I have an evga 680i A1 p26, 2G of GeIL 6400 ram, and a Tuniq 120 cooling stuff in a p180 case.
Right now Im OC'ed at 3.6gHz.
I have been using TAT and Orthos for testing.
TAT is brutal- temps are too high, ~65c under load, and I dont want to run that. The system crashes (of course) in short order, so I would not say 3.6 is stable with TAT.
But I can run Orthos for a while (havent gone longer than 30min yet, still tweaking). The temps there are ~55c, which I think is pretty good under load at 3.6. So I would say the system will most likely be stable with Orthos.
So my question is- should I test with TAT? Or Orthos? If I use TAT, I think I will have to bump down to 3.4 or maybe 3.45 to get stable (still nothing to cry about!). But if Orthos is the sufficient test, then I can stick with 3.6.
Which app is the standard for overclocking, and real-world use estimation?
If you are Orthos stable many would consider that good. If you are Orthos stable and the temps are good that probably represents as much load as any aps will ever put on your system. It would be nice if you're cooling could handle TAT at your overclock but the real world will never hit those temps. You're case and cooling are about as good as air gets so if you want to pass TAT you either need to lower your vcore or go water.
I consider my PC stable after running TAT for 5 minutes without crashing at first (to see how it handles the heat), then if it passes that, I see if it runs Dual Prime 95 Small FTT's for ~12 hours which is the most I would have my machine under stress when I go to LAN's, and even then that's no where near full load and there's constant breaks.
Really it depends on what your requirements are and how anal you are. If your system doesn't crash and you happy with that, that's fine. If your someone who refuses to touch a PC that won't pass a 48 Hour Orthos session, that's fine to although your probably giving yourself a lot of extra work you don't need!
I don't Fold or anything like that and I turn my PC off at night and while I'm at work unless I'm downloading so I really don't need an absolutely perfectly stable machine. As long as it doesn't crash while I'm already raging at a game it's fine!
I would say that if you are Orthos or Dual-prime stable for more then 8 or so hours then your probably good to go.
But most people dont think too much about full system stability. Meaning that people use a MemTest86 boot cd for 12 hours to test memory. Then use Orthos or prime95 to check for CPU errors and heat. Then use video benchmarks or an ATI tool type program to see if the video is good to go. But never stress everything at once.
But what you really need to do is run ALL of your system components maxed out at the same time. Or as best as you can. I personally prefer to do things like running orthos blend in the background, and looping a 3DMark so as to stress all 3 major areas at once. If you can run like this for over 12 hours then you ARE stable.
I quit using TAT for the simple fact that it gives artifically high temp readings. No matter how many stresses and programs you have running at once, it will never get as hot as TAT gets it. So why bother. I use Orthos because it gets my computer only a couple degrees hotter then anything I will ever do to the computer. Orthos is a more realistic "max usage" standard. Why compensate for the .04vdroop TAT gives when the computer will never need more then the .02vdroop orthos shows is needed. Its overkill.