Originally posted by: Idontcare
Originally posted by: OfficeLinebacker
I don't consider a system stable unless I can fold on it on air. Is that a reasonable stability for this forum?
I am pretty proud of getting a Pentium D 805 to 3.46GHz all day long, crunching, with casual internet use thrown in. Oh yeah one of the main reasons I'm proud is cos it's at stock Vcore with a sub-$50 ECS Mobo and the cheapest A-Data memory from Newegg. Is that pretty good, or do I need to scurry back into my e-cave?
Well there's really two forms of stability to be interested in, and to know when to distinguish between. (this includes XS)
There's (1) stable for my personal usage patterns and applications, and then there's (2) stable for some pre-defined usage pattern and software application which is expected to tax the system more-so than (1) does and is widely accepted as such.
What you have described is (1). For your software application of relevance and your specific usage pattern of that application your system experiences stability which you define to be acceptable to yourself. This is perfectly normal and acceptable.
However for the purpose of comparing your overclock results (my E2160 OC's to 3.6GHz, is this good or is it great!?) to those overclocks that others have achieved it is irrelevant as no one else is likely to be operating their computers exactly as you are.
That's where (2) comes in. We generally agree that Prime95 for small FFT for 8-12hrs is necessary to determine if the CPU is stable at a given clockspeed.
This is chosen because (1) Prime95 is freely available, everyone has access to the same application, (2) it is likely to stress your system more so than your specific application of interest (ergo if the computer is stable in Prime95 then it ought to be stable in your application), and (3) there is a rather large database of experience within the community on how to tweak your BIOS to make your system stable by way of using Prime95 as a debug tool when your overclock is not Prime95 stable.
Notice that XS falls into the (1) category just as you do. For XS their applications of interest are superPi 1M, and their usage pattern of interest is usually 1-3 runs between reboots.
My applications of interest are Metatrader 4 and Gaussian 03. Practically no one here is using these applications, so trying to debug my bad overclocks based on stability with these applications is not useful if I want help from the community. Likewise for boasting rights it matters to no one if I claim I can overclock a E8400 to 4.5GHz and run Metatrader 4.
Who cares if I can do that? It doesn't mean jack squat to anyone else (except those who run Metatrader 4) unless I put it in terms that are understood by the broader audience, which requires me to report that at best I can run an E8400 at 4.1GHz and be "Prime95 stable". That will communicate what needs to be communicated.
So you see it just comes down to whether you personally are interested in (1) or (2). Do you want to compare your overclock to other's overclocks (then you need (2)) or do you want to optimize your overclock to be specific for your software and usage patterns (then you are interested in (1)).