Being OCCT Stable: Does it matter?

[MaxAwesome]

Hey everyone,

What brings me here is the question whether or not we should let OCCT (overclocking Checking Tool) tells us if our system is stable or not.

Here is my case: Currently have E8400 @ 4ghz @ 1.4v BIOS (1.376v CPU-Z).

The computer is rock solid and stable AT ANYTHING. I play games for hours (CPU demanding games like GTA 4 and Prototype), I run countless loops of 3Dmark06 and Vantage, I run prime95 for hours... all stable.

Yet, it fails OCCT within minutes. Should I let this bother me?

NOTE: As I understand it, OCCT really pushes the system unlike anything else. I mean, really pushes and it seems to me that any real life application/game will NEVER push the computer this hard. Not even close.

So... Should I try to make it stable? Or just let it go? I don't really want to increase Vcore anymore.

 

[PaulieG]

Are you "old school" or "new school" when it comes to overclocking? What I mean is, do you care if your machine is truly stable, or stable enough? Old school overclocking demands total stability. Some of us hold true to this, because overclocking IS the hobby. For me this it true. It's seems like many of the younger, less experienced overclockers care less about this.

 

[MaxAwesome]

Well, the fact that it fails OCCT keeps poking at the back of my head... but it really is stable for ANYTHING else. Really, anything.

 

[FordGT90Concept]

I wouldn't trust a computer that isn't Linpack/Prime 95/benchmark stable.

 

[erocker]

Are you "old school" or "new school" when it comes to overclocking? What I mean is, do you care if your machine is truly stable, or stable enough? Old school overclocking demands total stability. Some of us hold true to this, because overclocking IS the hobby. For me this it true. It's seems like many of the younger, less experienced overclockers care less about this.

Some older experienced users also know when excessive stability testing is slowly ruining their processors which seems to be the case with many people. My computer is 1 hour Linpack stable and four months (so far) daily use stable. I trust the latter statistic over anything.

 

[kyle2020]

I run an hour of 95 then game. If its stable after that, fine by me.

 

[Evo85]

If OCCT finds an error, you CPU has a vulnerability. Now, it may never affect you. You may never notice it.

Or, you may be gaming one day, and get a BSOD that destroys your OS (yes, it HAS happened to me)... Never know.

 

[Kursah]

Some older experienced users also know when excessive stability testing is slowly ruining their processors.

Or wasting valuable gaming time which is another great way to test for stability overall. Especially with modern games utilizing 2-4 cores and running gpu's harder than ever!

I do prefer total stability, but I won't spend a week finding it/verifying it either. I've had times where stress tests have passed, yet while gaming I'd get a crash..doesn't mean there wasn't another issue with the game, but if it ran fine prior to the OC, that's always a good sign. But if I can pass OCCT's 1 hour stress anymore I'm pretty happy, then I'll run some furmark for the vid (have for a while, never killed a card, plus it's a great artifact tester), the usual memtest, and then use it for it's purpose and ultimately the stress test that matters most to me, being able to play all my games nice and smooth.

 

[PaulieG]

Some older experienced users also know when excessive stability testing is slowly ruining their processors which seems to be the case with many people. My computer is 1 hour Linpack stable and four months (so far) daily use stable. I trust the latter statistic over anything.

Obviously you can go too far. I was never a fan of the prime 12 hour runs. However, I do demand at least an hour of OCCT linpack or 20 cycles of LinX. If it can't handle that, the overclock is worthless to me. Then again, that's the hobby for me, finding the highest stable overclock. To each their own.

 

[etrigan420]

lolol, I went through this SAME EXACT scenario about a month ago...

e8400 @ 4.05 @ just over 1.3v. The ONLY way I could get it OCCT stable was to up the maximum temp that OCCT fails out at. I upped it from 80C to 90C, and it passed after an hour long run (at a max temp of 81C )

But you know what? I've had RealTemp running for about 2 weeks on this thing, and during NORMAL use, the most it's seen is 61C, which I'm comfortable with.

The fact is this: I've had it running at 4.05 for the entire month and have had exactly 0 issues with it. No blue screens, no lock-ups, no restarts, no alarms, no errors.

Am I justifying my lack of faith in hardcore benching because I (probably) couldn't run it for 27.5 hours? Maybe...but where does the line get drawn? I've seen everything from 1 hour to 12 hours being used as "It's not 'stable' unless you can run it for..." scenarios.

As such, I've come up with my own benchmark, it's like this:

<sarcasm>

Take a 3lb sledgehammer and drop it from a distance of 3 feet directly onto your processor, if it still works, you pass. Now we can argue about the weight of the hammer or the drop distance...but 3lbs. and 3 feet seem like the sweet spot.

</sarcasm>

That's what hardcore benching feels like to me...but I'm fucking weird anyway

End result? If it's stable...run with it!

 

[MaxAwesome]

I tend to agree with everyone here.

But what's bothering me is that this CPU can handle anything for hours on end, just not OCCT.

Perhaps OCCT pushes it just a tiny bit beyond what the current vcore can supply to the cpu, and it fails.

