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 14240385841D; Tue, 28 Nov 2023 12:02:23 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 14240385841D
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1701172943;
	bh=dUXVdCVZWSsAp1Z583QdRMPtuftv4I1iN66Kj0natTM=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=TFdpo0te2HhgQas2EiCoOfU1n3RhjjcKHpABALZW1RY79zaVFA1dKVwyUzUxIgq5N
	 yhYniQ735QwcdrFpgEdZ0YpRcSFxGUpn7JVpBsqHIgq830Djvqd+yqoqPfFvrMDxMs
	 Ixg98vlSC82aZKonXejlBxgrJ57sDcsI/Rz20TwM=
From: "rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/112742] missed vectorization with src[a*b+i]
 where a*b is not int rather than same percission as size_type
Date: Tue, 28 Nov 2023 12:02:22 +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: 14.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: enhancement
X-Bugzilla-Who: rguenth at gcc dot gnu.org
X-Bugzilla-Status: NEW
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-112742-4-LaPczPcqEI@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-112742-4@http.gcc.gnu.org/bugzilla/>
References: <bug-112742-4@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=3D112742

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah, row*stride is loop invariant and that saves us with -m32.  With -m64
we end up with

  <bb 3> [local count: 105119324]:
  row.0_1 =3D (unsigned int) row_13(D);
  _2 =3D row.0_1 * stride_14(D);

  <bb 4> [local count: 955630224]:
  # col_20 =3D PHI <col_17(6), 0(3)>
  col.1_3 =3D (unsigned int) col_20;
  _4 =3D _2 + col.1_3;
  _5 =3D (sizetype) _4;
  _6 =3D src_15(D) + _5;
  _7 =3D *_6;

which is "unfortuante" association of the (sizetype) conversion.  But
as the col addtition is unsigned it might overflow and we can't
associate the (sizetype) conversion but it makes the result non-affine.

A runtime versioning would be necessary, guaranteeing that _2 + col.1_3
never overflows.=