public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug optimization/14237] New: suboptimal comparison between long and long long
@ 2004-02-21 10:53 kazu at cs dot umass dot edu
  2004-02-23  1:23 ` [Bug optimization/14237] " pinskia at gcc dot gnu dot org
                   ` (3 more replies)
  0 siblings, 4 replies; 5+ messages in thread
From: kazu at cs dot umass dot edu @ 2004-02-21 10:53 UTC (permalink / raw)
  To: gcc-bugs

unsigned long long g;

unsigned long
foo (unsigned long a)
{
  return (a >= g) ? 1 : 0;
}

turns into

foo:
	pushl	%esi           ; we wouldn't need this if %esi were untouched
	movl	g+4, %ecx
	xorl	%esi, %esi
	pushl	%ebx
	cmpl	%ecx, %esi     ; we could use "testl" here because %esi == 0
	movl	12(%esp), %ebx
	movl	$1, %eax
	movl	g, %edx
	jb	.L3
	cmpl	%edx, %ebx
	jae	.L2
.L3:
	xorl	%eax, %eax
.L2:
	popl	%ebx
	popl	%esi
	ret

-- 
           Summary: suboptimal comparison between long and long long
           Product: gcc
           Version: 3.5.0
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P2
         Component: optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: kazu at cs dot umass dot edu
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14237


^ permalink raw reply	[flat|nested] 5+ messages in thread
* [Bug optimization/14237] suboptimal comparison between long and long long
  2004-02-21 10:53 [Bug optimization/14237] New: suboptimal comparison between long and long long kazu at cs dot umass dot edu
@ 2004-02-23  1:23 ` pinskia at gcc dot gnu dot org
  2004-02-23  1:36 ` kazu at cs dot umass dot edu
                   ` (2 subsequent siblings)
  3 siblings, 0 replies; 5+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-02-23  1:23 UTC (permalink / raw)
  To: gcc-bugs


------- Additional Comments From pinskia at gcc dot gnu dot org  2004-02-23 01:23 -------
Confirmed, basically the problem is subregisters support sucks (most likely in combine).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
  GCC build triplet|i686-pc-linux-gnu           |
   GCC host triplet|i686-pc-linux-gnu           |
 GCC target triplet|i686-pc-linux-gnu           |i386-pc-linux-gnu
           Keywords|                            |pessimizes-code
   Last reconfirmed|0000-00-00 00:00:00         |2004-02-23 01:23:46
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14237


^ permalink raw reply	[flat|nested] 5+ messages in thread
* [Bug optimization/14237] suboptimal comparison between long and long long
  2004-02-21 10:53 [Bug optimization/14237] New: suboptimal comparison between long and long long kazu at cs dot umass dot edu
  2004-02-23  1:23 ` [Bug optimization/14237] " pinskia at gcc dot gnu dot org
@ 2004-02-23  1:36 ` kazu at cs dot umass dot edu
  2004-02-29  4:10 ` kazu at cs dot umass dot edu
  2004-11-03 18:14 ` [Bug rtl-optimization/14237] " pinskia at gcc dot gnu dot org
  3 siblings, 0 replies; 5+ messages in thread
From: kazu at cs dot umass dot edu @ 2004-02-23  1:36 UTC (permalink / raw)
  To: gcc-bugs


------- Additional Comments From kazu at cs dot umass dot edu  2004-02-23 01:36 -------
Yes, it would be nice if the combiner could do a little more like
simplifying an insn using information from other insns instead of
direct substitution.

Oh, forgot to mention: -O2 -fomit-frame-pointer. :-)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14237


^ permalink raw reply	[flat|nested] 5+ messages in thread
* [Bug optimization/14237] suboptimal comparison between long and long long
  2004-02-21 10:53 [Bug optimization/14237] New: suboptimal comparison between long and long long kazu at cs dot umass dot edu
  2004-02-23  1:23 ` [Bug optimization/14237] " pinskia at gcc dot gnu dot org
  2004-02-23  1:36 ` kazu at cs dot umass dot edu
@ 2004-02-29  4:10 ` kazu at cs dot umass dot edu
  2004-11-03 18:14 ` [Bug rtl-optimization/14237] " pinskia at gcc dot gnu dot org
  3 siblings, 0 replies; 5+ messages in thread
From: kazu at cs dot umass dot edu @ 2004-02-29  4:10 UTC (permalink / raw)
  To: gcc-bugs


------- Additional Comments From kazu at cs dot umass dot edu  2004-02-29 04:10 -------
Basically, in the testcase code, we have the following sequence before
the first CSE.

(parallel [(set (reg:DI 61)
                (zero_extend:DI (reg:SI 59)))
	   (clobber (reg:CC 17 flags))])

  :
  :

(set (reg:CC 17 flags)
     (compare:CC (subreg:SI (reg:DI 61) 4)
		 (subreg:SI (reg:DI 64) 4)))

Note that (subreg:SI (reg:DI 61) 4) is 0 because of zero_extend.
Then I found this patch, which is for PR 13335.

  http://gcc.gnu.org/ml/gcc-patches/2003-11/msg01611.html

Unfortunately, the patch didn't solve the problem.  I haven't analyzed
whether it has made this PR any easier to solve, but it at least shows
one way in which we can solve this PR.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14237


^ permalink raw reply	[flat|nested] 5+ messages in thread
* [Bug rtl-optimization/14237] suboptimal comparison between long and long long
  2004-02-21 10:53 [Bug optimization/14237] New: suboptimal comparison between long and long long kazu at cs dot umass dot edu
                   ` (2 preceding siblings ...)
  2004-02-29  4:10 ` kazu at cs dot umass dot edu
@ 2004-11-03 18:14 ` pinskia at gcc dot gnu dot org
  3 siblings, 0 replies; 5+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-11-03 18:14 UTC (permalink / raw)
  To: gcc-bugs


------- Additional Comments From pinskia at gcc dot gnu dot org  2004-11-03 18:14 -------
For PPC, we have:
        cmplw cr7,r0,r9
        cmpw cr6,r0,r9
        bgtlr- cr7
        beq- cr6,L6
equality test can be done logically or arthimaticly so we don't need both cmp*'s.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2004-05-30 03:50:25         |2004-11-03 18:14:14
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14237


^ permalink raw reply	[flat|nested] 5+ messages in thread
end of thread, other threads:[~2004-11-03 18:14 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-02-21 10:53 [Bug optimization/14237] New: suboptimal comparison between long and long long kazu at cs dot umass dot edu
2004-02-23  1:23 ` [Bug optimization/14237] " pinskia at gcc dot gnu dot org
2004-02-23  1:36 ` kazu at cs dot umass dot edu
2004-02-29  4:10 ` kazu at cs dot umass dot edu
2004-11-03 18:14 ` [Bug rtl-optimization/14237] " pinskia at gcc dot gnu dot org

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).