Prime95 Is it the Be All--End All overclocking tool???

[midlifecr1515]

I've been wondering how many feel Prime 95 is a necessary step to the overclocking process? I have been hovering around 3.65 ghz on my AMD Phenom II 940 x4. I have typed in my Mushkin timings as 5-5-5-15 and am not sure whether it helps or hurts to manually enter in a mhz or leave at auto...same with voltage....type in 1.90v - 2.15v or leave auto. Anyhoo, P95 immediately or almost immediately crashes. I'm at work and don't have accurate memory on settings atm, but cpu voltage is between 1.45 and 1.53. I've gotten similar results by either keeping the bus low and turning up multiplier or vice versa. Last night I tried loosening the memory timings to hit 3.7 @ 6-6-6-18. WoW keeps crashing after an hour or two. (TBH, ALOT of people are having this problem, not just me. ) If I can play BC2, MW2, Sacred 2, Left for Dead 2 without too much trouble but it crashes Prime95, can I just write off P95 and call it good?
My mobo is AsRock 780slix3. After market cooling on everything. CoolerMaster HAF case.

 

[Rustyy117]

a few hours of game play doesn't mean your overclock is stable, but you can say with some degree of certainty that it is almost stable.

Programs like Prime95 at designed to push your CPU harder than anything else, so thats why its a general rule of thumb that if your CPU is 12+ Hours Prime95 stable then its 100% stable.

Personally I use OCCT which is like Prime95 and before that I used Orthos, they all do pretty much the same thing.

 

[wolfram23]

Linpack tests are faster and burn it harder. Try OCCT or Intel Burn Test or LinX. IMO, get it stable in one of those (makes the OC process faster as it only takes 10mins to see if it's fairly stable) and then run a Prime95 overnight, try at least 10-12hrs. Then you're good.

With all that crashing I think you need to do some tweaking. Sometimes a little more voltage will fix it. Could be a RAM issue, but fyi RAM timings depend on the RAM speed not the CPU speed (although they're related through ratios).

 

[notty22]

If your getting errors in prime 95 blend, almost immediately, its usually ram timings/to low a ram voltage, and/or running your ram to fast out of spec.
Those can be all related.
Those errors happening when processing 3d graphics could definitely lock up your system.

 

[ErnieN]

Running prime95 overnight is insane! Every hour you run it degrades all the electrolytic capacitors in your system by more than one month at normal temperatures. You are shortening the lifespan of your whole system.

 

[big_mike]

What Ernie said, you can run Prime for an hour or two in each mode and know your pretty stable.

 

[JofaMang]

Running prime95 overnight is insane! Every hour you run it degrades all the electrolytic capacitors in your system by more than one month at normal temperatures. You are shortening the lifespan of your whole system.

Proof Please. I don't disagree that running your system at max load is going to age your PC components faster, but one hour = 1 month? Was this hyperbole or do you have a source for this claim?

 

[RJR]

Running prime95 overnight is insane! Every hour you run it degrades all the electrolytic capacitors in your system by more than one month at normal temperatures. You are shortening the lifespan of your whole system.

Gotta love the electrical engineers that stop by here ,

OH, by the way, MB's have been using solid caps for quite some time now, just FYI.

 

[ErnieN]

I admit, it was just a guess, but I found this formula for determining capacitor degradation:

1/ESRt = 1 /ESR0(1-k * t ^ (-4700/(T+273))

Where

ESR = Electrical Series Equivalent resistance
T = the temperature in Centigrades
t = the number of hours
ESR0 = Start Electrical Series Equivalent resistance at t=0
k = the value of the capactor in Farads.

So... assuming ESR of 51 ohms, Capacitance at 50 uF, time of 100 hours, and temperature of 20 C, you get a degradation of .001% after 100 hours.

Same everything except changing the temperature to 100C and you get a .052% degradation after 100 hours.

So assuming failure will occur when the capacitance degrades by 10%, you can expect the same cap to last about 1 million hours at 20C but only 19200 hours at 100C. Yeah, I know. 19200 hours is about 2.2 years.

The point is that the same cap will degrade about 52 times faster at 100C than it would at 20C. So yeah... I overestimated. But still this is a terrible thing to do to your system!

And yeah, I know most better motherboards are now using tantalums in their DC-DC converters. But most power supplies, drives, graphics cards, memory, etc are still using electrolytics as tantalums are a non-renewable resource and cost much more.

 

[Pyroflea]

You will never hit 100*C, I can guarantee that, especially the capacitors.

 

[ErnieN]

You will never hit 100*C, I can guarantee that, especially the capacitors.

They are not only influenced by the temperature of the components around them but produce their own heat from the current flowing through them. Ask anyone who has ever worked with power supplies. Caps usually don't just degrade away. They blow when they dry out. They blow violently and they blow often - usually taking many other components with them.

I was being very generous predicting failure at 10% degradation. Chances are they'll take out your CPU or GPU way before the actually fail completely.

 

[redechelon]

If your getting errors in prime 95 blend, almost immediately, its usually ram timings/to low a ram voltage, and/or running your ram to fast out of spec.
Those can be all related.
Those errors happening when processing 3d graphics could definitely lock up your system.

+1 notty. Immediate crashes are most likely because of your ram if you're running on "blend".

