From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugs-return-437018-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 22876 invoked by alias); 9 Dec 2013 06:01:29 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Received: (qmail 22860 invoked by uid 48); 9 Dec 2013 06:01:25 -0000
From: "jmvalin at jmvalin dot ca" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c/59424] New: Optimization issue on min/max
Date: Mon, 09 Dec 2013 06:01:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c
X-Bugzilla-Version: 4.7.2
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jmvalin at jmvalin dot ca
X-Bugzilla-Status: UNCONFIRMED
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: bug_id short_desc product version bug_status bug_severity priority component assigned_to reporter
Message-ID: <bug-59424-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-12/txt/msg00673.txt.bz2

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59424

            Bug ID: 59424
           Summary: Optimization issue on min/max
           Product: gcc
           Version: 4.7.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jmvalin at jmvalin dot ca

gcc seems to be having problems optimizing float min/max code and behaves
unpredictably. If I compile the following code on an x86-64 (SSE enabled):

#include <math.h>

float func1(float a, float b) {
  return (a < b ? a : b);
}

float func2(float a, float b) {
  return sqrtf(a < b ? a : b);
}

float func3(float a, float b) {
  return fabsf(a < b ? a : b);
}

I get the following disassembly:

0000000000000000 <func1>:
func1():
   0:   f3 0f 5d c1             minss  %xmm1,%xmm0
   4:   c3                      retq   
   5:   66 66 2e 0f 1f 84 00    data32 nopw %cs:0x0(%rax,%rax,1)
   c:   00 00 00 00 

0000000000000010 <func2>:
func2():
  10:   f3 0f 5d c1             minss  %xmm1,%xmm0
  14:   f3 0f 51 c0             sqrtss %xmm0,%xmm0
  18:   c3                      retq   
  19:   0f 1f 80 00 00 00 00    nopl   0x0(%rax)

0000000000000020 <func3>:
func3():
  20:   0f 2f c8                comiss %xmm0,%xmm1
  23:   77 13                   ja     38 <func3+0x18>
  25:   f3 0f 10 05 00 00 00    movss  0x0(%rip),%xmm0        # 2d <func3+0xd>
  2c:   00 
  2d:   0f 54 c1                andps  %xmm1,%xmm0
  30:   c3                      retq   
  31:   0f 1f 80 00 00 00 00    nopl   0x0(%rax)
  38:   f3 0f 10 0d 00 00 00    movss  0x0(%rip),%xmm1        # 40 <func3+0x20>
  3f:   00 
  40:   0f 54 c1                andps  %xmm1,%xmm0
  43:   c3                      retq   


So gcc is correctly using maxss in func1 and func2, but not in func2. In
practice, I'm using MIN/MAX macros extensively in libopus and gcc is missing
the optimization most of the time, slowing down the code.