What's this thing about burning in a new motherboard?

[Felecha]

First time replacing a motherboard. When my current rig was built at a local shop 5 years ago (ABIT KT7RAID, Athlon TBird 900) the guy told me when it was delivered that he had "Burned it in", something about running hours of a stress program before delivering it to me.

Replacing it with an ASRock 939Dual-SATA2 and an AMD Athlon 64 X2.

From what I've found about it on the web, such a thing is more to verify that everything is good (useful before delivering to a customer) than to "break it in" like you might do with a new car.

Is that correct?

 

[evilharp]

There are 2 types of burn in:

1) Stability tests: make sure everything works as it should by running a series of utilities on the system.

2) Thermal burn in: heating and cooling TIM to optimize its efficiency.

System burn-in can involve both types, but most (reputable) white-box builders only concern themselves with #1. They test the system to make sure it works, and it was assembled completely, before the customer picks it up. This helps to reduce support, RMA and warranty headaches later on.

As an aside, I've seen major OEM boxes ship with critical parts missing (IDE cables, HSF, etc..) Clearly, not everyone believes in burn-in or quality control.

 

[Davegod]

You dont need to run any kind of "burn in". It can be prudent to run some stability tests on your new build, since its better to find out any problems early on while returning items is relatively easy and before you've got all your important files on there. Just leaving the computer on for several hours is perfectly fine for a good "burn in" to melt and settle a thermal pad, but just installing windows and maybe running a 3d or pc mark should be sufficient for non-overclockers. If you've got some arsey motherboard or premium ram it might be worth running memtest for a while, but if you're using SPD (stock) timings it really should be fine.

 

[Felecha]

Thanks

 

[tcsenter]

Burn-in = infant mortality testing. Trying to encourage any marginal components to fail before assembly or packaging. Burn-in testing is done with hardware based or specialized equipment under well controlled conditions (temperature, humidity, air pressure, higher frequencies, etc.), so that post-mortem analysis can be performed accurately. Also done for statistical reliability, manufacturing QA, design, and cost analysis exercises (e.g. MTBF and MTTF).

Stability testing = stability testing (self-explanatory). Loading a finished system or subsystem to the max using software methods to root-out any potential flaws that may not readily appear under less demanding use. Can also promote or lead to infant mortality, but is not the same thing as a burn-in procedure. Stability testing can only roughly suggest probability, whereas burn-in can definitively measure it (given a known set of variables).

Ok, I'm guilty of using burn-in and stability testing interchangibly, even though I know the difference.