From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:39859)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1VZkTg-0005RW-UO
	for qemu-devel@nongnu.org; Fri, 25 Oct 2013 12:43:21 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1VZkTX-0004vg-Gr
	for qemu-devel@nongnu.org; Fri, 25 Oct 2013 12:43:12 -0400
Sender: Richard Henderson <rth7680@gmail.com>
Message-ID: <526A9F91.104@twiddle.net>
Date: Fri, 25 Oct 2013 09:42:57 -0700
From: Richard Henderson <rth@twiddle.net>
MIME-Version: 1.0
References: <526947CA.4020504@gmail.com> <526949E1.3010405@gmail.com>
	<5269852E.9000601@twiddle.net> <526A9B6A.7080001@gmail.com>
In-Reply-To: <526A9B6A.7080001@gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [PATCH 13/19] Add VSX ISA2.06 Multiply Add
	Instructions
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Tom Musta <tommusta@gmail.com>, QEMU Developers <qemu-devel@nongnu.org>
Cc: "qemu-ppc@nongnu.org" <qemu-ppc@nongnu.org>

On 10/25/2013 09:25 AM, Tom Musta wrote:
> 
> I tried recoding xsmaddadp using float64_muladd.  The problem that I hit is the
> boundary case where the intermediate product and the summand are infinities of
> the opposite sign.  This is the case handled by the first "if" in the code
> snippet above.  PowerPC has a dedicated FPSCR bit for this type of condition
> (VXISI) as well as a general invalid operation bit (VX).  As far as I can tell,
> the softfloat code only has the equivalent of the VX bit.   Thus the
> implementation
> that I proposed is a more accurate representation of the Power ISA.
> 
> The VSX code was modeled after the existing fmadd FPU instruction.  I suspect
> the author of that code wrote it this way for similar reasons.
> 
> I am inclined to keep my proposed implementation, which is consistent with
> the existing PowerPC code.
> 
> Thoughts?

Hmm.  I won't object to your current implementation, since it does produce
correct results.

I believe that a better implementation could use float*_muladd, and check the
result for float_flag_invalid.  If set, compute the intermediate product so you
can figure out the VXISI setting.  But we'd expect that to be an unlikely path.


r~