Once again... I kinda have to *stress* this. Burning a CPU in like this takes a long LONG time. This really should only be for the guys who would kill for another 100mhz.
Im glad that some of you are seeing good results with this. I too am seeing some benefit but as I said earlier, it takes awhile.
As for the sticky, Ill do one later today. There are alto of skeptics on this subject, so I want to see that this is helping more people, and a higher level of consistancy, before I try to get it stickied
felinusz said:
My chip has seen a total burn-in time of 18 hours now, at 2700 MHz, 1.85V.
I'm up to 4 minutes of Prime95, from the 2 minute wall where I started at this speed and voltage.
I'll add that this is at the same room temperature, so it likely isn't an environmental variable. I've also been testing the Prime95 run-limit by shutting my rig down, leaving it for an hour, and then rebooting, and running Prime. Consistantly, I've been getting 4 minutes out of it now, where I got 2 minutes consistantly before the burning in.
So, it looks like a tangible gain, albeit an extremely small one.
I hate to say it, but thats all you are gonna see from the first few cycles
It generally takes me a week/month of this to see very noticable gains (50-100mhz)
One again, some are more responsive than others, so your results will very significantally. Also you must remember this works exponentially. I bet that if you prime it at your current speed / voltage, prime will then run @ 8 mins plus. It all depends on HOW unstable the CPU is. If you are really pushing it, you probably wont notice much change initally. However one you get it to a semi-stable speed you will see a great deal of benefit, since it works exponentially.
I personally prefer to push it as hard as I can then work my way from there. From my personal experience, doing this usually doubles stability. If you get yours to run @ 2-5mins, after the cycle it should run from 4-10 mins. Just enough to prove that it is working, which is why I suggested 2-5mins from the start
The greatest example is my old 2600+ and my new 3200+. My old AXP was very responsive to this. I shaved off a good .075v off of it, where as my A64 has gained maybe 30mhz thus far. Granted the 2600+ saw a great deal more of action than this one has, but the point remains the same.
However I have seen a trend. Good CPUs seem to benefit from this a hellova lot. Where as ****tier CPUs hadrly benefit at all
From ap personal perspective, stability tends to double or triple wiht each burn in (only successful ones, not each burn is) Your results fall in line almost exactally like mine. The reason why I suggested everyone to use 2-4 mins is that it is enough to show that stability has increased. If people decided to use 0-1, the stability gain would be so small it would appear as tho nothing had happened.
When you start getting into the 10mins stable range, that is where the most dramatic difference will occur (however it takes a long *** time to see such a difference since each prime now should last almost a half-hour)