Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm
Sender: cygwin-owner AT sourceware DOT cygnus DOT com
Delivered-To: mailing list cygwin AT sourceware DOT cygnus DOT com
Date: Sun, 2 May 1999 19:45:43 -0400
From: Chris Faylor <cgf AT cygnus DOT com>
To: Mumit Khan <khan AT xraylith DOT wisc DOT edu>
Cc: cygwin AT sourceware DOT cygnus DOT com
Subject: Re: (patch) newlib rint fix
Message-ID: <19990502194542.C950@cygnus.com>
References: <19990502193202 DOT B773 AT cygnus DOT com> <Pine DOT SUN DOT 3 DOT 93 DOT 990502183235 DOT 20497F-100000 AT modi DOT xraylith DOT wisc DOT edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.95.3i
In-Reply-To: <Pine.SUN.3.93.990502183235.20497F-100000@modi.xraylith.wisc.edu>; from Mumit Khan on Sun, May 02, 1999 at 06:36:32PM -0500

On Sun, May 02, 1999 at 06:36:32PM -0500, Mumit Khan wrote:
>On Sun, 2 May 1999, Chris Faylor wrote:
>>>The fdlibm assume correct rounding according to the type throughout the
>>>entire set of functions.  I'd suggest to compile all fdlibm functions
>>>with --float-store which should have the same effect.  This avoids
>>>fixing one function after the other.
>>
>>I'll ping the newlib maintainer again and see what he thinks about this
>>issue.
>
>This is a terrible idea, since the performance goes down the drain.
>Please reconsider the use of -ffloat-store.  I believe that this
>particular problem has more to do with code that was written without
>aggressive optimization in mind than rounding modes in x86 FPU.  Jacob
>Navia had suggested an alternative using assembly code that does do the
>right thing.

Ok.  I wasn't necessarily advocating the use of -ffloat-store, I was
just relaying the one response to the "volatile" fix that you mentioned.

It's entirely likely that we might (soon) have an assembly fix for this,
similar to Jacob Navia's.  Hopefully, the newlib maintainer will have a
strong opinion on how this should be fixed and will fix it soon once and
for all.

-chris

--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe AT sourceware DOT cygnus DOT com