From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-patches-return-297806-listarch-gcc-patches=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 18726 invoked by alias); 24 Jul 2011 12:02:40 -0000
Received: (qmail 18718 invoked by uid 22791); 24 Jul 2011 12:02:39 -0000
X-SWARE-Spam-Status: No, hits=-2.2 required=5.0	tests=AWL,BAYES_00,RCVD_IN_DNSWL_LOW,TW_EQ,TW_LQ,TW_QB,TW_TQ
X-Spam-Check-By: sourceware.org
Received: from mail-gx0-f175.google.com (HELO mail-gx0-f175.google.com) (209.85.161.175)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Sun, 24 Jul 2011 12:02:25 +0000
Received: by gxk3 with SMTP id 3so2169570gxk.20        for <gcc-patches@gcc.gnu.org>; Sun, 24 Jul 2011 05:02:24 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.101.106.25 with SMTP id i25mr3165226anm.80.1311508944495; Sun, 24 Jul 2011 05:02:24 -0700 (PDT)
Received: by 10.100.120.2 with HTTP; Sun, 24 Jul 2011 05:02:24 -0700 (PDT)
In-Reply-To: <CAKSNEw6E3fE9_z9hfV1NSHWn8Dvjg2ind8J0-dU33UW8gBxQ5A@mail.gmail.com>
References: <CAKSNEw4t=p7L1EBFyHAc0rx_CvgXotusEBmbp6LZvFL29DQFJQ@mail.gmail.com>	<201107201835.p6KIZQnF011722@d06av02.portsmouth.uk.ibm.com>	<CAKSNEw6E3fE9_z9hfV1NSHWn8Dvjg2ind8J0-dU33UW8gBxQ5A@mail.gmail.com>
Date: Sun, 24 Jul 2011 14:32:00 -0000
Message-ID: <CAKSNEw5zPPBn9+J9YO0fqmhYXmSVJr66AyT2kWipwR33BDyfZA@mail.gmail.com>
Subject: Re: [patch] Fix PR tree-optimization/49771
From: Ira Rosen <ira.rosen@linaro.org>
To: Ulrich Weigand <uweigand@de.ibm.com>
Cc: gcc-patches@gcc.gnu.org, Patch Tracking <patches@linaro.org>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-patches.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-patches/>
List-Post: <mailto:gcc-patches@gcc.gnu.org>
List-Help: <mailto:gcc-patches-help@gcc.gnu.org>
Sender: gcc-patches-owner@gcc.gnu.org
X-SW-Source: 2011-07/txt/msg02075.txt.bz2

On 21 July 2011 15:19, Ira Rosen <ira.rosen@linaro.org> wrote:
> On 20 July 2011 21:35, Ulrich Weigand <uweigand@de.ibm.com> wrote:
>>
>> The return value of foo with vectorization is 1249 instead
>> of 1999 for some reason.
>
> I reproduced the failure. It occurs without Richard's
> (http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01022.html) and this
> patches too. Obviously the vectorized loop is executed, but at the
> moment I don't understand why. I'll have a better look on Sunday.

Actually it doesn't choose the vectorized code. But the scalar version
gets optimized in a harmful way for SPU, AFAIU.
Here is the scalar loop after vrp2

<bb 8>:
  # ivtmp.42_50 =3D PHI <ivtmp.42_59(3), ivtmp.42_45(10)>
  D.4593_42 =3D (void *) ivtmp.53_32;
  D.4520_33 =3D MEM[base: D.4593_42, offset: 0B];
  D.4521_34 =3D D.4520_33 + 1;
  MEM[symbol: a, index: ivtmp.42_50, offset: 0B] =3D D.4521_34;
  ivtmp.42_45 =3D ivtmp.42_50 + 4;
  if (ivtmp.42_45 !=3D 16)
    goto <bb 10>;
  else
    goto <bb 5>;

and the load is changed by dom2 to:

<bb 4>:
  ...
  D.4520_33 =3D MEM[base: vect_pa.9_19, offset: 0B];
   ...

where vector(4) int * vect_pa.9;

And the scalar loop has no rotate for that load:

.L3:
        lqd     $13,0($2)
        lqx     $11,$5,$3
        cwx     $7,$sp,$3
        ai      $12,$13,1
        shufb   $6,$12,$11,$7
        stqx    $6,$5,$3
        ai      $3,$3,4
        ceqi    $4,$3,16


I manually added rotqby for $13 and the result was correct (I changed
the test to iterate only 4 times to make the things easier).

Ira

>
> Ira
>
>>
>> Bye,
>> Ulrich
>>
>> --
>> =A0Dr. Ulrich Weigand
>> =A0GNU Toolchain for Linux on System z and Cell BE
>> =A0Ulrich.Weigand@de.ibm.com
>>
>