Why torture test

The short answer would be for that .01% that you might hit. Some people let memtest only run one or two passes. That might not be enough to stress the memory enough to see what heat might do to it. If you game for more then a couple hours, that might be enough to start the memory failing.

I'm with you for some of this. 24hrs stable is certainly stable, but 8-10 is probably good enough. I'd only try for longer then that if you normally game for longer.

I game for 20 minutes every now and then, occasionally for an hour.

Any test that will load the PC but stop in the event of an error is good, I guess I misunderstood a few guys about torture testing.

Another reason to Torture test a CPU is just after you apply the thermal paste, you'll want to stress the CPU so the heat will help settle the paste quicker, just a thought. I do it every time i install a new CPU, it helps settle the paste while testing the OC settings.

If I may add a couple of thoughts to this...

Many of these tests really arent torture tests - you are still running the system within its design parameters and usually well under the limits. Its just that most of our normal computer uses do not run the components anywhere near their capabilites. It becomes a torture test only if your system cannot handle the heat or deliver the voltages - but then again the various components will slow down if some measurement gets too far out of spec.

It is a good idea to see if your system is stable at elevated voltages and temperatures. But people should run the same tests before and after changes to understand causes.

Another factor people use these tests for is bragging rights as you have said. The problem is since we cannot use most of the power sitting there and we cannot experience it, these tests are one way to document how we did and show off the performance.

When I run Prime95 on my 4.3GHz I7 975 I stabilize at about 85 C. The limit on that part is 100 C. So I can dissipate the heat and for me that was good enough. It may fail after 20 hours but I am not looking to explore its limits - I just want a healthy reasonably stable overclock - and I have it. When I game I have yet to cross 70C.

Very well written, thank you.

Its easy to say that normal use won't run our rigs at the 100% level. Recently I left coretemp running while I was working on 3dsmax (as well as running several other apps and games). While no game run my machine anywhere near to 100%, creating a high quality render in 3dsmax did run all cores to 100%. It certainly only ran this way for a few minutes, but I felt better having tested at 100% load.

Bob

I run the test sometimes, but if my computer does what i want it to do
at the overclock i set it at with out crashing....Im good.

I am one of those who stress tests for 24 hours. I used to think that 12 hours was a long enough period until one time the system under test failed at the 14 hour point. Made adjustments, reran P95, and it failed at 18 hours.

Since then, I have standardized on a 24 hour test period for all my systems.

When I first set up a computer or I change any significant component, I run Prime95 about 4 hours to make sure nothing is wrong. As I run my computers 24/7, I don't try to overclock them to the fastest that they will go, but rather a setting that is stable day in and day out. If an instability does set in, then I will run Prime95 longer to search for the problem. But like sirheck, as long as the computer works and does what I want it to, I don't get too excited.

Other than testing stability of your Overlock, there is another point...

You want to burn in a PC when you first get it, if it's going to break down, it's going to do it RIGHT THERE! So, if you want to know if your computer is gonna last, torture test it for 2 days right out the box!  

Burning a processor & ram once your happy with the settings will also give it a little added performance (provided you let it cool back down eh.), though this is not technically proven, it's a good idea to stress test your PC at least once ever 6 months, after cleaning of course, this in turn allows you to see if there are any problems with it, and it may even tell you if you have malicious hijacking software maliciously working in the back ground to kill your PC.   (not!)

I could go on and on, but to make it short, Torture Testing has a purpose, quite a few are valid too, especially the paste heating up (written above), this is a great way to set the paste fast!

Bad RAM settings will always stop your Torture test fast! But as the post directly above states, it sure has a way of helping you see there is an internal problem (somewhere!)

Hope this was a good answer..