From: leathm AT solwarra DOT gbrmpa DOT gov DOT au (Leath Muller)
Message-Id: <199701310251.MAA23590@solwarra.gbrmpa.gov.au>
Subject: Re: doubles vs. floats
To: eblazecka AT bc DOT sympatico DOT ca
Date: Fri, 31 Jan 1997 12:51:06 +1000 (EST)
Cc: djgpp AT delorie DOT com
In-Reply-To: <32F00F92.1662@bc.sympatico.ca> from "Ryan Blazecka" at Jan 29, 97 07:03:46 pm
Content-Type: text

> So then (disregarding calling the math functions, printf, etc), using
> long doubles would be even faster, right?
 
If you talking plain processing power , then you have to decide _which_
CPU your designing for. On a Pentium, it makes little difference (unless
you setup the FPU to do everything in single precision, which means you
can do a divide in 17 cycles instead of 33) because the conversion
from float->80bit takes no time. On pre-pentium, the
conversion takes time. :)  I think you have to first decide _what_
CPU/FPU you are programming for, and decide from there...

Leathal.