Brian Beesley mentioned not being able to ftp the Mlucas v2.7z SPARC
binary - turns out I had the permissions for all the new binaries set wrong,
so if you had problems downloading any of them (or any other files on my
ftp site), try it now.
Bill Rea reports he got significantly improved timings on his SPARC E450
by redoing them when the machine was idle - I tried this on my Alpha 21164
(the 400MHz machine in Table 1 of the timings page) and lo and behold, a
speedup of roughly 10% at nearly all FFT lengths. I've revised that timings
column. I did not observe a similar effect on my older 21064, and the 21264
I have a guest account on is almost never idle, so I haven't had a chance to
try it there.
I switched my own current-exponent run on the 21164 from v2.7y to v2.7z
today, and the timing for 2000 iterations at 384K has dropped from 11 minutes
to 10. That was nice to see, as it will cut 3 days off a full run at that
length.
Cheers,
-Ernst
ftp://209.133.33.182/pub/mayer/README
ftp://209.133.33.182/pub/mayer/gimps_timings.html
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
