-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
burn in?
- Thread starter saytan
- Start date
- Joined
- Jul 9, 2001
- Location
- Louisville, KY
...yet another example of a sucess after a burn-in and for those who haven't seen the article yet http://people.freenet.de/s.urfer/conditioning.htm#what
Have you stressed the CPU at all with prime95, sandra, SETi etc...
Have you stressed the CPU at all with prime95, sandra, SETi etc...
- Joined
- Mar 4, 2002
- Location
- Cincinnati, OH
Great! Congrats on the OC. 
deez - I have to disagree on that link. IMHO, that's rubbish. All of the graphs and discussion do nothing to scientifically support the theory.
Sometime ago there used to be an Intel CPU Engineer who frequented the OC/CPU forums on AnandTech. This is a guy that actually worked on the design team for Intel CPU's, including the P3 and P4. According to him, CPU's don't burn-in....contrary to the popular OCer's folklore. There was many lenthy dicussions of the topic. I have to believe him over anyone else unless you can supply a link to a credible source with actual scientific proof.
I know all of us overclockers have seen an apparent CPU "burn-in". I think the more appropriate theory should be "system burn-in". That is, all of the components "meshing" together in some mysterious fashion to allow higher clock speeds.
deez - I have to disagree on that link. IMHO, that's rubbish. All of the graphs and discussion do nothing to scientifically support the theory.
Sometime ago there used to be an Intel CPU Engineer who frequented the OC/CPU forums on AnandTech. This is a guy that actually worked on the design team for Intel CPU's, including the P3 and P4. According to him, CPU's don't burn-in....contrary to the popular OCer's folklore. There was many lenthy dicussions of the topic. I have to believe him over anyone else unless you can supply a link to a credible source with actual scientific proof.
I know all of us overclockers have seen an apparent CPU "burn-in". I think the more appropriate theory should be "system burn-in". That is, all of the components "meshing" together in some mysterious fashion to allow higher clock speeds.
- Joined
- Jul 9, 2001
- Location
- Louisville, KY
BadThad...yep that has been a topic of discussion here a bit too.
And I say voodoo majic or not cpu burn-in has worked for me.
I do agree that a system burn-in takes place during the first 0-48 hours though b/c faulty components will generally turn up at that time or within the first 1-2 months of normal use.
And I say voodoo majic or not cpu burn-in has worked for me.
I do agree that a system burn-in takes place during the first 0-48 hours though b/c faulty components will generally turn up at that time or within the first 1-2 months of normal use.
- Joined
- Mar 7, 2002
I disagree with that.
I have a celeron 1.0A on an abit ST6. When I first got it, it took 1.65V to be stable at 10x133 ([email protected] hours), with the retail heatsink.
After I'd had it for a week, my AX-7 came in the mail. The first thing I did after I installed that HSF, was drop my voltage to 1.6 and try prime. It failed in 10 minutes, so I pushed it back up to 1.65V, and left it there. It ran fine, for a little over a week, with something always running in the background (prime95, cpu stability test, burnp6, etc). Just tonight, I noticed my temps weren't as high as they were a week ago (but my case air temp was the same 30C it's always been). It kinda reminded me of some burn-in stories I had written off as mumbo-jumbo a long time ago. Anyway, I dropped my voltage a full .1V (to 1.55V), and began to run prime. so far, it has run for 1hr:33min. If prime is going to fail on me, it has always done it in the first 30 minutes. If it's still running when I wake up in the morning, I do believe I'm going to have to become a believer in this burn-in "voodoo-magic"
Is there anything else I should run to test my cpu for stability? I usually consider it stable after it has run prim95, burnp6, and cpu stability test (each individually) for 12 hours. Is that good enough, or should I do some more tests?
UPDATE: I may have spoken too soon...prime failed at 1 hour, 51 minutes...I'm gonna give the voltage a .025 bump, and see if that works.
I have a celeron 1.0A on an abit ST6. When I first got it, it took 1.65V to be stable at 10x133 ([email protected] hours), with the retail heatsink.
