Premise: do household voltage fluctuations play more of a roll in CPU's failing tests that involve many hours of endless calculation that in the REAL WORLD would be unlikely to cause a problem? i.e. if I wasn't running the CPU at 100% for 20 hours, a household voltage dip that occurs for a split second would normally have no discernable effect on my PC's stability... perhaps a power line conditioner or similar would be a better fix than upping the base CPU voltage?
I have two nearly identical PCs, one with an e6400 one with an e6700, both overclocked. Both have been run for many hours of single Prime95+CPUBurnin, and have gone through many gaming sessions, DVD authoring, etc., with no ill effects. [Same PSU on each (a 700watt OCZ, mobo P5WDH, memory, and harddrives (vid card wise, one is a 7950GX2, the other a 7900GTX)]
Last night, I ran Orthos(dual prime). On both my C2Ds (e6400 and e6700), it initially failed at the 20 minute mark (on both CPUs... both previously passed single Prime95+CPUBurnIn for 15+ hrs). I upped the voltage by 0.025V on both and ran it again and this time they both failed at the 2 hour mark... (I upped them AGAIN and they are running at home right now).
Now, I'm curious... on both PCs they seemed to fail at nearly the same time (as far as I can tell, I wasn't watching when it happened, but the time elapsed, allowing for different start times, seems to suggest a near simultaneous failure)... could it be a fluctuation in power line voltage that is causing the voltage to the CPU to dip, resulting in an error, and pumping the CPU volts just gives it more 'wiggle' room? (I noticed when my central air kicked in, the lights in the hallway dimmed for a split second). On neither PC did anything lock up or freeze -- the only thing that seems to have happened is a calculation error.
I upped the voltage on both CPUs and am running it again, but I remain curious:
-does it really matter if there is an error once every few HOURS of intense computation? everything else runs just fine and, other than video encoding, nothing I do would ever stress the CPU so much for so long.
Perhaps instead of bumping the CPU power up more and more, I should just get a line conditioner or a battery backup (which supposedly can smooth out the voltage as well, though I'm not sure how quickly it reacts, given the billions of calculations taking place).
Thoughts?
I have two nearly identical PCs, one with an e6400 one with an e6700, both overclocked. Both have been run for many hours of single Prime95+CPUBurnin, and have gone through many gaming sessions, DVD authoring, etc., with no ill effects. [Same PSU on each (a 700watt OCZ, mobo P5WDH, memory, and harddrives (vid card wise, one is a 7950GX2, the other a 7900GTX)]
Last night, I ran Orthos(dual prime). On both my C2Ds (e6400 and e6700), it initially failed at the 20 minute mark (on both CPUs... both previously passed single Prime95+CPUBurnIn for 15+ hrs). I upped the voltage by 0.025V on both and ran it again and this time they both failed at the 2 hour mark... (I upped them AGAIN and they are running at home right now).
Now, I'm curious... on both PCs they seemed to fail at nearly the same time (as far as I can tell, I wasn't watching when it happened, but the time elapsed, allowing for different start times, seems to suggest a near simultaneous failure)... could it be a fluctuation in power line voltage that is causing the voltage to the CPU to dip, resulting in an error, and pumping the CPU volts just gives it more 'wiggle' room? (I noticed when my central air kicked in, the lights in the hallway dimmed for a split second). On neither PC did anything lock up or freeze -- the only thing that seems to have happened is a calculation error.
I upped the voltage on both CPUs and am running it again, but I remain curious:
-does it really matter if there is an error once every few HOURS of intense computation? everything else runs just fine and, other than video encoding, nothing I do would ever stress the CPU so much for so long.
Perhaps instead of bumping the CPU power up more and more, I should just get a line conditioner or a battery backup (which supposedly can smooth out the voltage as well, though I'm not sure how quickly it reacts, given the billions of calculations taking place).
Thoughts?
Facebook
Twitter