• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Three Levels of Stability Methodolgy?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

RGone

Senior DFI Staff
Joined
Apr 20, 2004
Location
JAX, Mississauna
Okay we have been given a mission for this thread. The reason for starting this thread is still true and outlined below. But we have been asked to try for a consensus as outlined just below. Thank you.

Can we in this thread develop a basic set of guidelines as follows:

1. What is the recommended quick test method to check an overclock for stability (less than an hour of testing)

2. What is the recommended good test method to check an overclock for stability

3. What is the extreme recommendation to ensure you almost definitely won't see a crash under almost any condition

I feel some of the three points above will overlap, but we should do our best to fulfil the three criteria as outlined above. The problem with trying to develop such guidelines is personal bias. I have it. We all have it. In those instances I have to think out of my box and over at someone else's computer desk.
RGone...ster.

Dear Moderators:

You may feel free to position this thread in a more appropiate area if you so deem. Thank you.

Right now viewers per forum:
Cooling 75 viewers
Cpus 175 viewers
Motherboards 127 viewers
Video Cards 111 viewers
Storage 65
Memory 40
Display and Sound Technology 65 viewers
The rest of subforums under "Hardware" are on average showing less than 30 viewers.

The topic of how to do Stability Testing and probably a hot topic has been brought to my attention. Do we as posters have any responsibility for what we post? Looking at the numbers of viewers as shown above, I think we do have at least a modicum of responsibility to ensure that if we are posting an opinion that is inline with the general community; that we make our statements with a clear disclaimer that I am posting 'my personal opinion". We have far more viewers than posters. Far more viewers than 'helpers'. So I try to ensure my personal opinion is not going to be too great an influence on those that are in the viewer community, where they may have no idea about where I am standing when I voice an opinion.

Also open to discussion is the idea that my latest opinion, may be the new wave of technology and procedures. There is no sense in remaining in the past if there are newer and better procedures that we all might have consensus about.

LINK >>
What prompted me to start this thread was this thread and posts in it.

RGone said:
1. You have a C2 stepping processor. Not as good as a C3 stepping in general. After all you said you had it a pretty long time.

3. If Prime95 passes and we can see the Screen capture of HWMonitor we will know where the board stands and knowing you have an upward unlocked processor you can bump the muliplier to 18 from 17 and Prime95 again for at least 2 hours. If Prime95 fails immediately...then up Vcore 0.05 and retest.
Need to see Screen capture of the HWMonitor as it was running during the 2 hour Prime95 @ 18 multiplier.

4. When Prime95 is fully stable at the 18 multiplier for 2 hours...you can test the 19 Multiplier. Need to see Screen capture of the HWMonitor as it was running during the 2 hour Prime95 @ 19 multiplier.

storm-chaster said:
I think prime is overrated. Use the integrated system stability test within everest or aida64, just as effective. Like breaking in a new engine, you wouldnt stress it to the limit or 8000rpm redline for 17 hours, nor would you see 100% load on your cpu for a sustained 17 hours. Sure it can prove that your system is bulletproof, but if it fails after 3 or 4 hours it doesnt mean your system in its overclocked state is unstable. To each his own. I've seen many overclocks over the years that pass prime for a while, but eventually fail. In real life you never see BSODs or have any reliability problems. but again, just my experience.

So I see that the idea that Prime95 is over-rated is set forth as an opinion as it likely should be.

I can go back to my first experience with Prime 95. A forum I was a member of was doing some overclocking on a particular motherboard series. Our overclock submissions had to include captures of validations including 8 hours of Prime95. WTH? Prime who? Prime Rate? Prime my pump?

When I threw Prime at my glorious overclock...well my stuff was a failure. Deeyam P95. I was now sitting on the floor in the footspace of my computer desk. Shaken and feeling beaten.

The rise from the ashes of my first experience with Prime95 was one of the experiences of my lifetime. I learned more about overclocking in general with my quest to conquer my Prime95 failure, than I likely ever could in any other process. I learned what was needed to push my rigs in an over-speeded context and remain what was deemed stable by Prime95 and my peers at the time. I learned what relationships of hardware and software mattered. I mean if I was really going to do calculations that might unearth a real prime number, then I did not need a computer that could not pass Prime95 testing parameters.

