Burn-in duration?

Z

zachusaf

Limp Gawd
Joined
Jan 25, 2003
Messages
166
Just recieved an x2 4400, and I've been running prime95 on both cores to "burn in" the processor. How long should I do this?
 
Don't bother, it does nothing...Burn in is used at the factory to weed out early failing chips, not to 'bed them in.'
 
I know it does nothing "extra". I think I didn't clearly express myself. When you're "burning in" or checking system(CPU) stability after an overclock, how long is generally among the masses considered to be rock stable? 8 hours? 24 hours?
 
There is no standard way, in industry you run the chips loaded past the maximum they will ever see in the field for either a set time or until you reach a certain number of fails, then ship the ones that don't fail. There's a pretty wide range of times and scenarios even within a company, but whatever time you run it for make sure it's worst case scenario.

I'll tell you what I do that a lot of other people don't, I run my cpu at it's overclock, my video card at it's overclock, run one case of superpi (or whatever other stress program you want) and run atitool so that not only are both cores at 100%, but the video card is also stressed at 100%, then I turn the cpu and case fans down 15% from whatever the lowest I plan to run them at and let it run. If it survives that for an hour or two, I know that's beyond the worst case scenario as it will never get that hot when I use it so it shouldn't have a problem.
 
prime em for 48 hrs, and watch the temps

if you sense anything is wtong, stop the test and post here

we rather that you be safe than sorry
 
I usually run Prime 95 for a minimum of 24 hours. Then I run a 3d bechmark program (such as 3dmark 2001) for 24 hours. Keep an eye on temperatures as 2gigs has said. If your approaching 55 deg C., stop what you're doing and think about better cooling.
 
i'd do this:

memtest for 3 hours per gb of ram
windows memtest for 3 hours per gb of ram
run superpi 32m 2-3 times
run prime95 small-fft for 6 hours
p95 blend for 6 hours
loop favorite gaming bench for 3-6 hours
copy >3gb of files while windows memtesting and run MD5 hashes to make sure data is intact

that should be pretty stable. checks the cpu, gpu, ram, and hd's.
 
Burn in is a myth. Data corruption is not; so, I run Memtest86 to check the integrity of the RAM, then Prime95 to check the CPU.
 
burn in on bh5 ram isnt a myth

saw it first hand 220 out of the box 240 two weeks later with the same settings.

I really want to not beleive in burn-in, but i cant argue with the results, damned if I can find a way to explain it scientificly though.
 
Well, if you start up your computer and the CPU burns up, that could be burn-in.

However, the quality of manufacturing these days is such that you can probably get by with some basic motherboard/memory tests for a couple hours or so, and call it good.
 
BigTaf said:
burn in on bh5 ram isnt a myth

saw it first hand 220 out of the box 240 two weeks later with the same settings.

I really want to not beleive in burn-in, but i cant argue with the results, damned if I can find a way to explain it scientificly though.
Click to expand...
Strangely enough I have observed the same thing with my 2 gigs of Corsair XMS ram. The stuff sucked harcore out of the box, but every few days it is able to take bump in speed. Now, I have not observed the same thing with my 2 gigs of Ballistix
 
I've actually used CPUBurn for more than 24 hours and I was able to lower my CPU voltage considerably afterwards...I was at around 1.5 and I was able to go to 1.40...maybe even lower...maybe it was something else that caused my problems...but I'm pretty sure the CPUBurn actually helped me out
 
As far as cpu's are concerned, 'burn in' is a myth in what people usually believe it to be (some kind of electron migration that somehow improves the chip).

However, burning in of the TIM joint between the cpu and heatsink can have quite an impact on overclocking. With cpu die sizes getting smaller and smaller and the heat load ever increasing, the TIM joint becomes an increasingly critical area. You can probably gain 50mhz sheerly from 'burning in' a freshly seated heatsink over a couple of weeks.
 
(cf)Eclipse said:
i'd do this:

memtest for 3 hours per gb of ram
windows memtest for 3 hours per gb of ram
run superpi 32m 2-3 times
run prime95 small-fft for 6 hours
p95 blend for 6 hours
loop favorite gaming bench for 3-6 hours
copy >3gb of files while windows memtesting and run MD5 hashes to make sure data is intact

that should be pretty stable. checks the cpu, gpu, ram, and hd's.
Click to expand...

I have had seemingly stable OC's fail primer95 small-fft's after 8.5 hours. So I would strongly suggest running at least 12 hours, and if you can run 24 hours that would be better.
 
spine said:
As far as cpu's are concerned, 'burn in' is a myth in what people usually believe it to be (some kind of electron migration that somehow improves the chip).

However, burning in of the TIM joint between the cpu and heatsink can have quite an impact on overclocking. With cpu die sizes getting smaller and smaller and the heat load ever increasing, the TIM joint becomes an increasingly critical area. You can probably gain 50mhz sheerly from 'burning in' a freshly seated heatsink over a couple of weeks.
Click to expand...

That's more or less what I was going to say. I was under the impression that the burn in process is more or less for the thermal compound, in my case AS5. I've read reports of load temps dropping 3 - 5C by running through a burn in process.

When I first set my new system up with my Big Typhoon and AS5 I ran an 8 hour large FFT prime95 session, turned the system off for about 8 hours, then ran another 9 hour large FFT session. I definately dropped a couple degrees, went from 45C load on installaion to 42 - 43C load after the process.

I just lapped and polished my base last night and reapplied AS5. Ran 9.5 hours of Large FFTs last night while I was sleeping, woke up to a 1C - 2C drop. It's off now for the day but I'm going to run another 10 hours tonight, I'll report back tomorrow morning with the results. Between lapping and the burn process, I'm hoping to be at 41C under load. Lapping alone with fresh AS5 took my idle down from 29 - 30C to 28C solid :D
 
You must log in or register to reply here.
Back
Top