Tony Sivori wrote:

So I of course decided to put the resurrected box to work crunching for
seti.   :-)    This makes my third seti box; box number two is a 650E
(meaning 100 MHz front side bus) PIII. The PIII, when not overclocked,
tends to average about 10 1/2  hours per work unit.

I've only crunched a few work units on the 466, but it is averaging 14
hours per work unit. Considering the Celeron has only 71% of the clock
speed of the 650, has half the cache, runs on a 66 MHz front side bus, 
and is based on the PII, I was expecting it to take a lot longer.

I'm not surprised at the performance.

The P3 is not really a faster processor than the P2 except for clock
speed, cache, and SSE. Internally, the two cores are very similar. If
you set the P3 to run at 466MHz, I'm sure the floating point crunching
times would be very close. The extra cache and SSE instructions would
help the P3 in certain applications that are compiled for SSE, but
that's about it.

The same applies to the P4 basically...it's not any faster than the same
clock speed P3, in fact it's slower due to the longer pipeline. But it
ramps well with higher clock speeds than the P3, and that's why it's a
better processor.

I don't know if you remember, but the P3 clock speeds more or less took
over where the P2 left off (about 450MHz I think). The P4 was basically
the same, giving us more power by ramping up the clock speed at the
core, allowing the better buffering system to pump data faster.

Of course other factors improve overall performance as well, such as
faster memory, faster buses, faster HDD interfaces, and faster video
cards. And that's why I generally don't upgrade until I can get at least
a doubling in performance. I don't find it satisfying to buy a new
computer that "feels" a little faster, I want one that is noticeably
faster.