On Systems Stability Testing
...this is one of those topics which gets a lot of heated discussions going...enough to make me puke (though that may also be the penicillin I'm on). So I will try to enlighten you (
)...I start with a quote from Tom's Hardware re Intel's Extreme Tuning Utility, with Anandtech also having some great quotes on it:"...XTU is likely intended for “in the know” overclocking exhibitionists and members of the press who usually get their hardware long before our round-ups are published. Most end-users are forced to search for it, and the lack of references to the name makes that search even more difficult.
Intel has its own stress-testing program, which many professional-level overclockers use to prove that their systems are stable...."
...not that I know everything, but I head the firm that has been for over a decade now supplying a very large global organization with tens of millions of members with custom software (think cloud computing and such) running exclusively on hardware we also put together. Obviously, the commercial systems are not overclocked, but we apply the same rigor to those as I do to my own systems which run and bench up to 5.3 GHz...sorry for this piece of chest-beating but I think it is necessary in this context.
- IBT ...what you folks call the Intel Burn Test (IBT) is actually not the current IBT at all - the one I see here is an older one modified by 'agentGOD' - that should have been your first clue
- Prime95 isn't getting any younger - I know modern systems that can pass 24 hrs or longer on Prime95 but bomb as soon as they are supposed to run Cinebench 11.5 etc.
- I rather run 10 min of vCore at 1.75, and would purchase such a chip, but would not do so from someone who constantly runs Prime95, IBT or folds all day and all night - that will cause more degradation than short bursts of higher voltage - as much as you really should stay below 1.52v unless you have below-ambient cooling solutions
- we use specialty proprietary commercial software that is designed to bring any commercial system down in less than 2 min to test out...the stuff I see mentioned here as 'tough' and the 'final judge' of your system's stability does not compare.
- asking a CPU IMC to run 12 hrs of Prime95 on 4 or 8 GB is NOT the same as doing so for 16 or 32 GB, think of the extra work the IMC has to do...thus the CPU vCore voltages will be different - so no sense saying, 'my Golden chip' does this or that without any extra info on what is running.
- While I am at it, please stop talking about 'Golden Chips'
- it usually embarrasses you, in public. There are professional OC networks where you can buy such chips...one recent one offered booted up at over 5.5 GHZ w/1.55 vCore (un-deliided) - but the asking price was $1500...that probably qualifies for a Golden Chip
Now then, back to 'XTU'...that includes Intel's current System Stability Test...it comes with a CPU stress test AND a memory stress test which you can run from 1 min to
30 days ! That's what professional overclockers use as much as commercial system builders, not to mention Anandtech , Tom's Hardware etc...when I take an Intel product
such as an I7 3770K, I feel more comfortable with Intel's latest 'official' stress test
embedded in XTU than whatever some folks here seem to insist on with a certain degree of zealousness...usually when someone just posted a faster CPU time than they did.
My recommendation should be clear: Run Intel's XTU stress tests for CPU and memory for 12 to 24 hrs - and have a nice day