I'm not sure I know exactly what you're asking. I get my ranges from mersenneforum PSP and sieve using sr2sieve with the combined SoB.dat file. I have a large PowerPC farm (total of 168 2.5 GHz 970MP cores and 28 2.2 GHz 970FX cpus).
just curious how you found all of those low factors. You appeared to have sieved them being that they are all close together in size. But you have done it very quickly.
I'm not sure I know exactly what you're asking. I get my ranges from mersenneforum PSP and sieve using sr2sieve with the combined SoB.dat file. I have a large PowerPC farm (total of 168 2.5 GHz 970MP cores and 28 2.2 GHz 970FX cpus).
With the SoB.dat, each 970MP core peaks around 380 kps, each 970FX CPU at around 330 kps. They're not all dedicated sievers, and I also do Riesel. With riesel.dat, each 970MP core hits 235 kps and each 970FX CPU about 205 kps.
Right now I'm using 88 970MP cores on Riesel, the 28 970FX cpus on PSP/SOB combined and everything else on OGR/RC5.
Perhaps that explains thisOriginally Posted by Death
http://stats.distributed.net/partici...d=24&id=300568
Macs RULE!!!
I'm 12 in overall OGR-24 and goes straightforward to 10 but lost a BUNCH of 24 stubs from a power failure ((((
Just don't waste your time on RC5-72. It's almost as bad as using trial division to factor RSA-512.
If you want to do PRP testing on PPC, let me know. It cannot be used for the SoB or RieselSieve projects due to some issues with base 2 numbers. It works extremely well for the Sierpinki/Riesel base 5 project (over at mersenneforum). The software, called phrot (written by Phil Carmody and built on YEAFFT), is faster than LLR. To be more specific, phrot on a 2.5 GHz G5 is faster than LLR on a 3.0 GHz P4.