IntelBurnIn question

why would you care if someone types that their pc is completely stable, but it's not really stable?

What in the heck is wrong with you man? It depends on what you mean by "care".

Am I personally impacted or damaged by somebody putting misleading or false information on this hobbiest forum? Not unless I act on the information and loose data or hardware.

But the thing is I am a grown up. I understand that the world is made up of more than just myself. I understand that as a member of this community, I have engaged in a social contract with other people of like mind. We all have a obligation to each other, to move this hobby forward, and to help each other, that is why we are here.

Why are you here?
 
why would you care if someone types that their pc is completely stable, but it's not really stable?

Maybe when those same people start recommending voltage settings / overclock settings to others when those settings are not stable at all would be a cause for concern
 
I just finished running the Intel Burn Test myself. I'd been wanting to try it for a while, but put it off until I upgraded to Gigabyte's EP45-UD3P. I have a Core 2 Duo E4500 @ 3GHz (333x9) with vCore at 1.408v (according to CPU-Z 1.49), and a Xigmatek HDT-S1284EE + Arctic Cooling MX-2 for cooling. I used CoreTemp 0.99.4 and the Performance tab in Windows Task Manager to monitor activity.

Temps never went above 70 degrees C, it averaged about 69 with brief flickers up to 72 degrees, mostly near the end of the test. I only ran the test for 5 passes, but my system passed with flying colors. No BSODs or crashes, the IBT utility reported no errors. I should probably mention that I had the case panels off at the time, and the room temperatures would probably be considered 'cool' by some.

I'm actually rather proud of myself right now. :)

Again, I only ran it five times, so I can't say how my system would hold up under more stress, but IBT doesn't appear to be some insurmountable wall. It IS possible to have a stable system and pass IBT. :D
 
Why would you type that your PC is completely stable when it's not really stable?

i wouldnt. i was just wondering why he couldnt leave it at "both sidea are right, if your pc is stable enough for you, fine. if its not, make it more stable." without getting the last word in and insisting that no, he was right and the other guys are wrong. and why am i here? maybe i stated the reasons i am here more than 2 months ago. which would mean that you (lens pirate) werent here to read them.
 
I am going to have to agree with Vanilla_Guerilla. "both sidea are right, if your pc is stable enough for you, fine. if its not, make it more stable."
 
Stability is a relative term. Every system is different. You can have the exact same hardware as someone else, use the same overclocking settings, they are stable, you are not. That is just the way it goes. Overclocking is not guaranteed. I have seen a bone stock system with no overclocking fail OCCT, I have seen my Q6600 pass OCCT and Prime95, but lockup after playing Crysis for one hour. Gave the CPU some more voltage, Crysis never locked up again. So I do believe in the statement "stable enough". If it doesn't lockup or BSOD during whatever you use it for, who cares if it is "Prime95 or IBT" stable. People should use information from people here as a reference, not a blueprint to overclocking their own systems.
 
Stability isnt relative. It can either pass Prime/IBT or it cant.

That being said, sometimes my system can pass IBT, sometimes it BSODs. I dont consider it stable despite being able to pass IBT sometimes but not others. Im on water, so i believe temps have little bearing. If i had to guess, id say its memory related though, and not CPU OC related.
 
If it cant pass at least 10 passes of IBT, its not stable. Your PC may be stable for day to day use and gaming (which doesn't really tax your system that much) and most people are perfectly happy with that. For me that is not good enough. I'm a ProTools user so I demand 100% stability. At my studio I run a stock PowerPC dual core G5 but at home I use my PC which has Mac OS X and Vista installed. You know what the worst thing is? Having your computer crash in the middle of a studio session! Ahhh! Embarassing! That is why I test all my PC's to be at least 20 passes of IBT stable. I want the best of both worlds when it comes to high oc and stability so if I had the choice of 3.8ghz 100% stable and 4.0ghz 99.99999% stable I would take the 3.8 just because of that one instance where I may be doing something important and it might crash.
 
