Le dimanche 11 Mars 2007 13:15, Andrew Fildes a écrit :
> Unless of course they produced a 1.33GHz machine and then
> progressively disabled it downwards to produce a range and a sequence
> of machines of varying speeds, which was what I suspected.
> Sorry it was 800 to 1.33 (66%) - details here, scroll down a bit -
> http://www.lbodnar.dsl.pipex.com/eMac/eMac-upgrade.html
> #2 son did it with no problems at all.
I concur, that doesn't seem that far fetched. It's been well known for some
times that CPU manufacturers have become so good that there isn't any real
difference between cores of low and high frequency models, just marketing
range placement. For instance, AMD athlons are notorious to cope well with
wild overclocking. PowerPC G4 being risc cores too (as are all CPUs since 10
years), they should do equally well.
Rule of thumb : pushing any bottom of line CPU up to the max frequency of the
exact same part model should work perfectly. After that point, though, all
bets are open.
The 10~20% limit is a reminder of pre-1995 times when manufacturers were
building their frequency lineup by testing individually each CPU package up
to their locking point, then throtling frequency down 20% for safety margin
and labeling the chip accordingly.
Exception (very uncommon) : don't do that with an alpha CPU. They are so
optimized (hand laid core design !) that they run very close to their limit.
--
Manuel Viet
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|