The 24hr Prime Stable OC Thread
It's time to bring back the days of only 24 hours Prime Stable OCs as valid, normalized, easily comparable screens. Yep. While I respect what the Xtreme OCers have done to run their Venices and San Diegos up to 3.3-3.4ghz and whatnot using their Xtreme Badass High-End Cooling, I also want to see what their *stable* OCs are, that is, what they would use it on a day-to-day basis. To keep this thread tidy, I'd like for everybody who posts their *24hr Prime Stable screens ONLY!* to also clearly mark their posts with the first 2 lines of CPU markings blah blah, and also clearly state their cooling method. This way it's a lot easier to track who's cooling with what and how high their *24hr Prime Stable* clocks are.
My goal here is *NOT* to discredit the screens of a 1M Super PI run with very high clocks. I happen to think those screens are cool and I wish I had multi-stage Phase Change to run my San Diego to 3.4ghz, because that is very badass. :) Instead, I just want to see how high Venice and San Diego clock for stable 24/7 use. The idea of this experiment is to get a better idea of not what your max OC is, but what your stable OC is. I'm hoping we might get a better look at how high Venice/San Diego are clocking this way, since everyone has their own idea of what 'stable' is and by normalizing 'stable' to a single, well-defined, time tested, tried and true standard, we can get a consistent set of results which are easier to analyze. This will benefit all of us in the long run because we will have a consistent database of results.
To repeat, if you want to post here you will need:
(a) First 2 lines of CPU markings.
(b) Method of cooling.
(c) Max temp reached under Prime load after 24hrs.
(d) Type of Prime test used. I recommend large FFTs for more heat dissipation, plus it stresses the Hypertransport controller more than small FFTs. I do not recommend Blend as there is a lot disk thrashing with that setting and it's not as good a CPU stressor.
(e) A screen with a Prime95 run terminated at or after the 24hr mark.
(f) Optional: Feel free to also post 24hr runs of Memtest86 if this amuses you. You can also tell us about how you looped 3DMark2001 for several hours too if you like. This does not excuse you from running Prime for 24hrs!
You must run Prime95 for at least 24 hours! Don't bother posting here if you don't want to do this! Please don't try to cheat or Photochop, it does the community no good by doing this! This thread is NOT for discussing how much Prime is enough, so don't post any 12 hour or 14 hour screens telling me how that is good enough! Let's keep this thread tidy! I also don't want to hear about Super PI 32M or CPUBurn! That's for another thread! This thread is only for Prime stable 24hr screens! Period! Please try to abide by this guideline! Thank you!
I would like to encourage all of you to try and do this. Yes, 24 hours is a long damn time and it's a pain in the behind. But I think if we all try to do this, in the long run it will benefit all of us because we will have a huge database of standardized results.
Update (1:11am 5/17) - This thread ended up spawning a discussion of cheating in Prime95/SP2004. I think that this is an important thread, because it is very easy to cheat. However I would ask that people please do not cheat, because it benefits nobody in the community! Only honest results will be helpful to truly quantifying how Venice/San Diego is OCing!
Thread: http://www.xtremesystems.org/forums/...ad.php?t=62704
Testing in progress (preliminary results)
Received my Sandy from Monarch last thursday so didnt have alot of time to test it. So this is what I have so far (will update as soon as I have other results)
(a) ADA4000DAA5BN
CABHE 0515SPMW
(b) XP-90C
(c) 45
(d) Custom
http://yamaha.gotdns.org:2525/260x10%20Prime.JPG
