This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: AArch64 patches for glibc master.
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Carlos O'Donell <carlos at redhat dot com>
- Cc: Marcus Shawcroft <marcus dot shawcroft at linaro dot org>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Mon, 9 Dec 2013 21:40:25 +0000
- Subject: Re: AArch64 patches for glibc master.
- Authentication-results: sourceware.org; auth=none
- References: <52A60A93 dot 1050402 at redhat dot com> <Pine dot LNX dot 4 dot 64 dot 1312091852010 dot 17984 at digraph dot polyomino dot org dot uk> <52A61590 dot 9060303 at redhat dot com>
On Mon, 9 Dec 2013, Carlos O'Donell wrote:
> > * You're adding ulps for sqrt, which should always be exact.
>
> I wasn't aware that sqrt should always be exact. I'll try to
> educate myself on that. Do you have any pointers on where I can
> read about that?
If a function is bound to an IEC 60559 operation, it should always be
exact (correctly rounding according to the current rounding direction and
with a correct set of exceptions raised including "inexact" as
applicable). For example, C11 F.3 says "The sqrt functions in <math.h>
provide the IEC 60559 square root operation.". Other functions that
should never have ulps include rint, fma, scalbn.
--
Joseph S. Myers
joseph@codesourcery.com