Declaring a O/C'd CPU "stable"

[WobbleWobble]

Reading these forums, I see that many people O/C the CPU, bump it up the next notch and their game freezes on them. But I suspect that their O/C originally isn't 100% stable.

When I mean 100% stable, I mean being able to pass Memtest86 and a Prime95 torture test for at least a few hours. If my O/C fails in like the 4th hour of Prime95, I'll bump it down another notch.

But I did try running it when it doesn't pass Prime95 but running heavy games... 3DMark, that stuff and it all seems fine.

Am I being too cautious by having my O/Cs run through very vigorous tests? Or do you guys do the same?

Edit: Refined the topic name

 

[bjc112]

Well it's truly up to you..

P95 puts the CPU under stress that games won't. If your AGP card is not overclocked you for the most part should be ok.. ( Unless of course you game for 10 hours in a row... )

What is your Vcore at?

Bump your Vdimm to 2.8v if it is not.. That can sometimes increase stability.

I would consider anything over 8hours stable.

It's totally up to you, when your in that huge battle in Max Payne 2, do you want it to freeze??!!

 

[Corporate Thug]

at first i thought my OC was stable cuz it would run all my games for as long as i wanted it too but then i ran into halo and it wouldnt run for more than a few minutes. 24 hours of PRIME is my standard now

 

[WobbleWobble]

But my point is that it doesn't freeze during games, it seems like it's 100% stable.

I'm running my XP2500+ @ 200x10.5 with 1.177 (actual) vcore.

My RAM is fine... 2.5-3-3-6 2.8 vdimm.

 

[Corporate Thug]

yes, i ran all my games all summer for several hours a day without a hitch but then one single game made me realize it wasnt stable. priming is a basic task, if the computer cant do it, i dont see how it can be completely stable.

 

[bjc112]

Originally posted by: WobbleWobble
But my point is that it doesn't freeze during games, it seems like it's 100% stable.

I'm running my XP2500+ @ 200x10.5 with 1.177 (actual) vcore.

My RAM is fine... 2.5-3-3-6 2.8 vdimm.

I assume 1.77V

Your starting to get up there.. Try 1.8V if it's not stable through at least 8+ hours of Prime, you should back it down.

 

[WobbleWobble]

Whoops, you're right! 1.77V

It's Prime95 stable at that voltage, but I don't like my voltage being that high. I think I'm already pushing it a bit. I'm already taking into the account that I think my A7N8X-DX overvolts my CPU by .05V.

I originally wanted to do at least 3200+ speeds... I can do 3200+ at 1.825 (actual) but that's really pushing it I think!

 

[dullard]

It depends on your definition of stability. An overclock can mess many things up. Temperature is just one of them. However generally a CPU will reach a high enough temperature to crash after just a few minutes of an intense program (as seen my the examples above such as "i ran into halo and it wouldnt run for more than a few minutes"). Thus running a single program for 24 hours really is only testing one area of instability and is probably much overkill on testing that area.

Even if your computer can run Prime for years straight it might be a very unstable overclock. The simple reason is that it only tests one condition - the temperature of the CPU. But many times people don't just overclock the CPU (knowingly or by accident). For example if you bump up your fsb you may very likely are overclocking EVERYTHING in your box. So while Prime runs flawlessly, as soon as you get to a program that uses sound your overclocked sound card may freeze (replace that with memory, video card, and anything else which may be affected by a fsb). Or even maybe your CPU and other fsb dependant items work just fine but produce enough heat that your HD cannot cope - leading to a crash. Did you test your HD thoroughly while the CPU was hot running Prime (and all other components)? I bet not. So again just running one program for a long time tests next to nothing.

I've almost never seen a crash while running a program for hours/days/years. That is why everyone who says this is ignorant: "my uptime running just one program and me not touching is xxx months!" Any unstable computer can run just one program for months on end. No, a true test of stability is a computer that can open/close repeatedly many different programs (thereby testing all components in stressful situations). Almost all crashes are as a program loads/starts running - so a good stability test would load/run many DIFFERENT programs for 24 hours. Do that and lets see if your computer can keep running. If it passes that test, then it is stable to me.

