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 521203858D20; Mon,  8 Apr 2024 12:35:13 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 521203858D20
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1712579713;
	bh=cFTKCET8pZqQjJb8sx74s/G/FT0prz0Ux7eBQTUSyfs=;
	h=From:To:Subject:Date:In-Reply-To:References:From;
	b=WywZbemneOAYt7818rdOQGcX8IIkvjPE2ay2ckF65O1gdB9733ulct6uUB/wxMH30
	 HDLTml1+SYa/fHr96sznMWIl48R127jIRihFGZQWlGOj9j2TBFZJhwugObJQkcCPH6
	 c5Nuzx+EVZqeB6mPaekYC+2i+sKe6aMwvwn/nS7g=
From: "jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/114635] OpenMP reductions fail dependency
 analysis
Date: Mon, 08 Apr 2024 12:35:13 +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, openmp
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
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-114635-4-znq4kk3n7c@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-114635-4@http.gcc.gnu.org/bugzilla/>
References: <bug-114635-4@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=3D114635

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #3)
> As for the dependence analysis itself - the issue is that the D.5606[_33]
> style accesses have no access functions - I'm not sure how they evolve,
> but I see their domain is even [0, POLY_INT_CST [15, 16] ].  I guess
> we don't expect to have dependence analysis handle this but do dependence
> fully within 'safelen' here.

The magic arrays are indexed in the loop using .GOMP_SIMD_LANE (...) index,
which should map to the logical iteration modulo vectorization factor after=
 the
vectorization.  After the loop it can be indexed using .GOMP_SIMD_LAST_LANE
(...).=