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

Burn In Period

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

Mustanley

Member
Joined
May 23, 2003
Just got a Barton 2500+ setup and running on Tuesday. So far, with a Vantec Aeroflow HSF, I getting stable speed at 210x11.5 (2415) @ 1.75 Vcore. This is on an Asus A7N8X Deluxe Rev. 2.

Question: I've heard several references to the Athlons running at higher speed after they've been burned in for a while. How long should I wait before trying to raise FSB or multiplier? MBM5 shows idle temps at 44 C and load temps at 48 C.
 

fiji

Member
Joined
Jul 14, 2002
Location
berlin
do a 24 hour burn in using something like prime95, then OC some mhz, then burn in some more


at each speed burn it in a little
 
OP
Mustanley

Mustanley

Member
Joined
May 23, 2003
Started it running prime95 last night, so it's been going for almost 24 hrs. I'll see if there's any more headroom later this eve. I'm plesantly suprised at how fast this thing is running already.
 

altec

polka dot ninja
Joined
Dec 23, 2002
Location
Doylestown, PA
You don't really need to "burn in" after each new Mhz gain, all you should run Prime for is like 4 hours to test stability. When you hit the roof on the chip, lower it back down to stock speed, up the vcore to like 1.9v, and "burn in" for 24-48hours. Go back and see if the chip will go any higher after this. Sometimes it works, and sometimes it doesn't. Good luck.
 

altec

polka dot ninja
Joined
Dec 23, 2002
Location
Doylestown, PA
modenaf1 said:
I have a burn in question. does folding work?

Folding stresses the CPU a very good deal, so I don't see why it wouldn't work. I think Prime is a little better for burning in though since it will yield a little higher temps than folding.
 
OP
Mustanley

Mustanley

Member
Joined
May 23, 2003
I looped sandra 2003 burn in test 50 times at 217x11.5 without issue. But Prime95 locks up immediatly at anything over 210x11.5.
Prime95 is obviously a better stressor.

modenaf1, thanks for the warm welcome:beer:
 

altec

polka dot ninja
Joined
Dec 23, 2002
Location
Doylestown, PA
Well since 3DMark tests the memory card mostly, and Prime tests the CPU number crunching, yes Prime is a better CPU stress test. Lower your overclock a little and see where you can run Prime stable for four hours or more, and then use the burn in method i described above.
 

modenaf1

Member
Joined
Mar 1, 2003
Location
the terran system
you could also set sisoft sandra to do the arithmetic benchmark over and over and over again. that works to right?


i think prime is the best though because everybody uses it and it works.
 
OP
Mustanley

Mustanley

Member
Joined
May 23, 2003
I ran it about 20 hrs contimously last night and today at 205x11.5.
that was 1.75 Vcore. I'll bump it to 1.85 (the max on this board) tonight and back down the multiplier to 166 and see what happens tomorrow. Thanks ;)
 

altec

polka dot ninja
Joined
Dec 23, 2002
Location
Doylestown, PA
The CPU is a number crunching machine. What Prime95 does is searches for insanely large prime numbers. When you run the Prime95 Stress test, it is using numbers that the program knows are prime or not, but is testing how the CPU is rounding the numbers and computing them. So it is pitting the CPU against known results thus making sure it makes no errors so that you can insure that your CPU is stable. Sandra doesn't test for stability though, doesn't put the CPU up against know results for number as large, and therefore Prime is a better stress test.
 

modenaf1

Member
Joined
Mar 1, 2003
Location
the terran system
thanx. my 1700 is still shipping, but when i get it ill have to do this as well. What voltage should i run it at? it is a DLT3C. as long as it is cool voltage dosent matter right? hypotheticaly, if i used my central AC as a super duper phase change, i could hit 3.0V if i wanted too?
 

altec

polka dot ninja
Joined
Dec 23, 2002
Location
Doylestown, PA
The voltage does matter, because as the CPU is doing more and more intensive work per cycle, the chip needs more juice to keep it from becoming instable. Thats why a lot of the time a simple vcore upping will give you Prime stability when it was previously unstable at a certain Mhz. Yes the cooling is definatley a huge factor. When a chip gets too hot, it will begin to give out and make errors just like if it doesn't have enough voltage.
 

altec

polka dot ninja
Joined
Dec 23, 2002
Location
Doylestown, PA
When you are burning in you want to have a stock speed, high voltage, and pretty low temps at full load. The idea behind burning in is that the electromigration that occurs at high voltages "breaks the CPU in" making it perform better, like a car's engine. A lot of people say that the only thing that happens to a CPU with use is that it gets worse, but a lot of people diagree. I burned my 1700 in after hitting a wall at 2450Mhz, and afterwards I hit 2.5 with ease (and a lot of heat, hence my need for watercooling). So do it at your own risk.