If it cant pass at least 10 passes of IBT, its not stable. Your PC may be stable for day to day use and gaming (which doesn't really tax your system that much) and most people are perfectly happy with that. For me that is not good enough. I'm a ProTools user so I demand 100% stability. At my studio I run a stock PowerPC dual core G5 but at home I use my PC which has Mac OS X and Vista installed. You know what the worst thing is? Having your computer crash in the middle of a studio session! Ahhh! Embarassing! That is why I test all my PC's to be at least 20 passes of IBT stable. I want the best of both worlds when it comes to high oc and stability so if I had the choice of 3.8ghz 100% stable and 4.0ghz 99.99999% stable I would take the 3.8 just because of that one instance where I may be doing something important and it might crash.

The problem is that there is no stopping point. Who came down off the mountain and said that 10 passes of IBT was stable? Some people consider 20 to be stable, others 50. No matter how you want to slice it, it all comes down to what you the user thinks is stable enough. Bottom line, nothing is ever stable because stable isn't an absolute. It's not a black and white world.
 
Stability is a relative term.
No, it isn't. It's an absolute term. Either you're stable or you aren't. And if you fail even a single stress test, you aren't stable. Simple or that. There's a very significant difference between stable and stable enough (which is what some people in this thread seem to be attempting to pass off as real stability).
 
No, it isn't. It's an absolute term. Either you're stable or you aren't. And if you fail even a single stress test, you aren't stable. Simple or that. There's a very significant difference between stable and stable enough (which is what some people in this thread seem to be attempting to pass off as real stability).

IMO it is still relative. You are still limited by the duration of the testing. How long did run IBT/LinX, 1 hour, 24 hours, 48 hours? How would you know if it wont crash after that time? Cant you say for sure that you system can absolutely run IBT for the lifetime of the cpu, say 5 years? So it is still a relative term, limited by the duration of the testing and is not an absolute stability for the lifetime of the cpu.
 
IMO it is still relative. You are still limited by the duration of the testing. How long did run IBT/LinX, 1 hour, 24 hours, 48 hours? How would you know if it wont crash after that time? Cant you say for sure that you system can absolutely run IBT for the lifetime of the cpu, say 5 years? So it is still a relative term, limited by the duration of the testing and is not an absolute stability for the lifetime of the cpu.
In that sense it is sort of relative, simply because it is impossible to definitively prove that a system will be 100% stable at all times. However, it is not relative from the perspective that a system which is stable enough for someone's purposes but fails stress testing is not truly stable, even though that person might believe that it is.
 
IMO it is still relative. You are still limited by the duration of the testing. How long did run IBT/LinX, 1 hour, 24 hours, 48 hours? How would you know if it wont crash after that time? Cant you say for sure that you system can absolutely run IBT for the lifetime of the cpu, say 5 years? So it is still a relative term, limited by the duration of the testing and is not an absolute stability for the lifetime of the cpu.

lol intel uses IBT to test the lifetime stability of their cpus at set clocks, for short periods of time. Running IBT for 5 years straight is pointless.

Thats why people recommend 20 IBT runs or more to test out stability. IBT isn't as good of a measure of pinging northbridge voltage settings though, 20 IBT runs stable and 12 hours P95 stability = a rock solid stable system in my book.
 
No, it isn't. It's an absolute term. Either you're stable or you aren't. And if you fail even a single stress test, you aren't stable. Simple or that. There's a very significant difference between stable and stable enough (which is what some people in this thread seem to be attempting to pass off as real stability).

I never said that if you fail a stress test your system is stable, what I said was there is never a limit to how many tests you can run while chasing "stability". How many passes of IBT is "stable"? For you, 10. For some other guy, 20. How many is enough to call it stable? Obviously if I run 5 passes and it fails then it isn't stable. If I run 10 passes and it fails it isn't stable. But if I run 10 passes and it passes, is it stable? How do you it wouldn't have failed on that 11th pass? Do you get my point - you can follow the stability rabbit as far down the hole as you want, because there isn't an end. There isn't a shining light that says "37 passes of IBT = Stable".

