Hello,
I've compiled mprime on my own (after doing some minor changes to make it
compile under glibc2), and I've noticed something strange. If I run the program
with -m, and use `Test/Continue', I get an iteration time of 0.201 secs
(n=7398xxx, P2/400 overclocked to 448MHz (bus=112MHz, multiplier=4); I use
some programs to keep track of the CPU temperature, it is not even close to
overheating).
But, if I instead run it from init, on a separate tty (on the same machine,
kernel, binary etc.), the iteration time goes up to 0.203 secs! These results
are very consistent, usually only +/- 10,000 clocks (considering that the
iteration takes about 90,000,000 clock, this is nothing). Anybody know why?
(I've compiled the binary with `-O2 -fomit-frame-pointer -m486', and it seems
to be a few percent faster that George's, especially on smaller FFTs.)
/* Steinar */
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
