I've got a G4/400 with 1.2 gigs of ram at present.
When I first started the Seti challenge, this machine kicked butt and took names. I could do 4.5-6 hour
block times regularly. Every time. It wasn't surprising to sometimes see 3-4 hour blocks. But it
wasn't to last. People were running through blocks faster than the Seti folks could supply them and
they had to think about what they could do to increase processing time while adding value (science) to
the equation.
Then they added lots of science to the client which made things much worse in terms of pure processing
time. In fact, there was one time under the graphic client under OS X (I wasn't aware of the graphic client at the time) where a block had taken 16 hours and was still not done.
Last I checked under the text client under X with every speed trick I have in the book employed I was
getting 10-12 hour blocks, but some were much longer.
Now my main rendering machine for Seti is an 800 mhz PIII machine because the advantage in Seti 3.0x
and later is pure mhz speed whereas before if your L2 cache was large you'd win out over those who
were faster but had a smaller cache.
Now I use the G4 for the RC5 challenge and it does exceed there. Under OS 9 I'm getting just under 3.5
million kkeys per second and under OSX about 3.2 consistently.
You might also look at the distributed.net site as they have a really cool comparison of CPUs and speeds
and you can see how your speed will stack up versus other users. It doesn't help much on the Seti side
of things but there you go.
Interesting, when I had 4 Macs going I was getting about 12-15 Seti work units a day. Under RC5 of
course it's in the thousands but what can you do?
Top