Ernie, interesting stuff, but most don't run their CPU past 60-80.. so 100C is pushing it a little. The manufacturer states what is safe and what isn't, and most shoot to stay within boundaries.... I could run Prime95 for a MONTH, but it would never pass 70C. I'm not saying I would... I'm even cautious to run it overnight, but it's def not going to harm my CPU enough to notice a difference in life. I'm not keeping my chip for 10 years.

 

[Pyroflea]

They are not only influenced by the temperature of the components around them but produce their own heat from the current flowing through them. Ask anyone who has ever worked with power supplies. Caps usually don't just degrade away. They blow when they dry out. They blow violently and they blow often - usually taking many other components with them.

I was being very generous predicting failure at 10% degradation. Chances are they'll take out your CPU or GPU way before the actually fail completely.

I am aware of this, I was just pointing out that this isn't an overly realistic number, and was just for the purposes of demonstration.

 

[RJR]

But most power supplies, drives, graphics cards, memory, etc are still using electrolytics as tantalums are a non-renewable resource and cost much more.

I'm sorry, I have to rephrase my comment,

Gotta love the wanna-be electrical engineers that stop by here

You may want to have a look at the components you have specified, you will find electrolytic capacitors in the PSU, that's about it. As for your dooms day talk about everything blowing up because the caps dry out, it's very rare and as already stated almost all are solid caps now.

 

[RJR]

Here you go from 2007:
http://www.electronicspoint.com/solid-capacitors-t72242.html

And newer:
http://emailmarketing.msi.eu/display.php?List=42&N=248

And 2008, all solid:
http://www.pcstats.com/articleview.cfm?articleID=2239

GPU:
http://www.msi.com/index.php?func=newsdesc&news_no=837

Yeah, it's true you can still find a few electrolytic capacitors in a few "budget boards" but it's not a good idea to be OCing with them anyway, and it's still VERY rare to have them blow. Even the OP's AsRock board is all solid.

 

[JofaMang]

So pulled clear from the rectum, basically.

I have run Prime95 for dozens for hundreds of hours combined on this PSU alone, and easily hundreds of hours over the last year on my old motherboard. My new Mobo is only a month old, and it probably has 30hours P95 under its belt.

So if 1hour = 1 month
and lets say, 360 hours in the last year, my PSU must be aged about 30 years by now, surprising I can run still run my graphics cards! See, I can pull numbers out of my butt, insert them into an equation, and also come up with ridiculous conclusions. I love sharing time. Show and tell is next!

Oh wait, I only have solid capacitors. Got an equation for that?

Seriously though, I will be the last person to claim that OCing is harmless. I also wouldn't tell you that supercharging an engine makes it more reliable, or that polishing a floor doesn't make it slippery. P95 is only one of several tools you can use to determine stability, and each overclockers definition of stability is different, based on their own personal preferences. If you are afraid of being exposed to warm things, probably best to leave the kitchen to the tougher ones, you know, the woman and children.

 

[RJR]

Show and tell is next!

PLEASE, no show and tell rectum posts. :lol:

 

[JofaMang]

Haha, i didn't even realize that one might infer that I would be bringing the big 'ole brown ballon knot for show and tell. Does that make you filthier than me?

 

[RJR]

Does that make you filthier than me?

Probably, just can't resist a good rectum joke now and then, it tends to get a little stuffy around here and it's good to lighten up once in a while :lol:

 

[Scotteq]

Rectum!?!?!?


Damned near KILLED 'im!!

 

[Mugz]

Wrecked who?

 

[JofaMang]

"Dad! Dad! I was riding my horse home and tried to jump a fence when the horse got a fence pole up his ass"
"Son, don't you mean Rectum?"
"Rectum?!?!? Damn near killed 'im!"

Its an old joke that can take many forms, but hilariously regurgitated for this thread.

 

[Scotteq]

hell... that punch line is good enough to not need the rest of the joke!

 

[midlifecr1515]

I'm glad to have started this thread. It lead to some fun and some insight.
I seem stuck with my overclock around 3.5 with lowered NB multiplier and voltage than stock to keep WoW happy. WoW must have some really crappy code in how it handles memory. I read someone else's thread about it, too. Overclocking + WoW = CRASH. Seems true to me. ALL of my other games run so smooth at 3.7 and I love it. I haven't touched Prime 95 or any other testing program. I figure I don't really need them. The computer is mine so if it crashes, I just tweak the bios before the next reboot. If it was for a customer, however, I think the burn-in process would definitely be necessary.
I have been overly impressed with ASRock's mobo and the AMD 940 chip. Rock solid doesn't even come close to how amazing these components have been. I've abused the hell out of them, and they still keep on truckin'.
I had a question though: Should I mess with core voltage or PCIE bridge voltage, etc? It's below the DRAM voltage setting. Also, if I plug in 5-5-5-15, should I leave mhz on auto and also I've messed with flexibility. I've tried enabled and disabled...didn't seem to make any difference. But to get WoW to stop crashing/freezing up on me, I lowered the north bridge from 9x to 8x and voltage from 1.17v to 1.16v. Can I get someone to give me some knowledge about this? I still feel like a noob and that I'm really just guessing to numbers at the world's most complicated combination lock to computing bliss.
I know WoW is 5 years old and all, but I just love the art style, atmosphere and music it has.

Thanks for all the awesome responses.

 

[Scotteq]

Overclocking + WoW = Crash? I don't think so: my i7 is at 3.8Ghz now and WoW is stable. WoW is pretty processor/memory intensive, though, instead of the more usual GPU intensive. So you might want to visit/revisit your memory settings.