only when we try to get new contracts and this one happens to have a lot of specialized hand soldering required that we are shooting for and being one of only a handful of folks that can do that kind of work...i could pretty much work 24 hours a day last week. so no, not really regular, but not un-common in spurts.Originally Posted by rob2k
well, stability testing is a touchy subject but one thing to keep in mind is that a system is only as stable as you have tested it. if you're just testing with quad 32M SPi (dual 32M in C2D cases like me), then it is just that...only quad 32M SPi stable. if it is Prime Blend stable for a few hours, then it is just that...Prime Blend stable for a few hours. each program can and will stress different components at different times so for full stability, you need a combination of everything. multiple instances of SPi, Prime (I like Blend for its work on memory), 3D (games and benchmarks), HCI Memtest, OCCT, etc...they can all find a weakness in an overclock.
as an example for full stability testing in my reviews which are targeted at mainstream users i tend to go with 2 hours of Prime Blend, 2 hours of OCCT, dual 32M SPi using Hyper Pi, 30 mins of 3DMark 01 looped, 3 loops of 3DMark Vantage, and 2 hours of HCI Memtest for FSB and overall stability testing. for memory clocks, i do the same but drop the OCCT as it never finds memory issues in my experience. it is a compromise of time which is always a battle for a review, and actual stability...but a good one i have found.
for most anything i post in forums, dual 32M SPi is a good compromise of time and stability as it only takes 10~15 minutes to run and displays a 'usable' level of stability. dual 32M stable is usually 10~20MHz off the full stability testing above for memory, and most of the times fully 3D/Prime Blend stable for FSB clocking, or at least in the same ball park in my experiences. keep in mind, everyone has different levels of stability needs...sometimes you want to see 'decent' or 'full' stability testing which i would classify as dual 32M SPi or the whole slew of tests i outlined. some users here just want to see a single pass of 3DMark 01. you have a lot of guys here who can't wrap their head around the fact that a lot of us, when benching, just want to see bench stable...whether it be 1M SPi or 3Dmark 01, it is all some of us need in certain situations AKA benching.
as for temp/voltage monitoring, Everest and PC Probe II should be polling from the same sensors that show in the BIOS so aside from a bit of lag, should also be the same from the LCD Poster as i would imagine they are polling directly from the BIOS. i have shown all my screenshots with PC Probe II and the voltages/temps it displays are all bang on with the BIOS (aside from CPU when it goes wonky but that is what Real Temp is for). voltages are also pretty darn close...a couple hundredths off here or there from a calibrated DMM for vCORE, vDIMM, vMCH from the chokes and the DIMM slots. so needless to say, close enough to know what i set in the BIOS. as for Core Temp vs Real Temp...they use different tJunctions so they are 10C different, with Core Temp reading 10C higher. believe which ever one you want, to me, with my ambients...Real Temp seems more realistic under load
man, that turned out much longer than planned, but at least now you guys know how i feel about stability testing and monitoring software
Reply With Quote
Bookmarks