[Ti] 400>500

[Paul Russell]

>Alas, I hope it's not that simple. The party-line has always been 
>that the chips are manufactured and tested at a certain speed. If 
>they pass at that speed they are marked as such. If they happen to 
>be able to be "overclocked" then so be it. I would assume that the 
>chip in my 400 was tested at 500 and didn't pass certain criteria. 
>Either that or they needed a certain number of "400" chips and to 
>fill that order "500" chips were employed.
>

You're right up to a point. It's not cost effective to test 
individual chips. though, so typically a small sample from each batch 
is tested. If the testing of the sample indicates that the entire 
batch may not be good at 500 MHz then the whole batch might get sold 
as 400 Mhz chips. However this is quite a conservative method and it 
may well be that say 90% of the chips are perfectly good at 500 Mhz 
or greater. Of course in the worst case scenario then perhaps only a 
small percentage might be good at 500 MHz, but then due to the nature 
of statistical distributions this would probably also indicate more 
serious problems, i.e. that the production process was in a pretty 
bad state. Like everything in life, it's a gamble - you need to do a 
quick bit of analysis before you do the over-clocking, e.g.

                                 BLOW UP YOUR MAC              MAC WORKS FINE

     DO THE OVERCLOCKING             -1000                          +10

     DON'T DO IT !!!             (shouldn't happen !)                 0


You also need a value of p, the probability of blowing up your Mac if 
you do the overclocking. Let's say that for a skilled electronics 
technician p = 0.001, while for an accident prone hobbyist it might 
be p = 0.5. In the case of the former it's probably worth taking the 
risk, in the case fo the latter: probably not. (But you knew that 
already ;-)).

Paul

--