Date:	Wed, 4 Nov 1998 11:10:28 +0200 (IST)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender:	eliz AT is
To:	Kbwms AT aol DOT com
cc:	djgpp-workers AT delorie DOT com
Subject:	Re: Inlining math functions and ANSI/Posix
In-Reply-To:	<abd34bd8.363f3b8d@aol.com>
Message-ID:	<Pine.SUN.3.91.981104110504.25255N-100000@is>
MIME-Version:	1.0
Reply-To:	djgpp-workers AT delorie DOT com

Another implication of the code presented by K.B. Williams is that 
whatever clever argument reduction schemes are implemented in the library 
math functions, they are all but useless when the inlined code produces a 
``normal'' FP result, because the optimized code will accept it and never 
call the library.  Is this okay?

Can anyone look into the sources of GCC 2.8.1 and tell what math 
functions does it inline?  I bet those which have direct counterparts as 
FP instructions are on that list, but what about others?

And how would one disable this inlining?  I though that -fno-fast-math 
should do the trick, but I'm told that in GCC 2.8.1 it doesn't.

- Raw text -