After I'd had it for a week, my AX-7 came in the mail. The first thing I did after I installed that HSF, was drop my voltage to 1.6 and try prime. It failed in 10 minutes, so I pushed it back up to 1.65V, and left it there. It ran fine, for a little over a week, with something always running in the background (prime95, cpu stability test, burnp6, etc). Just tonight, I noticed my temps weren't as high as they were a week ago (but my case air temp was the same 30C it's always been). It kinda reminded me of some burn-in stories I had written off as mumbo-jumbo a long time ago. Anyway, I dropped my voltage a full .1V (to 1.55V), and began to run prime. so far, it has run for 1hr:33min. If prime is going to fail on me, it has always done it in the first 30 minutes. If it's still running when I wake up in the morning, I do believe I'm going to have to become a believer in this burn-in "voodoo-magic"
Is there anything else I should run to test my cpu for stability? I usually consider it stable after it has run prim95, burnp6, and cpu stability test (each individually) for 12 hours. Is that good enough, or should I do some more tests?
UPDATE: I may have spoken too soon...prime failed at 1 hour, 51 minutes...I'm gonna give the voltage a .025 bump, and see if that works.
Last edited:
- Joined
- Feb 25, 2002
- Location
- New Zealand
Anecdotally, it seems burning in has a wide body of support...
myself, i had an ancient tnt2m64 which i overclocked till it wouldnt go any further... after using this for gaming for a few months, to my total surprise, i was able to overclock it several more points on both the memory and gpu! To me this suggests some sot of effect with use. The physics of it may not be totally known, but it seems to work.
~ShiFtY out~
myself, i had an ancient tnt2m64 which i overclocked till it wouldnt go any further... after using this for gaming for a few months, to my total surprise, i was able to overclock it several more points on both the memory and gpu! To me this suggests some sot of effect with use. The physics of it may not be totally known, but it seems to work.
~ShiFtY out~
- Joined
- Mar 4, 2002
- Location
- Cincinnati, OH
Remember, I didn't say there was no such thing as component burn-in. However, I did say there is no such thing as a CPU or GPU burn-in to achieve high speeds. Again, this is based on a lengthy explaination from a REAL Intel CPU Engineer on the AnandTech forums. From what I recall, he said cpu's only get worse with use, definately NOT better. Electromigration is constant no matter ocing or not, it's the RATE of electromigration that gets you.
I have seen systems that run for awhile (days/weeks) that suddenly will overclock higher after the system "meshes" for an uncertain time. Not sure what causes this, my guess is the power supply, mainboard capacitors, resistors, etc. seem to operate "better" after they have been "broken in" a bit. I'm waiting to find an electronics expert that can scientifically explain the phenomena...actually, I work with one...I will ask him what he thinks.
I am a scientist, but for chemistry and not electronics/computers, this is just a hobby. I read a LOT of highly technical material everyday including research papers. The link to the article above is NOT scientific and, IMO, is total hogwash. Sorry to restate my point, but I do know quality, scientific research when I see it.
This is an interesting discussion. I would like to hear more from people out there so that we may all benefit from a greater understanding of "burn in".
I have seen systems that run for awhile (days/weeks) that suddenly will overclock higher after the system "meshes" for an uncertain time. Not sure what causes this, my guess is the power supply, mainboard capacitors, resistors, etc. seem to operate "better" after they have been "broken in" a bit. I'm waiting to find an electronics expert that can scientifically explain the phenomena...actually, I work with one...I will ask him what he thinks.
I am a scientist, but for chemistry and not electronics/computers, this is just a hobby. I read a LOT of highly technical material everyday including research papers. The link to the article above is NOT scientific and, IMO, is total hogwash. Sorry to restate my point, but I do know quality, scientific research when I see it.
This is an interesting discussion. I would like to hear more from people out there so that we may all benefit from a greater understanding of "burn in".
