From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugzilla@gcc.gnu.org>
Received: by sourceware.org (Postfix, from userid 48)
 id F06FD3858C3A; Wed, 22 Dec 2021 11:08:18 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org F06FD3858C3A
From: "hubicka at kam dot mff.cuni.cz" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/103797] Clang vectorized LightPixel while GCC
 does not
Date: Wed, 22 Dec 2021 11:08:18 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: unknown
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: enhancement
X-Bugzilla-Who: hubicka at kam dot mff.cuni.cz
X-Bugzilla-Status: WAITING
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-103797-4-meOOHTpwLg@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-103797-4@http.gcc.gnu.org/bugzilla/>
References: <bug-103797-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-BeenThere: gcc-bugs@gcc.gnu.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Gcc-bugs mailing list <gcc-bugs.gcc.gnu.org>
List-Unsubscribe: <https://gcc.gnu.org/mailman/options/gcc-bugs>,
 <mailto:gcc-bugs-request@gcc.gnu.org?subject=unsubscribe>
List-Archive: <https://gcc.gnu.org/pipermail/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-request@gcc.gnu.org?subject=help>
List-Subscribe: <https://gcc.gnu.org/mailman/listinfo/gcc-bugs>,
 <mailto:gcc-bugs-request@gcc.gnu.org?subject=subscribe>
X-List-Received-Date: Wed, 22 Dec 2021 11:08:19 -0000

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D103797

--- Comment #4 from hubicka at kam dot mff.cuni.cz ---
> -E and remove not needed code.
>=20
> > The
> > declaratoins are quite convoluted, but the function is well isolated and
> > easy to inspect from full one...
>=20
> Do we speak about:
> https://github.com/mozilla/gecko-dev/blob/bd25b1ca76dd5d323ffc69557f6cf75=
9ba76ba23/gfx/2d/FilterNodeSoftware.cpp#L3670-L3691
> ?
Yes.
>=20
> It should be possible creating a synthetical test that does the same (and=
 lives
> in a loop, right?).

Well, I tried that for a while and got bit lost (either code got
vectorized by both gcc and clang or by neither).  There are more issues
where we have over 50% regression wrt clang build at gfx code, so I
think I will first try to reproduce those locally and perf them to see
if there is more pattern here.

The releavant code is:

uint32_t mozilla::gfx::{anonymous}::SpecularLightingSoftware::LightPixel
(struct SpecularLightingSoftware * const this, const struct Point3D & aNorm=
al,
const struct Point3D & aVectorToLight, uint32_t aColor)
{

  <bb 2> [local count: 118111600]:
  _48 =3D MEM[(const struct BasePoint3D
*)aVectorToLight_25(D)].D.75826.D.75829.z;
  _49 =3D _48 + 1.0e+0;
  _50 =3D MEM[(const struct BasePoint3D
*)aVectorToLight_25(D)].D.75826.D.75829.y;
  _51 =3D _50 + 0.0;
  _52 =3D MEM[(const struct BasePoint3D
*)aVectorToLight_25(D)].D.75826.D.75829.x;
  _53 =3D _52 + 0.0;
  _80 =3D _53 * _53;
  _82 =3D _51 * _51;
  _83 =3D _80 + _82;
  _85 =3D _49 * _49;
  _86 =3D _83 + _85;
  if (_86 u>=3D 0.0)
    goto <bb 3>; [99.95%]
  else
    goto <bb 4>; [0.05%]

  <bb 3> [local count: 118052545]:
  _87 =3D .SQRT (_86);
  goto <bb 5>; [100.00%]

  <bb 4> [local count: 59055]:
  _29 =3D __builtin_sqrtf (_86);

  <bb 5> [local count: 118111600]:
  # _30 =3D PHI <_29(4), _87(3)>
  _88 =3D _53 / _30;
  _89 =3D _51 / _30;
  _90 =3D _49 / _30;
  _41 =3D MEM[(const struct BasePoint3D *)aNormal_26(D)].D.75826.D.75829.x;
  _39 =3D _41 * _88;
  _37 =3D MEM[(const struct BasePoint3D *)aNormal_26(D)].D.75826.D.75829.y;
  _33 =3D _37 * _89;
  _27 =3D _33 + _39;
  _45 =3D MEM[(const struct BasePoint3D *)aNormal_26(D)].D.75826.D.75829.z;
  _46 =3D _45 * _90;
  _47 =3D _27 + _46;
  if (_47 >=3D 0.0)
    goto <bb 12>; [59.00%]
  else
    goto <bb 6>; [41.00%]


With -Ofast it gets bit more streamlined:


  <bb 2> [local count: 118111600]:
  _48 =3D MEM[(const struct BasePoint3D
*)aVectorToLight_25(D)].D.75826.D.75829.z;
  _49 =3D _48 + 1.0e+0;
  _50 =3D MEM[(const struct BasePoint3D
*)aVectorToLight_25(D)].D.75826.D.75829.y;
  _51 =3D MEM[(const struct BasePoint3D
*)aVectorToLight_25(D)].D.75826.D.75829.x;
  powmult_78 =3D _51 * _51;
  powmult_80 =3D _50 * _50;
  _81 =3D powmult_78 + powmult_80;
  powmult_83 =3D _49 * _49;
  _84 =3D _81 + powmult_83;
  _85 =3D __builtin_sqrtf (_84);
  _86 =3D _51 / _85;
  _87 =3D _50 / _85;
  _88 =3D _49 / _85;
  _41 =3D MEM[(const struct BasePoint3D *)aNormal_26(D)].D.75826.D.75829.x;
  _39 =3D _41 * _86;
  _37 =3D MEM[(const struct BasePoint3D *)aNormal_26(D)].D.75826.D.75829.y;
  _33 =3D _37 * _87;
  _27 =3D _33 + _39;
  _45 =3D MEM[(const struct BasePoint3D *)aNormal_26(D)].D.75826.D.75829.z;
  _46 =3D _45 * _88;
  _47 =3D _27 + _46;
  if (_47 >=3D 0.0)
    goto <bb 3>; [59.00%]
  else
    goto <bb 9>; [41.00%]

But I do not quite see in the slp dump why this is not considered for
vectorization.

I attach the dump.
Honza=