From what I understand of Tom’s ‘benchmarking’ process, he actaully takes the total time taken for x units to run on an SMP capable system (where x is the number of CPU’s [virtual or real]) and then divides by x to get a (misleading) result.
So for that P4 system, it actually took 2h44m to run the benchmark, but actually knocked out 2 of them in that time…
Umm, a drag ass 3gig p4 takes a bit inder 3 hours for a woot, and even my ropey old slow 2gig amd does it in a fart over 3 hrs … intel playing catchup then … nothing we didn’t know already
SB
ps If we apply “toms maths” to this, the dual opteron rigs will be knocking them out in 33minutes then … Let’s see a p4 beat that
Originally posted by Stevieboy
[B]Umm, a drag ass 3gig p4 takes a bit inder 3 hours for a woot, and even my ropey old slow 2gig amd does it in a fart over 3 hrs … intel playing catchup then … nothing we didn’t know already
SB
ps If we apply “toms maths” to this, the dual opteron rigs will be knocking them out in 33minutes then … Let’s see a p4 beat that [/B]
Originally posted by richardlellisjr I only get nine per day out of a P4 2.67Ghz, 533Mhz bus, 512 Mb PC3300 Intel 845 motherboard, WinXP Pro. Am I doing something wrong?
no, don’t worry the only reason the 3.2 does 16ish WPD is because it uses hyperthreading to run two CLIs at the same time thus it does a work unit every 2hr40min but in this time it has run two clients and so produced 2 results.
this hyperthreading stuff seems to be good for WU production
Originally posted by lab_boy this hyperthreading stuff seems to be good for WU production
I think either Prescott or Tejas are going to step up the HT thing even more, possibly 4 threads instead of 2, just think what that will do for WU processing
I guess I will wait a bit before building another rig for SETI. The P4 2.67 I only built two months ago. I have a place for one more and will see what hyperthreading models come out later this year.
Originally posted by lab_boy
[B]no, don’t worry the only reason the 3.2 does 16ish WPD is because it uses hyperthreading to run two CLIs at the same time thus it does a work unit every 2hr40min but in this time it has run two clients and so produced 2 results.
this hyperthreading stuff seems to be good for WU production [/B]
Hmmm, 2 clients?
I’m running a P4 3.0Ghz ‘C’ with HT and at the moment it takes me +/- 2h26m per woo, single client divided over the ‘2’ cpu’s
I use Seti@Home as a service with cli 3.08 (click on my homepage to see current progress )