 

[FPSguy]

My definition of stable is that the computer runs the programs I use, the way I use them, without crashing unexpectedly. If my computer can run Prime for 10 minutes or so, that's usually more stress that I seem to put on it in real-world use.

 

[Duvie]

I use prime95 for 12hours and memtest for 10 passes....But I don't stop there....My p4b 2.4@3.24ghz would do that at 1.57v or so I thought...I then ran Divx encoding on several different movie clip testing and it would error. ..I ran several different ones ion case source vidoe was bad. I ran a program that ahs ran stable for me for many motnhs then. However I said Divx is quirky maybe it has the bug....But I didn't stop there I then ran autocad and rendered a 3d image with all the materials, smoothing, and shadows....Gave me a general exception error every time...Then I ran a Besweet test converting an ac3 file to a wav file...errored!!! So what does that tell you????

I know run those same things for more of a heat test and tress test, but when it comes to the C1 chips I run also an mpeg2 or Divx encoding, and my autocadd which I use regularly. If it passes those test I have no worry. Throw in I like to do those things why running F&H or Prime95......

 

[WobbleWobble]

Originally posted by: dullard
It depends on your definition of stability. An overclock can mess many things up. Temperature is just one of them. However generally a CPU will reach a high enough temperature to crash after just a few minutes of an intense program (as seen my the examples above such as "i ran into halo and it wouldnt run for more than a few minutes"). Thus running a single program for 24 hours really is only testing one area of instability and is probably much overkill on testing that area.

Even if your computer can run Prime for years straight it might be a very unstable overclock. The simple reason is that it only tests one condition - the temperature of the CPU. But many times people don't just overclock the CPU (knowingly or by accident). For example if you bump up your fsb you may very likely are overclocking EVERYTHING in your box. So while Prime runs flawlessly, as soon as you get to a program that uses sound your overclocked sound card may freeze (replace that with memory, video card, and anything else which may be affected by a fsb). Or even maybe your CPU and other fsb dependant items work just fine but produce enough heat that your HD cannot cope - leading to a crash. Did you test your HD thoroughly while the CPU was hot running Prime (and all other components)? I bet not. So again just running one program for a long time tests next to nothing.

I've almost never seen a crash while running a program for hours/days/years. That is why everyone who says this is ignorant: "my uptime running just one program and me not touching is xxx months!" Any unstable computer can run just one program for months on end. No, a true test of stability is a computer that can open/close repeatedly many different programs (thereby testing all components in stressful situations). Almost all crashes are as a program loads/starts running - so a good stability test would load/run many DIFFERENT programs for 24 hours. Do that and lets see if your computer can keep running. If it passes that test, then it is stable to me.

Let me narrow the scope of the topic a bit more. Declaring a O/C CPU "stable" .

 

[ponyo]

Originally posted by: Duvie
I use prime95 for 12hours and memtest for 10 passes....But I don't stop there....My p4b 2.4@3.24ghz would do that at 1.57v or so I thought...I then ran Divx encoding on several different movie clip testing and it would error. ..I ran several different ones ion case source vidoe was bad. I ran a program that ahs ran stable for me for many motnhs then. However I said Divx is quirky maybe it has the bug....But I didn't stop there I then ran autocad and rendered a 3d image with all the materials, smoothing, and shadows....Gave me a general exception error every time...Then I ran a Besweet test converting an ac3 file to a wav file...errored!!! So what does that tell you????

I know run those same things for more of a heat test and tress test, but when it comes to the C1 chips I run also an mpeg2 or Divx encoding, and my autocadd which I use regularly. If it passes those test I have no worry. Throw in I like to do those things why running F&H or Prime95......

You've very good method of testing stability. One of the better ones I've seen. Kudos.