 

[MaxAwesome]

lolol, I went through this SAME EXACT scenario about a month ago...

e8400 @ 4.05 @ just over 1.3v. The ONLY way I could get it OCCT stable was to up the maximum temp that OCCT fails out at. I upped it from 80C to 90C, and it passed after an hour long run (at a max temp of 81C )

But you know what? I've had RealTemp running for about 2 weeks on this thing, and during NORMAL use, the most it's seen is 61C, which I'm comfortable with.

The fact is this: I've had it running at 4.05 for the entire month and have had exactly 0 issues with it. No blue screens, no lock-ups, no restarts, no alarms, no errors.

Am I justifying my lack of faith in hardcore benching because I (probably) couldn't run it for 27.5 hours? Maybe...but where does the line get drawn? I've seen everything from 1 hour to 12 hours being used as "It's not 'stable' unless you can run it for..." scenarios.

As such, I've come up with my own benchmark, it's like this:

<sarcasm>

Take a 3lb sledgehammer and drop it from a distance of 3 feet directly onto your processor, if it still works, you pass. Now we can argue about the weight of the hammer or the drop distance...but 3lbs. and 3 feet seem like the sweet spot.

</sarcasm>

That's what hardcore benching feels like to me...but I'm fucking weird anyway

End result? If it's stable...run with it!

Oh man, you're awesome! lol

 

[Darknova]

I tend to agree with everyone here.

But what's bothering me is that this CPU can handle anything for hours on end, just not OCCT.

Perhaps OCCT pushes it just a tiny bit beyond what the current vcore can supply to the cpu, and it fails.

I don't trust OCCT at all, Linpack, Prime95, all manner of stability tests show my PC as being 100% stable, OCCT fails in minutes.

I've been gaming for months on this overclock, I watch HD movies, I encode audio and video, nothing lets me down, but OCCT seems to think it's unstable.

I may not be "old school", but I do do some type of stability testing, but my version is a few minutes of prime95 or linpack to pick up on any major instabilities, then I game, and I game hard and long, if it fails or crashes, I tweak the overclock.

Why waste time using a synthetic stability test when the best stability test is using your PC how you normally use it?

 

[ShadowFold]

I say, linpack for 2 hours or it's not stable. I need my clocks to be 100% stable because I don't want my games crashing or anything like that

 

[kyle2020]

^ a huge +1 on that (what nova said) . These synthetic tests OVERSTRESS your system, in reality, you will never push your system that hard.

 

[hat]

I'm an oldschool 17 year old... heh
That said Linpack for 4 hours is super stable, which is what I need. I don't like when there's a stability hole anywhere.

 

[MilkyWay]

i dont know ive never had a cpu that failed OCCT and was very stable, ive had overclocks where i though it was stable but it would be random run sometimes fine for hours then crash or just bsod straight away at desktop

if its OCCT stable its usually 99% stable, ive never had an oc that passed OCCT and bsod due to stability

if it fails OCCT i find it hard to believe it is going to be stable for hours of gaming or using intensive programs because that is the experience for me its allways stable if it passes OCCT

 

[MaxAwesome]

i dont know ive never had a cpu that failed OCCT and was very stable, ive had overclocks where i though it was stable but it would be random run sometimes fine for hours then crash or just bsod straight away at desktop

if its OCCT stable its usually 99% stable, ive never had an oc that passed OCCT and bsod due to stability

if it fails OCCT i find it hard to believe it is going to be stable for hours of gaming or using intensive programs because that is the experience for me its allways stable if it passes OCCT

I'm not going to force feed you, but I assure you, i have left 3dmark looping for an entire afternoon while i was at work, i have left prime95 running overnight... never crashed, rebooted whatever.


it just doesnt seem to like occt

 

[Tau]

I'm sorry but 2 hours of linpack and calling a machine stable? using games for stability tests?

When did this become acceptable?

Anyone who has been overclocking for anylength of time will tell you that unless your ATLEAST 12 hour (but 24 hour is the standard) stable then your PC is not 100% stable.

What people fail to realise is that the blanket term "my computer is 100% stable" after 4 hours of prime95/occt is garbage. Or the fact that you can "game for hours" without a crash means nothing. if you run OCCT and it crashes in minutes that tells you somethign right away, that an aspect of your machine is NOT stable.

Running a machine that is not 100% stable (by 100% i mean 24 hours of prime/occt/linepack (not just one, all of them) can actually cause harm to your hardware. Not haveing the correct voltage/skews can cause all kinds of issues, data corruption, at both the hardware, and kernal level. Why do you think your retail CPU has a 10%fudge factor in the vcore/coreclock? so that there can be no doubt its 100% stable.

And the argument that running the CPU @ 100% degrades it is an age old argument... personally one that i dont beleave, as i have cpus that are more than 10 years old and have been running 100% load for most of their lives (8+ years)

You also need to have a feel for diffrent stress tests.... thats were us oldschool people kind of have the advantage, i can pick out exactly what part of my machine is not stable based on how the test failed. As Prime95, OCCT, and Linpack are all used not JUST for stress testing, but for component stress testing. sure they test the overall stability, but each also tests its own segment of your machine more than the others and can give you faster quick tune results, and better long term results when used correctly.