So have we gone beyond the days of Prime95 and perhaps many of our tests or procedures for stability testing? Is it time to come to a new consensus for stability testing?

I don't know. I know I went kicking and screaming toward testing with programs like Prime95. After much perceived hardship, I came to have a fond if tempered affection for Prime95. But is it time to change? Do we have a new and better way? Again I do not know. That is what this thread is all about. Have we accrued enough valid information to perhaps develop a new consensus? I still don't know and is the reason for this thread. I suggest that opinions are like noses in that we all have one. Make your postulations with some valid background data.

You are now returned to your regularly scheduled programming. RGone...ster.
 
Last edited:

onefstsnake

Member
Joined
Nov 20, 2007
Location
Fburg, VA
Ive always used Prime95 to test for stability. If a rig will pass 12+ hours of Prime95 then 99% of the time I will have no issues with the overclock.
 
OP
RGone

RGone

Senior DFI Staff
Joined
Apr 20, 2004
Location
JAX, Mississauna
Ive always used Prime95 to test for stability. If a rig will pass 12+ hours of Prime95 then 99% of the time I will have no issues with the overclock.

But that is only 99% of the time. What will bring us that other 1%? Multiple tests?

Can we try to put that 1% into context? If I used the computer for 100 hours then I could expect stability for 99 hours. What of the 1% then? That 1% would mean I had problems for 1 hour. So 100 hours is 4 days and 4 hours and over that time period, I would expect 1 hour of issues? Or am I just plainly nuts? Don't all of you shout yes at one time! Hehehe.:screwy:
 
Last edited:

Archer0915

"The Expert"
Joined
Nov 3, 2008
Well it really is a matter of opinion not a fact. I have had overclocks pass prime but give me errors in WCG and I had to downclock. I personally do not use it lieu of a more stressful environment. I have found that first getting the rig crunching and folding before starting the testing gets the rig heated up. I then run OCCT Linpak. I will run as many tests as possible; concurrently, to crash or error out the rig. To me prime is good but does not stress enough.

So prime may be Good for some, over kill for others and it is too weak for me.
 

wingman99

Member
Joined
Dec 10, 2003
Well all the cpu's i have had will pass any test + prime95 till I overclock, then I go from there.:popcorn::cool:
 

Automata

Destroyer of Empires and Use
Joined
May 15, 2006
RGone, try LinX. http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface

When using all the RAM, it substantially shortens the time it takes to find instability. I went from finding errors an hour or two into the test down to minutes.

We do have to take responsibility for what we post. Regarding stability, it is a game of probability. If you get errors a few minutes in, the system is likely quite unstable while loaded and may do browsing/gaming fine but crash under heavier load (rendering videos, etc). Some don't mind a little instability if they get more speed, which is a personal preference. Personally, I need my system to be 100% rock solid when it comes to computing, so I make sure that it will past any test that I throw at it all of the time. Compiling code is not forgiving when it comes to instability and may present in very odd ways. If an error happens in a game, there is the possibility of nothing important failing (graphical artifacts) and everything continues working.

The longer a stability test runs without errors, the less likely you will see problems in day-to-day activity.
 
OP
RGone

RGone

Senior DFI Staff
Joined
Apr 20, 2004
Location
JAX, Mississauna
RGone, try LinX. http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface

When using all the RAM, it substantially shortens the time it takes to find instability. I went from finding errors an hour or two into the test down to minutes.

We do have to take responsibility for what we post. Regarding stability, it is a game of probability. If you get errors a few minutes in, the system is likely quite unstable while loaded and may do browsing/gaming fine but crash under heavier load (rendering videos, etc). Some don't mind a little instability if they get more speed, which is a personal preference. Personally, I need my system to be 100% rock solid when it comes to computing, so I make sure that it will past any test that I throw at it all of the time. Compiling code is not forgiving when it comes to instability and may present in very odd ways. If an error happens in a game, there is the possibility of nothing important failing (graphical artifacts) and everything continues working.

The longer a stability test runs without errors, the less likely you will see problems in day-to-day activity.

Okay so I might do better running LinX ahead of Prime95, IBT and OCCT which I generally use.
 

