public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "crazylht at gmail dot com" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug middle-end/102977] New: [GCC12 regression] vectorizer failed to generate complex fma. Date: Thu, 28 Oct 2021 01:34:36 +0000 [thread overview] Message-ID: <bug-102977-4@http.gcc.gnu.org/bugzilla/> (raw) https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102977 Bug ID: 102977 Summary: [GCC12 regression] vectorizer failed to generate complex fma. Product: gcc Version: 12.0 Status: UNCONFIRMED Severity: normal Priority: P3 Component: middle-end Assignee: unassigned at gcc dot gnu.org Reporter: crazylht at gmail dot com Target Milestone: --- Target: aarch64-linux-gnu #include<complex.h> #include<complex.h> void foo (_Complex _Float16* __restrict a, _Complex _Float16* b, _Complex _Float16 *c) { for (int i =0 ; i != 8; i++) a[i] += b[i] * c[i]; } gcc11.2 generate foo: mov x3, 16 ptrue p1.b, all whilelo p0.h, xzr, x3 ld1h z2.h, p0/z, [x1] ld1h z1.h, p0/z, [x2] ld1h z0.h, p0/z, [x0] fcmla z0.h, p1/m, z1.h, z2.h, #0 fcmla z0.h, p1/m, z1.h, z2.h, #90 st1h z0.h, p0, [x0] cntb x4 cnth x5 add x0, x0, x4 add x1, x1, x4 add x2, x2, x4 whilelo p0.h, x5, x3 b.none .L1 ld1h z2.h, p0/z, [x1] ld1h z1.h, p0/z, [x2] ld1h z0.h, p0/z, [x0] fcmla z0.h, p1/m, z1.h, z2.h, #0 fcmla z0.h, p1/m, z1.h, z2.h, #90 st1h z0.h, p0, [x0] .L1: ret current trunk foo: ptrue p1.h, vl8 ptrue p0.b, all ld2h {z2.h - z3.h}, p1/z, [x1] ld2h {z0.h - z1.h}, p1/z, [x2] ld2h {z16.h - z17.h}, p1/z, [x0] fmul z6.h, z0.h, z3.h movprfx z7, z16 fmla z7.h, p0/m, z0.h, z2.h fmla z6.h, p0/m, z1.h, z2.h movprfx z4, z7 fmls z4.h, p0/m, z1.h, z3.h fadd z5.h, z6.h, z17.h st2h {z4.h - z5.h}, p1, [x0] ret options: -Ofast -march=armv8.3-a+sve+fp16 refer to https://godbolt.org/z/4PPKnWvc1
next reply other threads:[~2021-10-28 1:34 UTC|newest] Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top 2021-10-28 1:34 crazylht at gmail dot com [this message] 2021-10-28 4:44 ` [Bug middle-end/102977] [GCC12 regression] vectorizer failed to generate complex fma with SVE pinskia at gcc dot gnu.org 2021-10-28 4:45 ` pinskia at gcc dot gnu.org 2021-10-28 4:47 ` pinskia at gcc dot gnu.org 2021-10-28 4:52 ` [Bug middle-end/102977] [12 Regression] vectorizer failed to use armv8.3-a complex fma pinskia at gcc dot gnu.org 2021-10-28 4:58 ` [Bug tree-optimization/102977] " pinskia at gcc dot gnu.org 2021-10-28 5:12 ` pinskia at gcc dot gnu.org 2021-10-28 6:48 ` crazylht at gmail dot com 2021-10-28 8:25 ` tnfchris at gcc dot gnu.org 2021-10-28 8:52 ` tnfchris at gcc dot gnu.org 2021-10-29 11:49 ` cvs-commit at gcc dot gnu.org 2021-10-29 11:49 ` cvs-commit at gcc dot gnu.org 2021-10-29 11:51 ` tnfchris at gcc dot gnu.org
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=bug-102977-4@http.gcc.gnu.org/bugzilla/ \ --to=gcc-bugzilla@gcc.gnu.org \ --cc=gcc-bugs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).