Something I've learned recently, each test has a certain "area" that it detects errors in better than the other programs. Like I ran an hour of OCCT, passed it and was happy. I flipped on Prime 15 minutes later, and got a blue screen in about 2 minutes of running it. I'm not saying that 1hr of OCCT is sufficient at all, I'm just saying clearly OCCT didn't stress my CPU in the same manner that Prime did. Its not that one is better, its that they're different, and to be thorough you want to use many of them.
Prime seems to be best overall according to what the community says. Linpack is probably a good one as well but its pointless to use it because all it does is limit your OC for no good reason (due to the insane temps it produces). So if you want to max your OC you just can't use this (well I guess there's always the expensive cooling options... but why would you pay more money for cooling JUST to handle this test?).
With Prime95 you really want to do it 24 hours... many people have said it is not unusual at all for it to catch errors around the 20th hour (I wish I knew why that was the case). Maybe its overkill and the whole thing with 24hrs is just a myth, maybe its not. I don't think there are a whole lot of people who can say they've tested a large sample size of CPUs at various OC settings w/ Prime95, then monitored and used those chips in the future to see how stable they were in normal use.
Originally Posted by Sethy666
For a purpose, its testing your thermal limits. Prime95 will never hit the heat IBT does. In this aspect, it works as advertised.
But why even bother... the "thermal limit" that IBT brings your CPU to is wildly unrealistic. Just my opinion, I understand if some feel the need to do it anyway.Edited by donkrx - 7/28/11 at 10:40pm