So you test to what you feel comfortable with, and you call it good. Otherwise you'd be running IBT in the background all the time.
 
lol intel uses IBT to test the lifetime stability of their cpus at set clocks, for short periods of time. Running IBT for 5 years straight is pointless.

Thats why people recommend 20 IBT runs or more to test out stability. IBT isn't as good of a measure of pinging northbridge voltage settings though, 20 IBT runs stable and 12 hours P95 stability = a rock solid stable system in my book.

My experience showed that 20 IBT runs does not necessarily translate into a rock solid system. My system was 20 IBT runs stable but BSOD 3 hours into a 1080p x264 Bluray encoding.

1080p x264 encoding with an insane/constant quality settings can take hours even with a highly overclocked i7. That's why im in the camp that you test your system for your usage. What good is a 20 IBT runs stable if fails a 4 hours 1080p x264 encoding session? The same argument goes for a system that fails even 1 run of IBT but can run games all day long without any problem and the user only uses the system for gaming and uses nothing that stresses it as heavy as say 1080p x264 encoding. You test the stability of your system for your own usage of the system. As long as it is stable with the program that you use, it is a stable system in my book.

Where do you get the information that Intel uses IBT to bin their CPU? AFAIK IBT uses Linpack and it is not from Intel or am I wrong here?
 
I've had a 10 IBT run, 8 hour Prime stable system BSOD during 1080p x264 encoding as well. I'm starting to wonder if the software itself just isn't very stable, cause Prime95 and IBT both make my cpu run hotter.
 
This discussion is very entertaining but not very technically accurate. To an engineer "stabilty" would be proper operation based on a set of specifications and a test critera. Without the parameters and conditions of "stablity" defined it is all bullshit. Every decent piece of hardware made has had some testing (call it "stablity" if you want) for proper operation under some set of conditions during design/manuf. I submit that under the right conditions I could make any machine the owner thinks is "stable" fail. How many of you have run your "stable" machines with 105 deg F ambient and an AC input voltage of 100V AC 24x7 for 2 weeks on a card table in the middle of a desert ? (an simplified example of condtions that might be required for "stablity" in a military computer or the device running a cell tower in Yemen). What about a rain forest? Remote data aquisition on the South Pole ? A good example is this sites "heat chamber" where to determine the "stablity" of a power supply a cardboard box is placed over the test machine. Anyone here claiming their machine is "stable" run it under a box for 4 hours ? Would such a test condition be vaild ? Under what machine use would such a test be vaild ? What would be a valid set of test conditions for whatever you guys want to call "stable" ? Looping some test program for some arbitary time while having at best fuzzy ideas of what it is doing to what in the hardware in your living room somehow seems lacking as a test specification.


Without a set of specifications as to the environmental, electrical, software and time constrains of a testing enviroment you are all blowing smoke out your .... And even when such conditions were specified the machine could only be considered "stable" when the envelope of those conditions were not exceeded. Outside of the those conditions the "stablity" of the machine would still be unknown.

Therefore as there is no published universal "testing for computer stablity" specification each person can and does set their own critera and determines if the machine is "stable" and all are valid.

----------------

Some thought should also be given to the fact that most of us use coretemp or the like to read the DTS sensors embedded in the chip die. Well here is a news flash, there are only a few sensors sprinkled around the die and while Intel did take (according to their technical journal) care to place them in areas they determined the most sensitive one could easliy write software that stresses a die area Intel deemed less likley to be heavliy used and has no close sensor and cause all kinds of havoc. My point is that a deeper understanding of what and how a microprocessor works and is physically laid out and how the thermal management is designed would result in posts that would not make me snort coffee on my keyboard. And then there is the motherboard with its I/O and memory subsystem and what about the graphics card ? I am tired of running my keyboard throught the dishwasher.

Applying your personal "stablity" critera to a machine belonging to someone else in a different situation is a vaild as teats on a Boar.

