Break/Burn in procedure for new Athlon64..

[AmeenR]

Going to start assembling my A64 3000/ASUS K8V deluxe system and was wondering if any certain steps should be taken to ensure a proper running rig? Is break-in even necessary? Please give some input!!! Thanks in advance.

 

[Hans5849]

don't, ive read that buring it in can make it slower in the long run.

 

[myocardia]

Originally posted by: Hans5849
don't, ive read that buring it in can make it slower in the long run.

Huh? Anyway, the answer to your question is that break-in isn't even necessary. Also, just about any question you would need to ask has already been answered, or can be answered, in this thread:
link to the Anandtech K8V thread.

 

[tcsenter]

There is no "break in" in the sense it applies to piston rings and rod bearings.

"Burn in" is a procedure used to weed-out marginal, sub-standard, or defective components by attempting to accelerate a failure or unstable condition that would likely have occured in time. By thrashing the dickens out of it under heavy loads, a marginal component or subsystem can be pressured to reveal its weakness much sooner than it would have under 'normal' use.

For CPU, cache, motherboard, chipset, HDD, and RAM subsystem burn-in, Hot CPU Tester 4 Pro by 7Byte is a bargain @ $20 and supports A64 (in 32 bit mode).

Other reasonably affordable utilities include:

PassMark BurnInTest V4.0
SiSoft Sandra Professional (Burn-in Mode)

There are others, but I suspect you would prefer not to fork out hundreds of dollars.

Benchmarking applications are not burn-in or stability testing applications, though benchmarks can certainly be used for that purpose, they are not specifically designed for it and are inferior to true burn-in applications.

If you want to go the cheap (or free route), there are many shareware or freeware programs to test individual subsystems such as CPU/Cache, RAM, video, or HDD, but few are comprehensive and none are used in commerical or professional circles.

 

[Soulkeeper]

i usually grab a 2 litre of root beer or a case of beer (depending on the mood i guess) and get set for a few hours of gaming to check stability

i would reevaluate any speeds/voltages/temps/timings/etc at the first sign of any instability (even just one crash in one day) and only take things one step at a time and only overclock one thing at a time.
No need to really burn anything in, but it just sounds cool

Have fun with the new rig

 

[AmeenR]

Cool guys...thats for all the replys! If I have any more questions, I definately know where to go.

 

[alexXx]

burning in a processor is a myth. THere is no scientific reason as to why burning in a processesor will do any good. Quite the contrary actually. It is not at all like an engine(as was said)

 

[WobbleWobble]

I think what you want to do is stress-test your system. Just run Prime95 for over 12 hours, Memtest86 for a few passes and 3D Mark 2003 looping overnight.

Burning-in is explained in the AT FAQs with respect to overclocking, but does explain the myth.

 

[Jaxidian]

Originally posted by: alexXx
burning in a processor is a myth. THere is no scientific reason as to why burning in a processesor will do any good. Quite the contrary actually. It is not at all like an engine(as was said)

Burning in your CPU will not change anything within your CPU but there could be changes with your thermal compound curing over time from the heat. The hotter it runs, the quicker it cures. In this sense, burning in may help you (although I still doubt it would affect stability or max OC speeds).

An additional benefit of burning in your system is to check for stability. When you burn it in properly, then your system should run at its maximum temperature that it will ever run at. If your computer runs fine when doing Windows stuff or surfing the web but crashes while you burn it in, then you know that something's wrong (too high overclock, too low voltage, not good enough cooling, etc.). So, again here, burning in can help you.

Personally, to burn something in, I first try to figure out the general range of all of my overclocks (CPU, RAM, and VGA). For CPU max overclock, I use Prime95. If it lasts 10 minutes without any errors, then I'm either good or really close to good. For RAM, I don't really have a specific way to test it other than general stability and RAM and VGA benchmarks (often, if you've OC'd too high, you'll crash when running RAM benchmarks or VGA benchmarks without your VGA card OC'd). For VGA, I run AquaMark and 3DMark 2001SE and 2003(?) to see how high I can go without crashing and without any artifacts. After I've done this, then I'm generally pretty close to my maximum overclock. At this point in time, I use 4-5 hours of non-stop gaming to do my final stability check. If it lasts for that long without any instability or graphic artifacts, then I usually consider it stable. Also, just as an additional check, I'll run Prim95 overnight just in case.

However, if you're not overclocking at all, then burning in is a waste of time.

Just my $0.02 worth.

 

[Jeff7181]

How bout instead of "burn in" we call it stability testing? That'll eliminate any confusion about what is meant.

 

[Degrador]

Just thought I'd mention that AMD would have already thrashed the crap out of the cpu to work out what rating it needs - all the cpus are coming out of the one line, the ones that can handle higher clock speeds are given the higher ratings and locked at the relative speed.