Lucas Wiman writes:
You could do factoring, the margin between factoring on a PC and an
UltraSPARC should be much slimmer than LL tests.
The last time I did timings like this - admittedly, probably over a
year ago, but the mers package hasn't changed much since, especially
in terms of performance - this is wrong. SPARC LL testing -
especially with MacLucasUNIX - is much closer to matching Prime95 LL
testing than SPARC trial factoring - with mersfacgmp, say - is to
matching Prime95's trial factoring, by, as I recall, a factor of about
12. Could someone out there do some new tests and send me the info?
I no longer have access to any SPARCs for Mersenne stuffs.
Mersfacgmp can probably be helped significantly by adjusting the size
of the sieve array and how many small primes are used to check the
primality of the trial factors, however; would anyone like to try some
experiments? I would add detailed info the the comments in the mers
Makefile so noone has to repeat the testing.
The UltraSPARC would probably significantly outperform a similar
megahertz PC, if we had similarly optimized code running on each.
Perhaps.
I know my friend's SPARC (running at 233 mhz, bus 233 mhz) sure
does outperform my PII (running at 233 mhz, bus 66 mhz) on most
things.
I have certainly seen this for most things involving large amounts of
I/O; again, not recently. But the advantage seems to disappear or
even reverse to the PC's favor once cost is taken into account.
Will
http://www.garlic.com/~wedgingt/mersenne.html
mers.tar.gz
`
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
