From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: by sourceware.org (Postfix, from userid 48) id CD00F3853C03; Wed, 28 Jul 2021 10:03:05 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org CD00F3853C03 From: "crazylht at gmail dot com" To: gcc-bugs@gcc.gnu.org Subject: [Bug target/101579] Suboptimal codegen for __builtin_shufflevector Date: Wed, 28 Jul 2021 10:03:05 +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: 12.0 X-Bugzilla-Keywords: missed-optimization X-Bugzilla-Severity: normal X-Bugzilla-Who: crazylht at gmail dot com 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: In-Reply-To: References: 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 List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 28 Jul 2021 10:03:05 -0000 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D101579 --- Comment #7 from Hongtao.liu --- (In reply to Jakub Jelinek from comment #6) > (In reply to Hongtao.liu from comment #4) > > I looks to me that middle end should be able to transform 64-byte vector > > shuffle to 32-byte vector shuffle when data flow analysis shows the upp= er > > part of the vector is never used. >=20 > That is not needed in this case. The permutation is such that all indices > for the first half read from one half (in this case the first) and all > indices for the second half read from one half (in this case the first > again), so it is > identical to a vector containing permutation of the first half with the > first half of the indices and permutation of the first half again with the > second half of the indices. U u =3D ((union { V a; U b; }) w).b + ((union { V a; U b; }) w).b[1]; return u; I means the result only use the first half, we can just create a tmp v1 with v1 =3D __builtin_shufflevector (g, g, 0, 1, 2, 0, 5, 1, 0, 1, 3, 2, 3, 0, 4, 3, = 1, 2, 2, 0, 4, 2, 3, 1, 1, 2, 3, 4, 1, 1, 0, 0, = 5, 2)(In reply to Jakub Jelinek from comment #6) > (In reply to Hongtao.liu from comment #4) > > I looks to me that middle end should be able to transform 64-byte vector > > shuffle to 32-byte vector shuffle when data flow analysis shows the upp= er > > part of the vector is never used. >=20 > That is not needed in this case. The permutation is such that all indices > for the first half read from one half (in this case the first) and all > indices for the second half read from one half (in this case the first > again), so it is > identical to a vector containing permutation of the first half with the > first half of the indices and permutation of the first half again with the > second half of the indices. U u =3D ((union { V a; U b; }) w).b + ((union { V a; U b; }) w).b[1]; return u; I means the result u only cared about the first half, we can drop the second half, it's redundant.=