public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "pinskia at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/106989] GCC fail to vectorize and clang succeed
Date: Wed, 21 Sep 2022 00:42:48 +0000 [thread overview]
Message-ID: <bug-106989-4-2H2AP62gO3@http.gcc.gnu.org/bugzilla/> (raw)
In-Reply-To: <bug-106989-4@http.gcc.gnu.org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106989
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
/app/example.cpp:20:25: note: Detected interleaving store a[i_27] and a[i_27]
/app/example.cpp:20:25: note: Queuing group with duplicate access for fixup
/app/example.cpp:20:25: note: zero step in outer loop.
/app/example.cpp:20:25: note: zero step in outer loop.
/app/example.cpp:20:25: missed: not vectorized: complicated access pattern.
/app/example.cpp:22:18: missed: not vectorized: complicated access pattern.
/app/example.cpp:20:25: missed: bad data access.
...
/app/example.cpp:21:27: note: dependence distance = 0.
/app/example.cpp:21:27: note: dependence distance == 0 between b[i_27] and
b[i_27]
/app/example.cpp:21:27: note: dependence distance = 1.
/app/example.cpp:22:18: missed: not vectorized, possible dependence between
data-refs a[i_27] and a[_9]
/app/example.cpp:21:27: missed: bad data dependence.
/app/example.cpp:21:27: note: ***** Analysis failed with vector mode V4SF
There is a missing DSE before hand:
# VUSE <.MEM_28>
_1 = bD.3768[i_27];
# VUSE <.MEM_28>
_2 = cD.3769[i_27];
# VUSE <.MEM_28>
_3 = dD.3770[i_27];
_4 = _2 * _3;
_5 = _1 + _4;
# .MEM_19 = VDEF <.MEM_28>
aD.3767[i_27] = _5;
# VUSE <.MEM_19>
_6 = eD.3771[i_27];
_7 = _3 * _6;
_8 = _5 + _7;
# .MEM_20 = VDEF <.MEM_19>
bD.3768[i_27] = _8;
# RANGE [irange] int [1, 31999] NONZERO 0x7fff
_9 = i_27 + 1;
# VUSE <.MEM_20>
_10 = aD.3767[_9];
_11 = _3 * _10;
_12 = _8 + _11;
# .MEM_21 = VDEF <.MEM_20>
aD.3767[i_27] = _12;
DSE does not notice the store defining MEM_19 does touch the load:
# VUSE <.MEM_20>
_10 = aD.3767[_9];
And that it is redudent with the store defining MEM_21.
next prev parent reply other threads:[~2022-09-21 0:42 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-20 23:05 [Bug c/106989] New: " juzhe.zhong at rivai dot ai
2022-09-20 23:21 ` [Bug tree-optimization/106989] " pinskia at gcc dot gnu.org
2022-09-20 23:46 ` crazylht at gmail dot com
2022-09-21 0:42 ` pinskia at gcc dot gnu.org [this message]
2022-09-21 0:45 ` pinskia at gcc dot gnu.org
2022-09-21 0:50 ` juzhe.zhong at rivai dot ai
2022-09-21 8:11 ` rguenth at gcc dot gnu.org
2022-09-22 15:56 ` pinskia 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-106989-4-2H2AP62gO3@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: link
Be 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).