RollingThunder

Destroyer of Trolls & Spammers
Joined
Jan 7, 2005
RGone, try LinX. http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface

When using all the RAM, it substantially shortens the time it takes to find instability. I went from finding errors an hour or two into the test down to minutes.

We do have to take responsibility for what we post. Regarding stability, it is a game of probability. If you get errors a few minutes in, the system is likely quite unstable while loaded and may do browsing/gaming fine but crash under heavier load (rendering videos, etc). Some don't mind a little instability if they get more speed, which is a personal preference. Personally, I need my system to be 100% rock solid when it comes to computing, so I make sure that it will past any test that I throw at it all of the time. Compiling code is not forgiving when it comes to instability and may present in very odd ways. If an error happens in a game, there is the possibility of nothing important failing (graphical artifacts) and everything continues working.

The longer a stability test runs without errors, the less likely you will see problems in day-to-day activity.

Okay so I might do better running LinX ahead of Prime95, IBT and OCCT which I generally use.

Bob, Thiddy,

This is a great idea.

We may want to have included exactly how a machine is tested; step-by-step. There are many valid ways to reach valid stability.

I never use Linpak, OCCT or Furmark. I'm a stock cooling overclocker, nothing high-end so I don't need to put my equipment through that kind of testing. Mine are daily working machines with mild overclocks. What works for me may not work for the aggressive guys. My methods have never failed me for my needs.

I would love to see how others who do high-end overclocking or those like Thiddy that require 100% stability do it. I can post my methods for a beginner quite easily, they aren't complicated.
 

Automata

Destroyer of Empires and Use
Joined
May 15, 2006
Mine are not complicated either. It is simply finding the most stressful program and running it for as long as you deem necessary. I usually run it for a few days, making sure that the ambient temperatures go higher than normal. :shrug:
 

grumperfish

Member
Joined
Nov 10, 2005
I don't really go for highly-aggressive settings, and end up relying more on 100% stability, as I don't really have a ton of free time to play with.

I usually just stick to OCCT/IBT for CPU testing and Kombuster for testing GPU overclocking. I don't have the time or patience for 10 hour P95 tests, so I might look into linx when I'm doing the SB update in a couple weeks.
 
OP
RGone

RGone

Senior DFI Staff
Joined
Apr 20, 2004
Location
JAX, Mississauna
I have no interest in RE-inventing the wheel...

There are good threads in this forum for sure. Wheels already do turn.


Big three in and of "Stability".
Of further importance, is the factor of general system stability at manufacturer settings. Before overclocking or tweaking your machine, one should ensure that all hardware is operating properly and with integrity. Overclocking an unstable machine then stress testing it, finding errors, and spending hours troubleshooting a problem inherent to some faulty component, isn't something anyone wants to put themselves through! Sometimes a component is simply faulty to begin with, and needs to be replaced or switched out.


Holy hotrod it is so hard for me not to interject any car corrolary here. Just a minute one then. How the heck does it perform before I start beating on it. There got that out the way.

When I saw "felinusz" speaking to establishing a baseline...I felt kinship for what I had been taught to do in anything I intended to mod. For my part I am convinced that is the first step to accomplish before twisting the wick up.
 

I.M.O.G.

Glorious Leader
Joined
Nov 12, 2002
Location
Rootstown, OH
My personal taste is run prime95 for a bit, not as long as most people would tell you is necessary, then go on using the computer for what I use it for. If it crashes or errors at any point, look at modifying frequency, voltage, or fan speeds, then go on using it some more.

Best test of stability is putting it to the task you intend to use it in, so long as its not going to crush your heart or nuts if it crashes. I use prime95 for a quick test typically, then take my chances after that. If I had the time or patience to run prime95 longer or run multiple stability tests, I would do that...

But I'm a strong believer in doing stuff and seeing what works. Running prime95 for 10 hours or something is more safety or assurance than I feel I need.

That said, I think this thread should develop a basic set of guidelines as follows:

1. What is the recommended quick test method to check an overclock for stability (less than an hour of testing)

2. What is the recommended good test method to check an overclock for stability

3. What is the extreme recommendation to ensure you almost definitely won't see a crash under almost any condition

If this thread can establish a consensus on those three things, the result could be posted as a sticky in all the CPU forums if that is put in the first post, and the thread title is made better.
 

