Prime95 error vs. everyday use

[MaDMagik]

Hi guys, sorry for asking the question that probably was asked many times before, but how important the prime errors actualy are? I mean, my e5200 is running at 3.6Ghz, everything is smooth but if i try prime testing(regardless of which test option i chose) im getting an error message on the first of the two test threads(fatal error, rounding was 0.5 expected less than 0.4) while the other one carry on forever with no errors. Now, my question is: how important this error actualy is? I mean, i can play crysis, cod4 or do whatever else for hours and hours and nothing happens, but do I risk anything in long term, like cpu damage or something? The stress.txt file that prime tells me to refer to doesnt exist on my machine.

 

[richardscott]

you dont risk cpu damage or any damage at all even its just not good for windows cos it will write wrong information to hd and then it will eventually get corrupted and then you have to reinstall windows. try upping ure voltage to get it stable, i find a few hours of p95 is fine.

 

[Priv]

I guess one or 2 notches of vcore are enough yes. If one core can run the test and the other cant you dont need much vcore anymore. Atleast, from personal experiencies that is.

 

[bobbknight]

Or you can drop the OC.

 

[MaDMagik]

Ok, I`ve been fiddlin with vcore for long time now but even going up to 1.35v doesnt fix the problem and I really dont want to go higher than that considering Im using stock cooler, so I just keep it at still stable 1.285 I dont want to lower the OC either. I guess I will carry on with current settings and see what will happen if Im not risking quick cpu damage. If windows goes tits up than I guess I can handle it

 

[richardscott]

okay its up 2 u your actually doing less damage at 1.28v than at 1.35v im currently running my e8400 at 4.6ghz with 1.6v.

 

[Priv]

Did you try other voltages, like north bridge, FSB, and RAM?

 

[orangegator]

okay its up 2 u your actually doing less damage at 1.28v than at 1.35v im currently running my e8400 at 4.6ghz with 1.6v.

1.6V ! You realize that will kill the chip pretty quickly. It'll probably die in a few months to a year. The safe 24/7 voltage is 1.4V, regardless of temperature. People typically will run it up to 1.6V for benchmarking for very short amounts of time, but then drop it back down.

 

[RetiredChief]

As pointed out the normal problem is an occational corruption of Data.

However; There is a smaller possibility (still there) that you could wipe out your HDD, or corrupt your BIOS. This happens when the corrupted Bit is a pointer. IF that pointer is changed from say a Location to write data to your hard drive to Lets say partition table, or worst your BIOS. Then that is where the Data gets written. Although you loss everything on your HDD (I KNOW YOU HAVE A BU) you can recover. IF BIOS its 50/50 - If clearing will recover - Great, other wise its RMA time.

 

[blackwidow_rsa]

I had a few errors with prime95, played games with it no probs but the pc felt faster once i got rid of the error since it doesnt have to error check, recalculate blah blah etc

 

[one-shot]

Always make sure it is stable before you leave it at that speed. I ran my cpu for 8 hours each time i raised to Core frequency. 8hrs is enough for me. some people may do it longer but I have a Xigmatek CPU cooler and keep my e6750@3.7ghz pretty cool.

 

[MaDMagik]

So I guess my safest option is to drop the speed to a no-error level, get some proper cooler and try again with higher vcore. I dont normally touch other voltages like NB or fsb, but might as well give it a try, Ram is running on 2.1 anyway

 

[iluvgillgill]

have you lower the memory multiplier?that could be the problem being memory running too fast.

 

[MaDMagik]

I kinda excluded memory from causing the problem as it only happens when cpu reaches 3.6ghz or over, regardless of ram speed i.e: cpu 3.6ghz ram 800mhz(spec) -problem exists, cpu 3.2 ram 960 - problem doesnt exist. But we`re drifting away a little bit here, Im not really looking for cause or solution as its not THAT important, i just wanted to know whether or not i risk hardware damage

 

[iluvgillgill]

there wont be any hardware damage. it would probably one day you turn your computer on and found out that the BIOS had reset itself before its not stable.

thats all.

 

[richardscott]

1.6V ! You realize that will kill the chip pretty quickly. It'll probably die in a few months to a year. The safe 24/7 voltage is 1.4V, regardless of temperature. People typically will run it up to 1.6V for benchmarking for very short amounts of time, but then drop it back down.

its at that voltage to see how long it lasts i rekon it will last well before i swap it out at xmas tho feel sorry for who buys it on ebay lol.

 

[richardscott]

voltage kills your cpu in 2 main ways, it can increase the chance of quantum tunneling where the electrons will force there way through the wrong part of the chip resulting in death, or it can the electrons can actually move the atoms which make up the transistors resulting in a larger charge loss which will also result in cpu death but is slower then the tunneling effect. to reduce both of these effects you have to keep he chip cool the cooler the longer it will last, for every 10c lower in temps your double the lifespan of the chip. if you swap hardware every couple of years then ure fine overvolting the chip.

 

[V3NOM]

well i don't know what multi and fsb you're using but 11x multi is much more stable oc than 12.5x multi on e5200.

imo, if it fails prime or orthos it's unstable. while it might not be unstable in other applications, it's still not good for the cpu imho.

 

[MaDMagik]

Im using 12x300 atm but I will give 11x a try

 

[V3NOM]

yeah 333x 11 = 3.6ghz should be happy with that if it works! make sure you set your ram synch or manually @ 1:1 though.

 

[V3NOM]

if you even bought it you already shortened its lifespan. intel overvolts from the factory.

 

[V3NOM]

true fact dont deny it

 

[MaDMagik]

Well as for me i dont really care if it goes up in smoke in some time, i just dont want it to happen before xmas, that was my only concern But 11x didnt help, still i get exactly same error message on the first thread, BUT while it was something like: Rounding was 0.5 expected less than 0.4 now its Rounding 0.4785748574 expected less than 0.4 So maybe if i bump the vcore to over 1.4 it will eventually go below 0.4 But i decided to drop the oc to fully stable 3.4ghz until i get a better cooler. Any sugestions on that btw? Will anything of around 20 quid be better than the stock cooler?

 

[V3NOM]

10x multi?

 

[antiacid]

quick info: higher voltages might end up "killing" your chip but the chances are that it really won't matter at all for the duration of ownership of the cpu.
To go back to your original problem, I would definitely not run an OC not prime stable. Try bumping up the NB voltage by a notch or two. That should help. Also, do you know if you have a lot of vdroop? That could cause problems.