Instead of debating a point on which both sides are on shaky ground get [H]ard and come up with a spec or specs (one for gaming machine one for general use/web/office etc. etc. ) and define the parameters of "stable". Good luck reaching an agreement. Gotta go wash my keyboard, again.
 
Nice post Bill,

About time someone brought some Sanity to this "discussion" . :D

To test "stability" on my machine I game, browse the web, read e mail, run 3D Mark Vantage and check my voltages with a VOM. If Vantage or gaming and just day to day use does not cause me any headaches or crashes in my experience my rig is stable period, no matter what some nimrod claiming that unless I run 24 hours of Prime95 or 10 passes of IBT says.

In fact 3D Mark I have found is the best indicator of whether my system is stable or not. It'll crash pretty damn quick and either hard lock the machine or make the machine reboot or hard lock and make my rig squeal like a stuck pig. When this happens I either up my voltages a bit or back my clock down till my rig is happy, and that makes me happy. :D
 
My computer sits in a hello kitty dream house! and I wear a tin foil hat...my pc never crashes but I have BSOD on occasion.
 
I too think 3D mark is overlooked on testing a gaming machine. I use the free version all the time being the cheap bastid I am .
 
Totally agree with you there Bill Parrish. Now lets wait from the people that say you have to run xxx for xxx hours or else you are not stable.

If Vantage or gaming and just day to day use does not cause me any headaches or crashes in my experience my rig is stable period, no matter what some nimrod claiming that unless I run 24 hours of Prime95 or 10 passes of IBT says.
QFT
 
While being 10 pass IBT and 16 hour Prime stable does not prove absolute stability, they're still very useful programs for two reasons.

First, we HAVE established that if a system is 10 pass IBT and 16 hour Prime stable, it's VERY unlikely that it will have stability issues with any other programs. Who really wants to use their daily applications, which might be manipulating important data, to test an overclock? I would rather just run IBT and Prime (and be 99.99% certain that every other application will work perfectly) than worry about instability every time I open a new application.

Second, IBT and Prime provide a great reference for comparison (to other systems or just for our own sake to see how a setting affects stability).

People who don't use these programs are lazy, don't care if their system crashes sometimes, or they just don't know enough about overclocking to make their system stable.
 
Bill I thought more about your post last night. Your logic is flawed.

Each user sets the environmental variables by virtue of using the PC on his or her desktop.

Using these testing programs establishes stability for the user’s unique circumstance and are more valuable than picking arbitrary criteria and imposing it on everyone.

The argument is about what is enough testing, and what to test with to ensure your machine is solid and not likely to go tits up on your desktop.
 
While being 10 pass IBT and 16 hour Prime stable does not prove absolute stability, they're still very useful programs for two reasons.

Any one program is just going to give you a narrow check of stability. I had my q6600 up to 3.6 last night crunching small FFTs in Prime without error and CPU-Z just crashed for no reason as Prime kept going.
 
While being 10 pass IBT and 16 hour Prime stable does not prove absolute stability, they're still very useful programs for two reasons.

First, we HAVE established that if a system is 10 pass IBT and 16 hour Prime stable, it's VERY unlikely that it will have stability issues with any other programs. Who really wants to use their daily applications, which might be manipulating important data, to test an overclock? I would rather just run IBT and Prime (and be 99.99% certain that every other application will work perfectly) than worry about instability every time I open a new application.

Second, IBT and Prime provide a great reference for comparison (to other systems or just for our own sake to see how a setting affects stability).

People who don't use these programs are lazy, don't care if their system crashes sometimes, or they just don't know enough about overclocking to make their system stable.

From my experience 10 IBT runs is no where near stable enough to complete a 1080p x264 Bluray encoding.
 
I think the point being made here is not that there is some sort of absolute measure of stability that everyone should be adhering to. Your system being 'stable enough' for you is perfectly acceptable....for you. I think the problem some people are having with this when you come to a forum dedicated to maximizing the performance of your system (what this forum is dedicated to) and, because your system is 'stable enough' for you without any kind of objective verification to back it up, you feel its acceptable to participate in its discussion.

