Torture test questions

[SamuraiInBlack]

I know in the system reviews the torture test seems to be the standard in determining system stability and possibly even build/hardware quality. That being Prime, I/O Meter, and 3DMark all at once if I'm not mistaken. And I know some folks run Prime95 to torture test also for these reasons.

Do any of you guys torture test your own rigs running all three of those programs simultaneously? And what would be a good gauge in determining whether or not a system is a turd or not, in terms of hours? Twenty-four hours seems to be the passing point, but what about anything beneath that? Is there a certain cutoff where it's considered stable enough for everyday use, or specific uses?

Also, obviously this is going to ream unholy terror on a system. Would this build up a lot of wear and tear on it to where the hardware might not be usable anymore? Or would that more than likely happen during the torture test itself?

 

[ChingChang]

I PMed Chris Morley a couple months ago asking about how they do the stability testing, but I never got a reply.

I'm not sure how exactly they do it. I don't even know how to run 3D Mark 05 continuousy.

I run Prime95, I/O Meter, and Luna (nvidia demo, stresses GPU). I have ran them all together before but I am not sure how well that works compared to stressing each component seperately.

These types of tests can't really harm the system.. only thing that happens is heat is increased, and power draw is increased too. So if the PSU is crap and it blows up in your face, it might take out a component. Modern systems have thermal shutdown protection, so computer will shut off before CPU gets too hot, and if GPU gets too hot you will see artifacts and possibly a crash or a BSOD.

 

[Stu55]

You shouldn't run all of those benchmarking programs together because you aren't really getting correct results from each program because you are using memory and cpu usage to all the programs.

 

[ChingChang]

You shouldn't run all of those benchmarking programs together because you aren't really getting correct results from each program because you are using memory and cpu usage to all the programs.

sometimes it's good to know that the system is stable when it is under a very heavy load. Running different programs at once stresses several different components, increasing the load.

 

[CpuMan]

For most normal folk, running prime 95, or for duel core systems, 2 instances of prime 95 overnight is sufficient for most rigs. This will test mainly your cpu, system cooling, and memory. To thoroughly test your memory run memtest. To test your gpu, run 3d mark 05. If you get no errors or crashes you could in most cases consider your system to be stable.

 

[4keatimj]

For most normal folk, running prime 95, or for duel core systems, 2 instances of prime 95 overnight is sufficient for most rigs. This will test mainly your cpu, system cooling, and memory. To thoroughly test your memory run memtest. To test your gpu, run 3d mark 05. If you get no errors or crashes you could in most cases consider your system to be stable.

I agree, for most people there is a point that you can say that 'enough is enough', and say that the system is stable. Myself, well I got rid of the random BSODs and crashes (damn MSI motherboard) by upgrading mobo, ran 3DMark05 and 01 for a while, ran some memTest86+, and finally some real world testing (aka games playing) to determine that the system was running sweet. At the end of the day, for most people, the PC should do what they want to it to do, and do it well. Hours of prime95 won't always determine this, the user will. If it doesn't crash or produce problems, then its fine.