From mboxrd@z Thu Jan  1 00:00:00 1970
From: Joern Rennecke <amylaar@redhat.com>
To: dewar@gnat.com
Cc: aoliva@redhat.com, gdr@codesourcery.com, amylaar@redhat.com, gcc@gcc.gnu.org, moshier@moshier.ne.mediaone.net, torvalds@transmeta.com, tprince@computer.org
Subject: Re: What is acceptable for -ffast-math? (Was: associative law in combine)
Date: Mon, 30 Jul 2001 21:39:00 -0000
Message-id: <200107310439.f6V4dKh09152@phal.cambridge.redhat.com>
References: <20010731035353.271D3F2B65@nile.gnat.com>
X-SW-Source: 2001-07/msg02070.html

> If you see  a / 2.0 / 2.0
> 
> it is absolutely definition OK, to compute this particular expression
> as a / 4.0 in any language no matter WHAT the standard says, since the
> result is exactly the same. But for other divisors this is not necessarily

No, if you have IEEE denormals, double rounding can give a different result.