- Joined
- Mar 4, 2002
- Location
- Cincinnati, OH
OK, I talked to the eletronics expert I work with (Hi Giles if you're reading this! hehehe). He has an excellent theory on "burn in".
The capacitors used in electronics are mainly to smooth current to devices. Capacitors are chemically based and typically have a "coil" of aluminum inside them. They are made VERY cheap so they tend to be a bit inconsistant. As they warm, the bits inside the capacitor actually "move" and "settle in" to a position. The chemical "goo" inside tends to evenly distribute itself over time making the capacitor more efficient as smoothing the current.
Overclocking, as we all know, is VERY dependent on a quality supply of power. The slightest fluctuation in voltage will send the system to a halt. My theory is that "burn in" actually brings consistancy to all of the capacitors in the system, i.e mobo, power supply, etc., thus, overclocking ability IMPROVES with time.
There you go folks....THE BADTHAD THEORY OF OVERCLOCKING BURN IN: "Capacitor Burn In".
The capacitors used in electronics are mainly to smooth current to devices. Capacitors are chemically based and typically have a "coil" of aluminum inside them. They are made VERY cheap so they tend to be a bit inconsistant. As they warm, the bits inside the capacitor actually "move" and "settle in" to a position. The chemical "goo" inside tends to evenly distribute itself over time making the capacitor more efficient as smoothing the current.
Overclocking, as we all know, is VERY dependent on a quality supply of power. The slightest fluctuation in voltage will send the system to a halt. My theory is that "burn in" actually brings consistancy to all of the capacitors in the system, i.e mobo, power supply, etc., thus, overclocking ability IMPROVES with time.
There you go folks....THE BADTHAD THEORY OF OVERCLOCKING BURN IN: "Capacitor Burn In".
- Joined
- Jul 9, 2001
- Location
- Louisville, KY
I do agree that the article I posted is definitely not from a scientific journal or anything like that...it is simply the opinion of somebody based on some stuff they read. Kind of like the articles at OC.com are the opinion of Ed, Joe, Skip etc....
I do agree with the capacitor burn-in b/c I've heard or read of something like that. But how would you account for upgrading the CPU only and getting better results after running it for several days/weeks?
I do agree with the capacitor burn-in b/c I've heard or read of something like that. But how would you account for upgrading the CPU only and getting better results after running it for several days/weeks?
- Joined
- Mar 4, 2002
- Location
- Cincinnati, OH
I respect people opinions on technical issues. I'm just saying the facts he presents don't support his theory. Not trying to be hard on the author, I just think people believe too much of what they read on the web when it's convoluted by technical jargon.
Maybe the change in voltage with an only cpu change-out affects the stability of the capacitors or perhaps the cpu's resistance is different. Of course, all this is pure theory on my part. I don't have scientific proof either, LOL. This is simply brought up as a topic of discussion and I was hoping somebody more technical than I would perhaps have some insight.
It would benefit all of us to understand overclocking better. It would be awesome if someone took on the challenge of setting up a test with identical cpu's and mobo's....but the problem of varibility exists with electronics. No two cpu's nor mobo's seem to be exactly alike even if they come off the line one after another. Perhaps this will forever remain a question unanswered.
Does anyone know a person in the business of manufacturing capacitors? It would be interesting to have their perspective.
Maybe the change in voltage with an only cpu change-out affects the stability of the capacitors or perhaps the cpu's resistance is different. Of course, all this is pure theory on my part. I don't have scientific proof either, LOL. This is simply brought up as a topic of discussion and I was hoping somebody more technical than I would perhaps have some insight.
It would benefit all of us to understand overclocking better.
It would be awesome if someone took on the challenge of setting up a test with identical cpu's and mobo's....but the problem of varibility exists with electronics. No two cpu's nor mobo's seem to be exactly alike even if they come off the line one after another. Perhaps this will forever remain a question unanswered.Does anyone know a person in the business of manufacturing capacitors? It would be interesting to have their perspective.
