Brian J. Beesley writes:
I put a development version on my anon ftp server about three days
ago. ftp://lettuce.edsc.ulst.ac.uk/gimps/DecaMega/factor95.c
Um, that's an unfortunate choice of name; George's P-1 factorer is
"Factor95" (though there's a newer version, renamed to Factor98). O
> Will change the "engine" to keep going to 2^33 after finding the
> first factor & report the results from that.
OK, here's the results. (All factors to 2^33 found, input is 159,975
largest primes < 36 million)
Sieve 6 smallest primes, 3517525 calls, 40.15s
Sieve 10 smallest primes, 2972446 c
> That does sound like it's faster, but let me go thru some numbers. I
> don't recall exactly, but I'm sure my earlier run took less than 72
> hours; 72 hours divided by 20 seconds is 72*3600/20 = 12960. Brian
> stopped at 2^32 and I went to at least 2^33 for a factor of two;
> 12960/2 = 6480.
