Originally Posted by EarlZ
Originally Posted by Belial
No... that's not true at all. If I'm going for gaming stable I can do firstname.lastname@example.org. If I want streaming stable, which is roughly the same stability required for AVX stress testing, I can only do email@example.com (and it's looking I'll need even more, ie 1.47v).
Gaming doesnt come close to stressing out a CPU. you need way more than .01v from 'gaming stable' to be 99.9% stable, or even just stable enough to do CPU analysis or streaming.
Another thing to consider is that some who are 24hrs stable in Linpak can sill crash with games like BF3, I would like to know why that is.
Well yea, linpack is awful for stability testing, I don't think anyone claims it is. It's decent for temp testing (p95 small fft is same/better), but it's awful for stress testing. You can do 50+ passes of linpack easily and not be close to 24 hours stable, and that's pretty much always been the case with my overclocks on multiple chips.
I said prime95, not linpack, there's a huge difference. But to speak to the phenomenon you speak of, sure, that happens. That's why the best way to test for stability is the most comprehensive one, which is what P95 does (70-80 different tests over ~23 hours). That said, if you can pass 23+ hours of prime95, you can pretty much guarantee it will pass anything else. And it'll come far closer to being actually 100% stable, than anything else.
On top of that, while it's very likely prime95 24+ hours still might crash on some weird occassion in BF3, I don't think I've ever heard of that happening. But just as why it's important to test for 23+ hours of prime95, is the same reason you might crash in BF3 after 24+ prime95. There still might be some certain, untested sector of your system that another program exploits.
That said, if you can pass 24 hours of Linpack I don't think you'll crash at all in BF3. Still possible, of course, and linpack is awful for stability testing, but games don't stress your system nearly as much as a weak stress test. So I'm not sure if you are speaking from experience or what with your comment.
Hey TheHunter, thanks for the benches. Really interesting results, look at what I've got on Haswell and Ivy:
Warning: Spoiler! (Click to show)
Your haswell is basically the same as my 5ghz Ivy, and slower than my faster haswell. It's cool to see such a big difference, that haswell really was a big leap from ivy.
Haswell 4.6ghz 2133 CL9
Superpi32m: 7m 8
Ivy 5ghz 2400 CL8
Haswell 4.8ghz 2800 CL12
Cinebench: 10.56 (on ss it's 10.41 only because i did all those other benches first for the screenshot, but even 10.41 is a big step up).
Superpi 32m: 6m 37
Just really interesting. I've had to drop it to 4.7, 2666 CL11 for better stability (p95 is way more stressful than h264 benchmark and aida!) but really cool to compare results.Edited by Belial - 7/30/13 at 6:32pm