- Joined
- Jul 9, 2001
- Location
- Louisville, KY
Well maybe some of that article is BS but I've followed it and got better results with my last 2 CPU's and here is what I did. I first checked the CPU at its default speed then attempted to overclock as far as possible increasing my voltage accordingly for stability.
The first CPU was a P3 1100E. I was able to run this CPU in the 1380's but b/c I had to keep my PCI speed low to avoid HDD corruption I ran at 1309. Before a burn-in I had to increase my vcore a bit to run stable at 1309 and after a burn-in using prime95 and sandra simultaneously for several days at 66FSB and 2.10V I was able to run at 1309 default voltage. Now the mobo and PSU was also new at the time so the capacitor theory may be the reason with this chip. Although I sold the chip and the guy I sold it too was also able to run at the same speed with default voltage.
The second chip (my current chip) is a cele 1.0A. I slapped this chip in the same mobo as above and was able to run at 1400 sucessfully 100% rock stable with my voltage at 1.675. After this I burned in for about 5 days at 66FSB with my voltage as high as possible also running prime95 and sandra and was able to run at 1430 at 1.675v. In order to hit 1450 stable I had to do the wire trick and later on I removed the IHS to drop temps a few degrees.
so maybe its only a few extra Mhz or a bit less voltage but thats my input
The first CPU was a P3 1100E. I was able to run this CPU in the 1380's but b/c I had to keep my PCI speed low to avoid HDD corruption I ran at 1309. Before a burn-in I had to increase my vcore a bit to run stable at 1309 and after a burn-in using prime95 and sandra simultaneously for several days at 66FSB and 2.10V I was able to run at 1309 default voltage. Now the mobo and PSU was also new at the time so the capacitor theory may be the reason with this chip. Although I sold the chip and the guy I sold it too was also able to run at the same speed with default voltage.
The second chip (my current chip) is a cele 1.0A. I slapped this chip in the same mobo as above and was able to run at 1400 sucessfully 100% rock stable with my voltage at 1.675. After this I burned in for about 5 days at 66FSB with my voltage as high as possible also running prime95 and sandra and was able to run at 1430 at 1.675v. In order to hit 1450 stable I had to do the wire trick and later on I removed the IHS to drop temps a few degrees.
so maybe its only a few extra Mhz or a bit less voltage but thats my input
- Joined
- Mar 4, 2002
- Location
- Cincinnati, OH
When you sold the P31100E perhaps it was going into a used system? Do you know if his system was new? That may explain why he could use the default voltage.
With you cel 1.0A were you able to do 1430 right away, i.e. before you burned it in?
Interesting stuff! Only a geek can appreciate, HAHAHA.
With you cel 1.0A were you able to do 1430 right away, i.e. before you burned it in?
Interesting stuff!
Only a geek can appreciate, HAHAHA.
- Joined
- Mar 4, 2002
- Location
- Cincinnati, OH
deez, BTW, that's a nice little cel system you have there bro.
- Joined
- Jul 9, 2001
- Location
- Louisville, KY
Thanks BadThad...
been out of town for a few days just getting back to the thread
Yes the 1100E was going into a used system, and No the celeron was not able to do 1430 stable initially. I either got lock-up or prime95 errors until after the burn-in. Which was totally uninterrupted for 5 days <---it was Mardi Gras down here in New Orleans so I didn't touch the thing at all from Friday night till Wed. night
been out of town for a few days just getting back to the thread
Yes the 1100E was going into a used system, and No the celeron was not able to do 1430 stable initially. I either got lock-up or prime95 errors until after the burn-in. Which was totally uninterrupted for 5 days <---it was Mardi Gras down here in New Orleans so I didn't touch the thing at all from Friday night till Wed. night
All i know is that i don't care if it's been scientifically proved or not.
There's not really much harm done in leaving your computer on overnight, as you'll be using it past those hours easily anyway.
If after running it a few nights can make you CPU run at a lower voltage/higher mhz, well thats a good thing. If it doesn't it's not exactly the end of the world either. Personally I'd go for dropping the voltage for long term reliability's sake.
