public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug middle-end/35341] Early exit loop with short known trip count not unrolled
       [not found] <bug-35341-4@http.gcc.gnu.org/bugzilla/>
@ 2015-03-24  6:37 ` aldot at gcc dot gnu.org
  2022-01-10 11:06 ` [Bug tree-optimization/35341] " pinskia at gcc dot gnu.org
  1 sibling, 0 replies; 2+ messages in thread
From: aldot at gcc dot gnu.org @ 2015-03-24  6:37 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=35341

Bernhard Reutner-Fischer <aldot at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |aldot at gcc dot gnu.org

--- Comment #1 from Bernhard Reutner-Fischer <aldot at gcc dot gnu.org> ---
Did you forget to specify -funroll-loops?

gcc-4.2 -O2 -funroll-loops and gcc-4.4 as well as 5.0 with these options
basically generate this optimized dump, which IIUC is what you want.

;; Function foo (foo)

Analyzing Edge Insertions.
foo ()
{
  int temp.34;
  int temp.31;
  int temp.29;
  int temp.26;
  int temp.25;

<bb 2>:
  temp.25 = a[0];
  temp.26 = temp.25 + temp.25;
  a[0] = temp.26;
  if (temp.26 == 10)
    goto <bb 7>;
  else
    goto <bb 3>;

<bb 3>:
  temp.29 = a[2] + a[1];
  a[2] = temp.29;
  if (temp.29 == 10)
    goto <bb 7>;
  else
    goto <bb 4>;

<bb 4>:
  temp.31 = temp.29 + a[4];
  a[4] = temp.31;
  if (temp.31 == 10)
    goto <bb 7>;
  else
    goto <bb 5>;

<bb 5>:
  temp.34 = a[6] + a[3];
  a[6] = temp.34;
  if (temp.34 == 10)
    goto <bb 7>;
  else
    goto <bb 6>;

<bb 6>:
  a[8] = [plus_expr] a[8] + a[4];

<bb 7>:
  return 0;

}


^ permalink raw reply	[flat|nested] 2+ messages in thread

* [Bug tree-optimization/35341] Early exit loop with short known trip count not unrolled
       [not found] <bug-35341-4@http.gcc.gnu.org/bugzilla/>
  2015-03-24  6:37 ` [Bug middle-end/35341] Early exit loop with short known trip count not unrolled aldot at gcc dot gnu.org
@ 2022-01-10 11:06 ` pinskia at gcc dot gnu.org
  1 sibling, 0 replies; 2+ messages in thread
From: pinskia at gcc dot gnu.org @ 2022-01-10 11:06 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=35341

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Currently we get at -O2:

Estimating sizes for loop 1
 BB: 3, after_exit: 0
  size:   1 _1 = i_13 * 2;
   Induction variable computation will be folded away.
  size:   1 _2 = a[_1];
  size:   1 _3 = a[i_13];
  size:   1 _4 = _2 + _3;
  size:   1 a[_1] = _4;
  size:   2 if (_4 == 10)
 BB: 6, after_exit: 1
 BB: 4, after_exit: 0
  size:   1 i_10 = i_13 + 1;
   Induction variable computation will be folded away.
  size:   1 ivtmp_7 = ivtmp_12 - 1;
   Induction variable computation will be folded away.
  size:   2 if (ivtmp_7 != 0)
   Exit condition will be eliminated in peeled copies.
   Exit condition will be eliminated in last copy.
   Constant conditional.
size: 11-5, last_iteration: 11-5
  Loop size: 11
  Estimated size after unrolling: 20
Not unrolling loop 1: size would grow.

We do unroll at -O3.

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2022-01-10 11:06 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <bug-35341-4@http.gcc.gnu.org/bugzilla/>
2015-03-24  6:37 ` [Bug middle-end/35341] Early exit loop with short known trip count not unrolled aldot at gcc dot gnu.org
2022-01-10 11:06 ` [Bug tree-optimization/35341] " pinskia at gcc dot gnu.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).