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 06E943858D32; Thu, 13 Apr 2023 07:30:59 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 06E943858D32
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1681371059;
	bh=Cwuu8Ji1A3KQ1XkTY/SfB0zyXFUp/ZmQp9sy6rQPpLs=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=gV5ZE4LBVTKpvdgxHPOTkB/53c3SIN5f/wMqeU2aLAx1GCnw3Xx3mDkEdqVOO7ty8
	 5jumO32gluFqA27PCUM04EgTDpH+AZzER4rc/Upy6ex9SC/nCRt2TWV3QSf0e/4OG2
	 HFSP56br6RefbxHXmvMhH00/K0+cWGVVBJRtvAK0=
From: "rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/109048] [13 regression] redundant mask
 compare generated by vectorizer.
Date: Thu, 13 Apr 2023 07:30:58 +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: 13.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
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: 13.0
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-109048-4-k6UUjDLIY5@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-109048-4@http.gcc.gnu.org/bugzilla/>
References: <bug-109048-4@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=3D109048

--- Comment #11 from Richard Biener <rguenth at gcc dot gnu.org> ---
The recent patch improved this to avoid some of the compares.  We still have
the three-argument PHI and thus three VEC_CONDs.

.L10:
        vmovups (%rdi,%rdx), %ymm0
        vcmpltps        %ymm6, %ymm0, %ymm3
        vcmpltps        %ymm2, %ymm0, %ymm1
        vpandn  %ymm1, %ymm3, %ymm1
        vblendvps       %ymm1, %ymm5, %ymm4, %ymm1
        vblendvps       %ymm3, %ymm7, %ymm1, %ymm1
        vaddps  %ymm1, %ymm0, %ymm0
        vaddps  (%rax,%rdx), %ymm0, %ymm0
        vmovups %ymm0, (%rax,%rdx)
        addq    $32, %rdx
        cmpq    $1024, %rdx
        jne     .L10

vs. GCC 12

.L6:
        vmovups (%rdi,%rdx), %ymm1
        vcmpltps        %ymm5, %ymm1, %ymm0
        vcmpltps        %ymm6, %ymm1, %ymm4
        vblendvps       %ymm0, %ymm3, %ymm2, %ymm0
        vandps  %ymm3, %ymm4, %ymm4
        vaddps  %ymm4, %ymm0, %ymm0
        vaddps  %ymm1, %ymm0, %ymm0
        vaddps  (%rax,%rdx), %ymm0, %ymm0
        vmovups %ymm0, (%rax,%rdx)
        addq    $32, %rdx
        cmpq    $1024, %rdx
        jne     .L6

which at least overall looks comparable.=