There's not really much harm done in leaving your computer on overnight, as you'll be using it past those hours easily anyway.
If after running it a few nights can make you CPU run at a lower voltage/higher mhz, well thats a good thing. If it doesn't it's not exactly the end of the world either. Personally I'd go for dropping the voltage for long term reliability's sake.
- Joined
- Dec 15, 2001
well, as unscientific as this analogy may sound, it's the best I can do at the moment...
Burning in your processor is very similar to breaking in a baseball mitt. The more you squeeze it open and shut, the looser it gets...the looserit gets, the easier it is to catch a fly ball. Same with the CPU, to me anyways. It seems that when I use a processor out of the box, I can only go so high, with so much voltage. But after increasing the voltage and "burning" the chip in for a few days, I effectively "loosen" it up a bit, and I'm able to go with a higher FSB at default voltage (sometimes even less!).
I don't know why this happens, and it may never be explained fully. Intel and AMD certainly won't tell us, because the simple fact of the matter is that they don't want us to overclock their chips! So unless we find out on our own, we may never know. I do not believe in voodoo magic, but I do believe in results...which is what I have gotten out of every single burn-in I've ever done...to new chips in an old system (w/ the chip being the only upgrade).
People that have had negative results with burn-ins will go against it, people with positive results will be for it. I think everyone should try it, and form their own opinion of it, so I recommend it.
Burning in your processor is very similar to breaking in a baseball mitt. The more you squeeze it open and shut, the looser it gets...the looserit gets, the easier it is to catch a fly ball. Same with the CPU, to me anyways. It seems that when I use a processor out of the box, I can only go so high, with so much voltage. But after increasing the voltage and "burning" the chip in for a few days, I effectively "loosen" it up a bit, and I'm able to go with a higher FSB at default voltage (sometimes even less!).
I don't know why this happens, and it may never be explained fully. Intel and AMD certainly won't tell us, because the simple fact of the matter is that they don't want us to overclock their chips! So unless we find out on our own, we may never know. I do not believe in voodoo magic, but I do believe in results...which is what I have gotten out of every single burn-in I've ever done...to new chips in an old system (w/ the chip being the only upgrade).
People that have had negative results with burn-ins will go against it, people with positive results will be for it. I think everyone should try it, and form their own opinion of it, so I recommend it.
- Joined
- Jul 9, 2001
- Location
- Louisville, KY
Makaijin said:
Perhaps you misunderstand the "common" burn-in method. It is more than just leaving your computer on. You have to run your processor at full load. You can do this using a variety of programs...prime95, sandra burn-in, SETI, folding, 3D Mark 2001 etc....
Another thing some of us have done to get better results is to set the FSB as low as possible and set the voltage as high as possible during the burn-in
- Joined
- Jul 9, 2001
- Location
- Louisville, KY
one more chip I forgot to mention...my cel 566 would not do 850 until I put 1.85V into it and then after a few monbths I was able to drop this down to 1.75V completely stable
- Joined
- Mar 4, 2002
- Location
- Cincinnati, OH
Just had to add that ALL my systems are "burned-in". They all run 27/7 at the max speed crunching SETI.
I'm a firm believer of NEVER shutting a PC down as long as you have good power. If I could afford it, all 6 of my main PC's would be on a UPS. As it stands now, only 2 of them are but my power is pretty good...no problems in 5 years.
When a PC is shut off the available CPU power is going to waste IMO. Everyone's PC should be doing at least one of the many distributed computing projects available on the web. It's my sole reason for overclocking!
I'm a firm believer of NEVER shutting a PC down as long as you have good power. If I could afford it, all 6 of my main PC's would be on a UPS. As it stands now, only 2 of them are but my power is pretty good...no problems in 5 years.
When a PC is shut off the available CPU power is going to waste IMO. Everyone's PC should be doing at least one of the many distributed computing projects available on the web.
It's my sole reason for overclocking!
