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 1370C3858D33; Fri, 19 Jan 2024 00:33:49 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 1370C3858D33
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1705624430;
	bh=OCEHh73OQmSVvucZpJqpbm/ILAOfzvGVCcLh5eUjg04=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=ATLShn1WRMNftRaBImSfKgEJ7gi7oH+xFx1unXmIGrRVS3KCAsNwsyuqGKHRgTouU
	 7pUAEYfXh71zoKUSiMVnUSud/IuvtYLoc/HHlpGHlled+F/EJxEwZAZnxGstIBNH/F
	 CHaptmeAethqNwXcvoCoBP3D93qpZa2YiJ44/Poo=
From: "liuhongt at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/113458] Missed SLP for reduction of
 multiplication/addition with promotion
Date: Fri, 19 Jan 2024 00:33:48 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 14.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: enhancement
X-Bugzilla-Who: liuhongt at gcc dot gnu.org
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: pinskia at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-113458-4-psEqo2fWlp@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-113458-4@http.gcc.gnu.org/bugzilla/>
References: <bug-113458-4@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
List-Id: <gcc-bugs.sourceware.org>

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D113458

--- Comment #6 from Hongtao Liu <liuhongt at gcc dot gnu.org> ---

> thus
>=20
>   vect__16.5_40 =3D MEM <vector(4) short int> [(short int *)a_22(D)];
>   vect__17.6_41 =3D (vector(4) int) vect__16.5_40;
>   vect__18.9_44 =3D MEM <vector(4) signed char> [(signed char *)b_23(D)];
>   vect_patt_36.10_45 =3D (vector(4) signed short) vect__18.9_44;
>   vect_patt_37.11_46 =3D (vector(4) int) vect_patt_36.10_45;
>   vect__20.12_48 =3D vect__17.6_41 * vect_patt_37.11_46;
>   _49 =3D VIEW_CONVERT_EXPR<vector(4) unsigned int>(vect__20.12_48);
>   _50 =3D .REDUC_PLUS (_49); [tail call]
>   _51 =3D (int) _50;
>=20
>
Ideally, it should be recognized as DOT_PROD_EXPR, but
vect_recog_dot_prod_pattern only works for loop vectorizer. We may add some
pattern match in match.pd for


vect__17.6_41 =3D (vector(4) int) vect__16.5_40;
vect_patt_37.11_46 =3D (vector(4) int) vect_patt_36.10_45
vect__20.12_48 =3D vect__17.6_41 * vect_patt_37.11_46;
_49 =3D VIEW_CONVERT_EXPR<vector(4) unsigned int>(vect__20.12_48);
_50 =3D .REDUC_PLUS (_49); [tail call]

to

vect__20.12_48 =3D DOT_PRDO_EXPR (vect__16.5_40, vect_patt_36.10_45, 0);
_49 =3D VIEW_CONVERT_EXPR<vector(4) unsigned int>(vect__20.12_48);
_50 =3D .REDUC_PLUS (_49); [tail call]=