posted on Feb, 16 2004 @ 08:03 AM
Hi EXR!!
Again, coming to this discussion late (I'm not usually but it's been kind of hectic recently!)!!
The figures you quote seem about "right": my 2.6Ghz P4 with WinXP Home and 512 MbRAM takes about 4 hours on average to complete 1 WU. I run
SETI@home client 3.08 constantly in the background incidentally. Times can change dependent on many factors but significant ones are:
a) The data in the WU itself: if corrupt data is detected eg because of locally rec'd RFI (radio frequency interference) captured at the same time at
Arecibo, these results are "scrapped" but you still receive a "credit" for them ie it registers as a WU completed. This happens rarely in my
experience but I have known a WU to be returned as completed after about 20 minutes!!
However, some units contain *so much* data that processing
takes longer - again, I've known units take up to 6 hours to process using the same platform. The resultant signals don't necessarily "look" any
different it's just the processing seems to be more complex.
b) Local conditions: as others have written, much depends on what else is happening on your machine at the time, as SETI@home, whilst being very
CPU-demanding, also releases the CPU readily if other programs or system resources need it (it is designed, afterall, to run in the background and not
interfere with what you want the machine to do - S@h wouldn't gain many friends if it stopped you using your machine!!
)
So, for example, when I use my machine for game playing, I might as well disable S@h, as effectively it goes to sleep until I've finished playing the
game. However, programs like MSWord etc - and other "killer apps" - are so undemanding of CPU pwer that I never notice any significant difference in
the cycle times of the WU's (there might be, but they are insignificant). But, as they say, "normal" apps are so dependent on our "slow" input
(from my poor typing at least!) that the CPU simply gets "cracking" whilst waiting for that next input...
I hope this helps you EXR?