public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "marxin at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/107767] [13 Regression] switch to table conversion happening even though using btq is better
Date: Fri, 02 Dec 2022 15:07:14 +0000	[thread overview]
Message-ID: <bug-107767-4-FQBXbRQcgT@http.gcc.gnu.org/bugzilla/> (raw)
In-Reply-To: <bug-107767-4@http.gcc.gnu.org/bugzilla/>

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107767

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|marxin at gcc dot gnu.org          |unassigned at gcc dot gnu.org
             Status|ASSIGNED                    |NEW

--- Comment #7 from Martin Liška <marxin at gcc dot gnu.org> ---
So what happens with the current master: we first convert the if-else-if series
to switch in iftoswitch pass:

  dst_port_5 = MEM[(const uint16_t *)data_3(D) + 64B];
  switch (dst_port_5) <default: <L25> [INV], case 1: <L22> [INV], case 2: <L23>
[INV], case 3: <L24> [INV], case 15: <L17> [INV], case 23: <L18> [INV], case
42: <L21> [INV], case 45: <L20> [INV], case 47: <L19> [INV]>

  <bb 3> :
<L17>:
  // predicted unlikely by early return (on trees) predictor.
  goto <bb 11>; [INV]

  <bb 4> :
<L18>:
  // predicted unlikely by early return (on trees) predictor.
  goto <bb 11>; [INV]

  <bb 5> :
<L19>:
  // predicted unlikely by early return (on trees) predictor.
  goto <bb 11>; [INV]

  <bb 6> :
<L20>:
  // predicted unlikely by early return (on trees) predictor.
  goto <bb 11>; [INV]

  <bb 7> :
<L21>:
  // predicted unlikely by early return (on trees) predictor.
  goto <bb 11>; [INV]

  <bb 8> :
<L22>:
  // predicted unlikely by early return (on trees) predictor.
  goto <bb 11>; [INV]

  <bb 9> :
<L23>:
  // predicted unlikely by early return (on trees) predictor.
  goto <bb 11>; [INV]

  <bb 10> :
<L24>:
  // predicted unlikely by early return (on trees) predictor.

  <bb 11> :
  # _2 = PHI <1(4), 1(5), 1(3), 1(10), 1(9), 1(8), 1(7), 1(6), 0(2)>

then convert tree-switch-conversion which prefers bit test if possible.
However, the CFG is not collapsed and thus it fails due to:

bool
bit_test_cluster::is_beneficial (unsigned count, unsigned uniq)
{
  return (((uniq == 1 && count >= 3)
           || (uniq == 2 && count >= 5)
           || (uniq == 3 && count >= 6)));
}

as count == 7. and so tree-switch-conversion happens. So one can mitigate that
with:
1) use switch statement instead of if series
2) reduce -param=switch-conversion-max-branch-ratio= that will not create so
big CSWTCH array
3) disable tree-switch-conversion pass

  parent reply	other threads:[~2022-12-02 15:07 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-20 13:00 [Bug regression/107767] New: GCC has some problems in optimizer of trivial case socketpair at gmail dot com
2022-11-20 13:02 ` [Bug regression/107767] " socketpair at gmail dot com
2022-11-21  8:46 ` marxin at gcc dot gnu.org
2022-11-21 10:33 ` socketpair at gmail dot com
2022-11-21 15:57 ` [Bug tree-optimization/107767] [13 Regression] " pinskia at gcc dot gnu.org
2022-12-02 14:27 ` [Bug tree-optimization/107767] [13 Regression] switch to table conversion happening even though using btq is better marxin at gcc dot gnu.org
2022-12-02 14:54 ` socketpair at gmail dot com
2022-12-02 15:00 ` marxin at gcc dot gnu.org
2022-12-02 15:07 ` marxin at gcc dot gnu.org [this message]
2022-12-02 15:15 ` socketpair at gmail dot com
2022-12-02 15:17 ` jakub at gcc dot gnu.org
2022-12-02 15:27 ` rguenth at gcc dot gnu.org
2022-12-02 15:31 ` jakub at gcc dot gnu.org
2022-12-02 15:44 ` jakub at gcc dot gnu.org
2022-12-05 14:49 ` marxin at gcc dot gnu.org
2022-12-21  9:46 ` rguenth at gcc dot gnu.org
2022-12-23 14:05 ` marxin at gcc dot gnu.org
2023-01-04 14:26 ` jakub at gcc dot gnu.org
2023-01-06 12:36 ` marxin at gcc dot gnu.org
2023-01-09 10:02 ` rguenth at gcc dot gnu.org
2023-01-11 12:14 ` cvs-commit at gcc dot gnu.org
2023-01-11 12:14 ` rguenth 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-107767-4-FQBXbRQcgT@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).