MattNo5ss

5up3r m0d3r4t0r
Joined
Aug 11, 2008
I use 5-10 LinX, IBT, or LinPack passes for quick stability since they seem to find errors quickly. However, I don't use those programs for temp testing a 24/7 OC since they heat up your CPU far more than any real-world program that I know of. That shows there are varying degrees of 100% load, P95 does 100% and LinX does 100%, but Linx gets the CPU much hotter. So, typically, what I do is run LinX for stability, and not worrying about temps as long as the CPU isn't throttling at TJmax. Then once LinX passes, I'll use P95 for 30-60 mins for testing my max temps and thermal headroom.

Once I've reached my vcore and/or thermal limit using the above method, I'll start using the PC and hope for the best :D
 
OP
RGone

RGone

Senior DFI Staff
Joined
Apr 20, 2004
Location
JAX, Mississauna
I'll start using the PC and hope for the best :D

I'll start using the PC and hope for the best. Humh? That sounds as if you have planned for the 1st and 2nd parameter set forth in the first post now but what about that "hope for the best"?

Should we have increased the time frame of #1 and/or #2 testing that you did and thus perhaps not have to "hope for the best"?

What really puts us in the frame of mind that the system is truly stable and the only thing we believe will cause a crash is going to Windows Update?
 

Bobnova

Senior Member
Joined
May 10, 2009
Short testing: IntelBurnTest.
Medium testing: MoreIntelBurnTest.
Really *&@% sure testing: Run stock.

Seriously, if you absolutely cannot have crashes you have absolutely no business OCing in the first place!

If you simply must, my long test would be:
HyperPi32m a few times for ram.
Lots of IntelBurnTest and lots of Prime95 for CPU stability.
 

MattNo5ss

5up3r m0d3r4t0r
Joined
Aug 11, 2008
I'll start using the PC and hope for the best. Humh? That sounds as if you have planned for the 1st and 2nd parameter set forth in the first post now but what about that "hope for the best"?

Should we have increased the time frame of #1 and/or #2 testing that you did and thus perhaps not have to "hope for the best"?

It's because you can never guarantee CPU stability outside of stock CPU/IGP speeds and supported RAM speeds of that CPU's IMC. It doesn't matter how long you run stability tests b/c the CPU could have failed 1 minute after you stopped the test, but you'll never know.

What really puts us in the frame of mind that the system is truly stable and the only thing we believe will cause a crash is going to Windows Update?

:rofl:
 
OP
RGone

RGone

Senior DFI Staff
Joined
Apr 20, 2004
Location
JAX, Mississauna
Okay, Okay and humh!

I don't want to chase a rabbit as I am wont to do, but do we need to describe who should not overclock? I was associated with DFI for 6.5 years. I had to reach the internet everyday and my email had to work and my files needed to remain in place. I had one computer that was never overclocked and I did not run willy nilly to windows update either. That computer ran for 6 years with only replacing the memory twice. Ram just failed.

I did that because those who maintained computers for companies, said to leave it stock if I wanted to be pretty darn sure it would not fail.

I did images of my partitions regularly. Data security was paramount in my mind.

I had other computers to clock to heck and back.

Are we too lax in our remembrance that overclocking is running stuff out of the stuffz expected specs?
 

Archer0915

"The Expert"
Joined
Nov 3, 2008
I think what we are looking at here is not necessarily a bad thing and perhaps a stability guide may be in order.

Level one stability should be rock solid both crash and error free and this is really impossible because even stock clocked crashes from time to time and this is due to the software having a conflict or hickup. I have had plenty of new setups and when I was setting my baselines and updating everything the drivers may crash or the scary update may cause problems.

The question of what rock solid is becomes more of a pot luck thing. I personally say if it can pass all calculation tests without errors, pass every common torture test without shutting down while running some DC in the background wothout errors it is rock solid.

Level two stability would be in line with what many suggest. A crash free day even though it may error out on some calculation tests and the heat is not so much that it causes any issues.

Level 3 stability would be what many of us consider unstable. This is the level where you may get an error from time to time along with the normal daily crash. You games do fine and that is all that matters though you have to downclock to play story based ofline games so you do not lose your progress. This is just stable enough.