% load for a week.

Jesus, so you wait for a week to use your computer? I just overclock and do 2x30 runs on ITB and then do 4 hours of Prime95 and then I just game till I either experience weird issues or crash.

 

[jackeyjoe]

Quote:

Originally Posted by error10 About 2,000 runs, give or take. I don't consider the system stable until it's run at 100% load for a week.

Most people can't run their computers for a week straight

I just run mine for 20 times and it hasn't crashed on me once but if you aren't comfortable i'd be aiming for 30 runs so it is garanteed stable

 

[error10]

Most of you don't do much serious work with your computers. I'm constantly running something and my computer is on 24x7. (At this moment, 276 processes.) It absolutely HAS to be stable. That's why I only have this moderate 3.6GHz OC.

 

[Tnlgg]

I use IBT with maximum memory & 70 runs or more.

If it can pass 60 runs or more than its ROCK SOLID & it will pass the Prime95 & Orthos tests too.

 

[zootielolo]

Alright, thanks all, I let it run for 70 passes and then quit, could wait no longer

 

[error10]

I hope you weren't just sitting there staring at it while it ran. You can do other things, ya know

 

[TwoCables]

The Read Me file that comes with LinX explains that 50-100 runs is the minimum. It also explains that we should get the most out of it by having it use the most amount of memory possible (or by making the Problem Size as big as possible).

Therefore, my vote is 100 runs.

I also recommend running stress tests while you're sleeping. There's no point in running a stress test when you'd rather be using the computer.

 

[FSF-Foxhound]

Quote:

Originally Posted by zootielolo So I am trying to get my system stable at 4.4 Ghz, and I decided to let it run overnight for 200 runs. Anyways, I woke up this morning to find out it had failed at run 21. I find this funny as it seems many people consider 20 runs stable. I upped the voltage one notch to 1.375 (i think) and started the test again; failed this time at run 27. This has led me to the conclusion that 20 runs of LinX at max memory is NOT stable. How many runs do others here use to test for stability for their 24/7 clock?

You can run 53 trillion runs of linx, but it could fail on prime95. although, assuming you already did 12+ hours of prime95, I'll say a hundred using all of your RAM.

 

[TwoCables]

Quote:

Originally Posted by FSF-Foxhound You can run 53 trillion runs of linx, but it could fail on prime95.

This is a great point. Linpack is not the best stress testing tool. It's just one of 2-3 that you should use. Prime testing is another one that should be used.

 

[Murlocke]

I've failed at 48 and 74 on this overclock before getting it stable.

I usually go for 100.