Now im sure alot of the youth clockers will go ahead and tell me that their 3 hours of call of duty 5, or crysis, and 2 hours of linpack means their computer is stable. And you know that may be true you MAY have lucked out and set your machine up right in that short ammount of testing and have it turn out to be 24 hour stable. But the VAST majority of clocks will not be.


/end rant

I just dont like how everyone magically decided one day that "gaming, and a few hours of prime" constitute a machine as being stable.

 

[kyle2020]

*Back away youth members, we've got a seasoned overclocker on our hands*

Different people deem different levels of stressing / testing to be stable. It doesnt mean that one is right and the rest are wrong, however. Personally, I couldnt run my rig for 24 hours straight stability testing - why? A. Heat. B. Electricity. C. Conscience.

Nobody 'magically' decided anything, its just a shared opinion.

 

[Tau]

*Back away youth members, we've got a seasoned overclocker on our hands*

Different people deem different levels of stressing / testing to be stable. It doesnt mean that one is right and the rest are wrong, however. Personally, I couldnt run my rig for 24 hours straight stability testing - why? A. Heat. B. Electricity. C. Conscience.

Nobody 'magically' decided anything, its just a shared opinion.

You cant affoard the extra $1 it would cost to run it at full load for 24 hours?


As for stresses overstressing your components, what do you think the wafers are run through at the factory before they pass out into the public? I'm sure its a hell of alot more than what we put them through in a day.

 

[kyle2020]

Its not me that pays the electricity bill, neither am I in the states, so it adds up and can produce some angry parents

 

[etrigan420]

I'm sorry but 2 hours of linpack and calling a machine stable? using games for stability tests?

When did this become acceptable?

Anyone who has been overclocking for anylength of time will tell you that unless your ATLEAST 12 hour (but 24 hour is the standard) stable then your PC is not 100% stable.

What people fail to realise is that the blanket term "my computer is 100% stable" after 4 hours of prime95/occt is garbage. Or the fact that you can "game for hours" without a crash means nothing. if you run OCCT and it crashes in minutes that tells you somethign right away, that an aspect of your machine is NOT stable.

Running a machine that is not 100% stable (by 100% i mean 24 hours of prime/occt/linepack (not just one, all of them) can actually cause harm to your hardware. Not haveing the correct voltage/skews can cause all kinds of issues, data corruption, at both the hardware, and kernal level. Why do you think your retail CPU has a 10%fudge factor in the vcore/coreclock? so that there can be no doubt its 100% stable.

And the argument that running the CPU @ 100% degrades it is an age old argument... personally one that i dont beleave, as i have cpus that are more than 10 years old and have been running 100% load for most of their lives (8+ years)

You also need to have a feel for diffrent stress tests.... thats were us oldschool people kind of have the advantage, i can pick out exactly what part of my machine is not stable based on how the test failed. As Prime95, OCCT, and Linpack are all used not JUST for stress testing, but for component stress testing. sure they test the overall stability, but each also tests its own segment of your machine more than the others and can give you faster quick tune results, and better long term results when used correctly.

Now im sure alot of the youth clockers will go ahead and tell me that their 3 hours of call of duty 5, or crysis, and 2 hours of linpack means their computer is stable. And you know that may be true you MAY have lucked out and set your machine up right in that short ammount of testing and have it turn out to be 24 hour stable. But the VAST majority of clocks will not be.


/end rant

I just dont like how everyone magically decided one day that "gaming, and a few hours of prime" constitute a machine as being stable.

Wow man...just...wow. This is exactly what I was talking about.

Tell you what, when referring to my rig...just to fit into your definitions, I'll use the term "Stable Enough". Would that work?

And until you do the 3lbs. at 3 feet test, I'd appreciate it if you did the same.

 

[Tau]

You cant affoard the extra $1 it would cost to run it at full load for 24 hours?


As for stresses overstressing your components, what do you think the wafers are run through at the factory before they pass out into the public? I'm sure its a hell of alot more than what we put them through in a day.

Wow man...just...wow. This is exactly what I was talking about.

Tell you what, when referring to my rig...just to fit into your definitions, I'll use the term "Stable Enough". Would that work?

And until you do the 3lbs. at 3 feet test, I'd appreciate it if you did the same.

Stable Enough would be fine, but why settle for stable enough? I see it as lazyness and a copout.

People expect to be able to overclock their computer in a day maybe 2, then wounder why it doesent work, then decide to settle for "good enough"

For a SOLID overclock (heavy tweeking) i would expect to spend a week or two at it, granted most of this is in testing single changes (omg ram timings suck) but this way you actually KNOW what is going on and can dial your machine in to whatever you want basically.


And i dont understand your 3lb 3 foot reference

 

[hat]

Tau is correct, this is an unarguable absolute. When something fails, wether it fails only on the extreme outer limits of the occt linpack test or fails something as simple as superpi 1m, it is not stable.