A forum like this exists to exchange and provide information. In order to maintain any kind of credibility, that information needs to be as objective and accurate as we can possibly make it. Dispensing information based on what you 'feel' or what's 'good enough for me' without any kind of objective basis behind it is dishonest and possibly destructive to the people who don't know any better. Asking 'why do you care if I say my system is stable when it really isn't?' just demonstrates the kind of behavior that doesn't belong in that kind of environment. Its not a matter of whether its stable enough for you or me, its a matter of 'is your system stable enough that you can honestly advise someone to use the same settings as you'?

Yes, everyone's system is different. What is stable for one person will not necessarily work for the other. But taking that attitude and using it as a basis to completely discard objective testing renders this entire forum meaningless.

"Yeah, I could help you with your overclock, but since my system is not 100% exactly like yours its not guaranteed that my advice will work for you. So I won't bother."

If that's what you believe. Don't post on this forum. Don't help other people on Overclocking & Cooling. If, on the other hand, you are interested in helping and contributing despite that, which do you think will look better in the eyes of a newbie whose looking for some good advice?

1). "I've overclocked my CPU to <insert_awesome_speed_here>. My system is not exactly like yours, so I can't guarantee this will work for you. However, I have run several objective stress tests that you can download and recreate on your own system. My overclocked system has passed these stress tests successfully. So, based on reliable and repeatable information (that disregards my own personal feelings on the matter), I can say these settings have a GOOD CHANCE of working for you."

2). "I've overclocked my CPU to <insert_awesome_speed_here>. My system is not exactly like yours, so I can't guarantee this will work for you. However, my system has performed well in day to day use. It doesn't crash when I play my games or burns discs or during various other activities that may or may not actually stress the CPU or expose any underlying instability. However, my system FEELS stable to me, and that is GOOD ENOUGH for me."

Who are you more likely to listen to?

I can download Prime95, Intel Burn Test and whatever other tools he used, for free, to recreate the testing conditions of 1). to determine if his results are indeed applicable to my system. He can say, objectively, that his overclock has resulted in a stable, higher performing system. And his tests, that are specifically tailored to stress areas that apply to the overclock, can verify this to a convincing degree (though not an absolute degree).

I cannot download and recreate the 'feelings' of 2)., I cannot download and recreate what is 'good enough' for 2).. Because subjectivity doesn't work like that. He can only say that he 'feels' his overclock has resulted in a stable, higher performing system. Yes, he said he played a bunch of games and burned some discs, but what if I don't want to spend money to buy those same games? What if I don't have blank discs to burn with? How does any of that even matter when those activities do not necessarily apply to the overclock?

Now, if you DON'T contribute in this manner, when everything I said DOES NOT APPLY TO YOU. However, I would perhaps question what you're doing here to begin with if that's the case.

If you DO contribute in this manner, you're socially obligated to provide information that is as objective and as accurate as possible. Your 'feelings' be damned.
 
Im sorry but no 2 settings will be the same from 2 different setups with the same components. Overclocking is all about trial and error. You can find a good starting point, but thats about it. I like stress testing because it's very repeatable and makes it alot easier to narrow down problems. Thats my personal opionion, this subject will allways be of debate and is only useful for an oc database on a forum.
 
I think the point being made here is not that there is some sort of absolute measure of stability that everyone should be adhering to. Your system being 'stable enough' for you is perfectly acceptable....for you. I think the problem some people are having with this when you come to a forum dedicated to maximizing the performance of your system (what this forum is dedicated to) and, because your system is 'stable enough' for you without any kind of objective verification to back it up, you feel its acceptable to participate in its discussion.

