How do you burn-in your processor?

[n00bien00blar]

Hi

I've just gotten my P42.26B CPU and I was wondering how you burn it in? My understanding is that burning it in allows higher overclocks at lower voltages. Is this correct, because that is what I want becasue I want to keep the chip for quie a while so I don't want to have to increase the voltage too much while overclocking. An what about the RAM and video card? My Corsair XMS3000 RAM isn't running stable @ 375MHZ like it should, even at a CL of 2.5, which im not happy about (it crashes in both prime95 and 3DMark within 10secs). Would burning it in help this (its not my CPU because it runs fine when the ratio is set to 1:1)?

Thanks for your help

 

[Mikki]

The general concensus is that the whole burn in thing is crapola, but I'll let someone else answer that. I just let my computer run for a day or so, using normal applications, before I start overclocking it. That's just me though.

 

[kimber]

loop 3dmark for a few hours.

 

[mastertech01]

Go to ZDNET and download a program called HOT CPU Tester.. I think the burnin issue may be a little flawed at best, but this program WILL test the stability of your system at a given setting. Run it overnight and if you are still running in the morning you should be good.

 

[Mikki]

mastertech01, how does that compare to Prime95 (which is the stability test program standard)? Also, n00bien00blar was asking about burn in, if that program isn't really good for that, can you suggest one that is? And do you have any thoughts on whether burn in is necessary or not? TIA

 

[gururu]

Burning in is thought to optimize a processor for a given voltage/clock. This is done by running the processor under
load for an extended period. Supposedly, a permanent physical effect (expansion, conductivity) on the metal circuitry occurs. Although I am not sure about the theory, I tend to do it myself. Some people even think that once burnt in at the highest stable clock speed , even higher clock speeds may be accessible after. Burning in might involve the motherboard and memory as well. Is the concept sound? Beats me.

 

[mastertech01]

Originally posted by: Mikki
mastertech01, how does that compare to Prime95 (which is the stability test program standard)? Also, n00bien00blar was asking about burn in, if that program isn't really good for that, can you suggest one that is? And do you have any thoughts on whether burn in is necessary or not? TIA

Ive never used Prime95 personally. I did download the proggy I mentioned last night as I am doing a dually MP processor overclocking project as we speak. I had to do my overclocking by multiplier using Golden Sockets and the voltage mod to the processor(closing the open bridge) and needed something that would put a heavy load on the processor overnight. This program will put such a load on the cpu,ram, etc. that you cant even navigate because it is a TRUE 100 percent load. I ran it overnight and my processors held up to the torture.

I dont have enough proof either way whether it will improve overclockability to burn in, but I have heard many swear that it will. I believe a processor has limits and no matter how much you load it,it wont gain much if anything. But I do believe you should give the processors a heavy workout to ensure stability. And since my setup requires removing the processors to make multiplier settings, I dont want to find out I have a stability issue after everything is all put back together a few days down the road. IMHO.

 

[Wingznut]

I started to write a reply, but then I couldn't figure out why I should go through the effort when I can just copy Patrick's post.

Originally posted by: pm
"Burn in" as described here doesn't do anything. There is no chemical, physical or quantum effect that would allow CMOS semiconductor components to run faster after having been run at elevated voltages or temperatures for a period of time. Burn-in is performed at manufacturers as a form of accelerated stress testing to weed out marginal parts. The process of burn in actually results in an overall reduction in speed due to nMOS hot-e and pMOS BTI effects introducing charge defects into the silicon dioxide of the gate during the burn-in process. So it doesn't speed up silicon, it actually slows it down. Manufacturers live with the hit in frequency performance in return for reduced customer returns.

Patrick Mahoney
Microprocessor Design Engineer
Intel Corp.

 

[TheSentinel]

As mentioned here and elsewhere, "burn-in" is a myth, an urban legend...

Now if you mean stability testing, well, that's a different story.

 

[gururu]

This is a great article about how to burn in your processor and the physics behind it

http://www.dwpg.com/print.php?artid=89

gururu

 

[jiffylube1024]

Here's my favourite way to burn-in : pick your favourite 3d game and just play it for a few hours. If required, repeat a couple of times.

 

[newbiepcuser]

Here's my favourite way to burn-in : pick your favourite 3d game and just play it for a few hours. If required, repeat a couple of times

thats my favorite way too

 

[newbiepcuser]

Here's my favourite way to burn-in : pick your favourite 3d game and just play it for a few hours. If required, repeat a couple of times

thats my favorite way too

 

[zephyrprime]

loop 3dmark for a few hours.

That wouldn't do anything but test your video card.

 

[JC]

Originally posted by: TheSentinel
As mentioned here and elsewhere, "burn-in" is a myth, an urban legend...

I dunno....my old K6-2/500 was marginally stable at stock speed/voltage, but after a good burn-in at 550/increased vcore, it would actually run stable at under stock voltage...I used good 'ol Final Reality for the burn.

I'm pretty sure burning-in doesn't really help with the newer, smaller process CPUs, nor such a good idea.

m2c

JC

 

[zephyrprime]

I just tried out "HOT CPU Tester" from ZDnet and it seems no good at all. It's another one of those testing programs that use math functions to test stuff and I don't think such techniques are nearly as good as test designed to test instructions and generate as much heat as possible. My CPU only heated up a little with Hot CPU but it heated up much more with CPUBurn.