Best Stress Testing Program - Page 3 - Overclock.net

TranscendentalProtagonist · 10-01-09

I'd initially stick with the test that fails the most. Once you're stable with that, proof it with the other.

flyin15sec · 10-02-09

Quote:

Originally Posted by Exostenza

TwoCables: wow dood.

I'd suggest Linx if you want to get it over with in a hour or 2 or prime 95 if you are willing to run 8+ hour tests.

I know TwoCables will have a problem with this, but I've seen that if I can do 50 passes in linx with most of my available RAM (about a hour and a half) prime95 can run forever.

My E8400 passed 100 IBT only to fail Prime95 Blend in 15mins. I spent 3 weeks looking for the cause in all the wrong places. It turned out to be low vcore.

Lesson learned, I Use Linpack testing along with Prime95 and Memtest now for all my systems.

TwoCables · 10-02-09

Quote:

Originally Posted by Exostenza

TwoCables: wow dood.

I'd suggest Linx if you want to get it over with in a hour or 2 or prime 95 if you are willing to run 8+ hour tests.

I know TwoCables will have a problem with this, but I've seen that if I can do 50 passes in linx with most of my available RAM (about a hour and a half) prime95 can run forever.

I have a problem with this because I don't want people to be frustrated by problems due to instability.

Just because it was stable enough to pass linpack runs with the maximum amount of memory and then pass Prime95 "forever", it doesn't mean that Prime95 is worthless. It just means that you happened to have tweaked it just right. Either that, or you only ran the Blend test or only the Small FFTs test and didn't check with the other 2 tests out of the 3 prime tests.

Please follow along with me in the quotes below:

Quote:

Originally Posted by ChickenInferno

It's not always true though and I have seen prime95 error out after LinX passes indefinitly. This is why multiple tests are a must.

It's not that LinX stresses harder than prime95 and therefore if you are LinX stable you are prime95 stable; it is simply that they test differently and find different errors (Some errors overlap of course).

Yep. I see it all too often where people are confused because their overclock can do 100 linpack runs, but yet fail one of the 3 tests in Prime95. They always find it hard to believe that this is an indication that they need to increase the stability of their overclock. Most usually blow it off and say "bah. It's stable. If it can pass linpack - which is far more stressful than Prime95 -, the I'm not worried about it". But they don't know the truth that linpack testing isn't superior to prime testing; it's simply different. There's no way around it.

Quote:

Originally Posted by tha d0ctor

if I were to use solely one program i'd use prime but two is always better than one so why not include linx. I wouldn't just use linx though cause i've passed linx and before and failed minutes into prime w/ the same settings

This is a great example of why it's so important to avoid relying on either linpack or prime testing, and utilize both instead.

Quote:

Originally Posted by tha d0ctor

although I do find linx better to determine the max temperatures under load, prime never seems to put me at 99.9 percent utilization

Then try the Small FFTs test in Prime95 instead. This test is a CPU-only stress test and it gets the temps hotter than the Blend test.

Quote:

Originally Posted by flyin15sec

My E8400 passed 100 IBT only to fail Prime95 Blend in 15mins. I spent 3 weeks looking for the cause in all the wrong places. It turned out to be low vcore.

Lesson learned, I Use Linpack testing along with Prime95 and Memtest now for all my systems.

This is another great example. This is why I emphasize the importance of making the overclock stable enough to pass both extreme linpack testing and all 3 tests in Prime95 for nearly a dozen hours on average. It may seem like it's not worth it, and it may seem to be very extreme, but for those of you who refuse to make your overclock that stable, then all I can say is this: that's your choice. But don't be confused if you run into problems later on down the road that seem to be impossible to fix. When that happens, just increase the stability of the overclock. But unfortunately, since you let it be that unstable for such a long period of time, you have to work a little harder now at making it stable because it degraded.

Clockadile Dundee · 10-02-09

It seems to be a common misconception: people think any of these programs are alleged to find any and all errors. No one has ever claimed that one program can find all errors; therefore, failing one and passing the other, does not mean one of them is faulty.

sintricate · 10-02-09

Quote:

Originally Posted by XFreeRollerX

Stable pass of linX= 20 minutes

Stable pass of Prime95= 8 to 12 hours

You still want to try P95?

I usually do 1 hour or so... I set to MAX MEM and 20 passes, takes about 1 hour, that will prove ultimate stability. If you can pass that, your PC won't crash for sure lol.

I've had OCs that passed either one of those and sometimes not both. You need to do more than one type of test.

TwoCables · 10-02-09

Quote:

Originally Posted by Clockadile Dundee

It seems to be a common misconception: people think any of these programs are alleged to find any and all errors. No one has ever claimed that one program can find all errors; therefore, failing one and passing the other, does not mean one of them is faulty.

This is exactly true. When somebody's overclock can pass one or two tests but fail another, then that means they need to increase the stability of their overclock so that it can pass that failed test!

Quote:

Originally Posted by sintricate

I've had OCs that passed either one of those and sometimes not both. You need to do more than one type of test.

Exactly. The more stable the overclock is, the safer it is. Not only is it safer, but it decreases the chances that problems will occur later due to instability.

Jtwizzle · 10-02-09

May I also suggest Folding@home(SMP client). =]

Prime/Orthos + IBT/LinX

Clockadile Dundee · 10-02-09

Quote:

Originally Posted by TwoCables

This is exactly true. When somebody's overclock can pass one or two tests but fail another, then that means they need to increase the stability of their overclock so that it can pass that failed test!

Exactly. The more stable the overclock is, the safer it is. Not only is it safer, but it decreases the chances that problems will occur later due to instability.

Agreed. Failing one test, and passing the other, if anything, means you should be running more tests, not less.