A forum like this exists to exchange and provide information. In order to maintain any kind of credibility, that information needs to be as objective and accurate as we can possibly make it. Dispensing information based on what you 'feel' or what's 'good enough for me' without any kind of objective basis behind it is dishonest and possibly destructive to the people who don't know any better. Asking 'why do you care if I say my system is stable when it really isn't?' just demonstrates the kind of behavior that doesn't belong in that kind of environment. Its not a matter of whether its stable enough for you or me, its a matter of 'is your system stable enough that you can honestly advise someone to use the same settings as you'?

Yes, everyone's system is different. What is stable for one person will not necessarily work for the other. But taking that attitude and using it as a basis to completely discard objective testing renders this entire forum meaningless.

"Yeah, I could help you with your overclock, but since my system is not 100% exactly like yours its not guaranteed that my advice will work for you. So I won't bother."

If that's what you believe. Don't post on this forum. Don't help other people on Overclocking & Cooling. If, on the other hand, you are interested in helping and contributing despite that, which do you think will look better in the eyes of a newbie whose looking for some good advice?

1). "I've overclocked my CPU to <insert_awesome_speed_here>. My system is not exactly like yours, so I can't guarantee this will work for you. However, I have run several objective stress tests that you can download and recreate on your own system. My overclocked system has passed these stress tests successfully. So, based on reliable and repeatable information (that disregards my own personal feelings on the matter), I can say these settings have a GOOD CHANCE of working for you."

2). "I've overclocked my CPU to <insert_awesome_speed_here>. My system is not exactly like yours, so I can't guarantee this will work for you. However, my system has performed well in day to day use. It doesn't crash when I play my games or burns discs or during various other activities that may or may not actually stress the CPU or expose any underlying instability. However, my system FEELS stable to me, and that is GOOD ENOUGH for me."

Who are you more likely to listen to?

I can download Prime95, Intel Burn Test and whatever other tools he used, for free, to recreate the testing conditions of 1). to determine if his results are indeed applicable to my system. He can say, objectively, that his overclock has resulted in a stable, higher performing system. And his tests, that are specifically tailored to stress areas that apply to the overclock, can verify this to a convincing degree (though not an absolute degree).

I cannot download and recreate the 'feelings' of 2)., I cannot download and recreate what is 'good enough' for 2).. Because subjectivity doesn't work like that. He can only say that he 'feels' his overclock has resulted in a stable, higher performing system. Yes, he said he played a bunch of games and burned some discs, but what if I don't want to spend money to buy those same games? What if I don't have blank discs to burn with? How does any of that even matter when those activities do not necessarily apply to the overclock?

Now, if you DON'T contribute in this manner, when everything I said DOES NOT APPLY TO YOU. However, I would perhaps question what you're doing here to begin with if that's the case.

If you DO contribute in this manner, you're socially obligated to provide information that is as objective and as accurate as possible. Your 'feelings' be damned.

Well said.
 
I agree with both points discussed on this thread, although I do lean towards BP's post a bit more.

For a beginner, just stepping into the world of overclocking, it is imperative that he or she sticks to a rigid set of stress testing guidelines, such as what utilities to use and how long to run them for, to ensure that the overclock is "stable"(by community's general standards). Since prime95 and IBT are considered as the community's best tools to gauge stability of CPU overclocks, these tools should be used and recommended as such by more experienced overclockers.

However, such rigid set of rules (or even specific stress testing programs) are not always necessary once a person acquires a certain level of overclocking experience. Many come to this realization as have I. You should still use the above tools, but the length of time or passes do not need to be an exact science nor do you need to run them both or at the same time as some have suggested (on a different thread). What you run and for how long you run them for should be completely up to the user as dictated by the user's experience.

It does not matter if you believe that running small fft in prime95 for 72 hours vs 3 hours is better. If your experience has taught you that running 72 hours of prime95 is a must, then that is what works for you. On the other hand, if your experience has taught you that running only 3 hours of prime95 is enough and such testing method has yet to fail you, you are good to go.

You may even be one of those rare individuals that runs some program that no one else runs for stability testing. Again, if such method has worked for you and your system has always proven to be stable for any real application that you use it thereafter, then no one can deny that you have a stable system.
 
Back
Top