🚀 go-pugleaf

rmarsh@plains.nodak.edu (Ronald Marsh) writes:
>I have been running a neural network program (that I compiled using gcc
>2.5.8) and I have noticed that the program (while running under Linux) is
>about 40% slower than the same program (running under DOS and compiled
>with Turbo C). I have narrowed the possible causes of the poor performance
>down to the transcendental functions (cos, exp, etc). Does anybody know of
>a "better" math library or version of gcc, with respect to math performance?

You might try experimenting with the following flag (quote from the
gcc manual):

@item -ffast-math
This option allows GCC to violate some ANSI or IEEE rules and/or
specifications in the interest of optimizing code for speed.  For
example, it allows the compiler to assume arguments to the @code{sqrt}
function are non-negative numbers and that no floating-point values
are NaNs.

I don't know if the Linux C library has improved any in the last year
or so (if you have gcc 2.5.8 you presumably have a relatively old C library)
-- 
-- Joe Buck 	<jbuck@synopsys.com>	(not speaking for Synopsys, Inc)

Work for something because it is good,
not just because it stands a chance to succeed.	   -- Vaclav Havel