CPU "Burn In" -- necessary or Not?

[LuckMan212]

I am not an EE or anything but common sense would dictate that "burning in" a piece of silicon is unnecessary.

The reason I am asking is, I recently got 5 SL6WUs (3.0Cs) and not a single one of them was a good clocker. One of them even degraded quite strangely from 3.85 all the way down to ~3.3 in a matter of minutes, at "normal" OC voltages of 1.65~1.675v.

Now, I tested all of these by putting them in and trying to boot immediately at 250fsb (3.75Ghz). From there, I used ClockGen to see if I could go higher. I have heard some people say that they run their CPUs at stock voltage and clockspeed for 24hrs of Prime95 before attempting an OC. Not only is this very time consuming, but I am not sure this actually gains anything. BUT after my dismal results (see THIS post) I am wondering-- is there some merit to this testing method? Have I "ruined" the OCability of these CPUs by clocking them right away? They all still work fine at stock voltage so nothing seemes "damaged" but I am just curious what people's thoughts are in general on burn-in.

Thanks.

 

[Jimbob7]

Some people run 24hour prime before they overclock to make sure that there is no stablity issues. Another thing done is to overclock as high as you can on stock vcore, and prime for 24hours as to not let them get used to running high vcore.

 

[I.M.O.G.]

Necessary? No.

Possibly help? Yes.

Running the chips before you burn them in does not "destroy the OCability" of the chips. You can still burn in after first running them.

The best way to burn in is to lower the FSB so that the chip will run as cool as possible while raising the Vcore to ABOVE stock and running PRIME95 for 24 hours. The point of burning in is running with a high vcore with low heat output - that is how you are supposed to go about getting the alleged benefit that may come from burning-in.

A lot of people don't understand or know about the proper way to burn in... Common misconceptions are that the chip should be run hot, or that you just need to run prime at stock for 24 hours before anything else. Both those are incorrect. Also, often times people confuse burning in a CPU with another meaning of burn-in. When you get new parts and are testing them under load for a period of time that is also called "burning them in" sometimes - but its just testing them under load to ensure stability. CPU burn in has different intentions than that.

The intent of burning in is to raise the vcore, and cause the CPU to draw a lot of power (prime95), while keeping the temperature as low as possible (thats why you lower the FSB). It has nothing to do with finding stability of the chip at this point.

There are a lot of threads on burning in asking this same question... If you want you can search for them to get more opinions. This isn't new or something that ever changes really.

I'm searching for old threads right now, and I will link you if I come across any good ones.

 

[slawson]

If 24 hrs (or more) at stock voltage is good then wouldn't 48 hrs under volted be better?

 

[I.M.O.G.]

No, read my post above, 24 hours at stock is not burning in the chip. That is the wrong way to burn in.

 

[LuckMan212]

Originally posted by IMOG
The best way to burn in is to lower the FSB so that the chip will run as cool as possible while raising the Vcore to ABOVE stock and running PRIME95 for 24 hours

Do you mean lower the FSB to BELOW stock (ie, <200) ?? Or simply, slightly OC'ed (215-220 for example) and how high VCORE are we talking, 1.6, 1.650, 1.7, etc? Is it different for D1 steppings vs. M0?

There are a lot of threads on burning in asking this same question [...] I'm searching for old threads right now, and I will link you if I come across any good ones.

Ok, please do, thanks!

 

[I.M.O.G.]

I mean run it at considerably under stock. Not sure how high of Vcore is right for Intel, something above stock should be used though... For AMD I would just put my mobo to its highest setting and check temps - this might not be a good idea for everyone though. Go with what you are comfortable with if you want to burn in.

I wouldn't worry about it if I were you... like I said burning in isn't required and it isn't exactly proven IMO that it works. A lot of people report that it helps their overclock become stable at a lower voltage, while others say it made no difference. That doesn't mean much though because theres no way of knowing who's doing what the right way, or who's testing thoroughly enough before and after, if you know what I mean.

EDIT: I hadn't found any links that were very good when I first searched. If I search again I will link you.

 

[Fuzion]

This is the method I use:

I find my max OC on the default vcore and run my system at that speed for about 3-5 days. During that time, I'm not running all Prime95 or any burning program. I just use the comp like I normally would. After about 3-5 days, I up the vcore and find my max stable OC.

I've done this on 3 out of my current 5 P4s, and I've noticed better overall results when doing this. I'm not sure if it will work for everyone, but I know it works for me.

 

[Lancelot]

I usually find the sweet-spot at default Vcore and run the system for a few days at that speed using both Prime95 and SETI. After that I start bumping Vcore and FSB until I find a reasonable OC of my likings and run it like that for a few days. After some time you'll find you're able to lower the Vcore a notch or two at that speed while keeping Prime95 stability...

 

[I.M.O.G.]

FOUND IT! This thread explains both sides of the burn in issue, and it is well worth the read:

http://www.ocforums.com/showthread.php?s=&threadid=245696&highlight=burnin+how+to

This is the other really good burn in thread I knew about:

http://www.ocforums.com/showthread.php?s=&threadid=237004&highlight=burnin+how+to

Sorry, these took me quite a while to find. Usually don't have much trouble. You can see my search terms highlighted in red in those links.

 

[LuckMan212]

IMOG--

thanks a mill for digging these up. Very interesting reading!!! For sure, on my next batch of 3.0C's I will be following a more rigorous burn in procedure to try and squeeze more Juice out of my CPUs! Whether or not it helps-- who knows but I guess It "can't hurt"

I will post back here with the results.

thanks for your help.

 

[I.M.O.G.]

No problem glad you found them useful, and glad I now have a thread in which I know where to find those threads.