[Bug middle-end/27139] Optimize double INT->FP->INT conversions with -ffast-math

[Bug middle-end/27139] Optimize double INT->FP->INT conversions with -ffast-math

[marc.glisse at normalesup dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27139

Marc Glisse <marc.glisse at normalesup dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |marc.glisse at normalesup
                   |                            |dot org

--- Comment #2 from Marc Glisse <marc.glisse at normalesup dot org> 2012-04-24 06:35:43 UTC ---
(In reply to comment #0)
> int test (int a)
> {
>         return (double) a;
> }

I just wrote the very same testcase today, extracted from my code...

> Produces:
> 
>         cvtsi2sd        %edi, %xmm0
>         cvttsd2si       %xmm0, %eax
>         ret

Still does. Did you have any idea how to handle it?

> However, following code does the same (at least for -ffast-math):
>         movl    %edi, %eax
>         ret

I don't think -ffast-math is relevant here, on x86 the int->double conversion
is exact hence the reverse has to be as well.

(In reply to comment #1)
> Confirmed, I doubt this shows up that much anyways.

Just posting to mention that it does show up...

[Bug middle-end/27139] Optimize double INT->FP->INT conversions with -ffast-math

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27139

--- Comment #3 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-04-27 10:34:18 UTC ---
Author: rguenth
Date: Fri Apr 27 10:34:13 2012
New Revision: 186898

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=186898
Log:
2012-04-27  Marc Glisse  <marc.glisse@inria.fr>

    PR middle-end/27139
    * tree-ssa-forwprop.c (combine_conversions): Handle INT->FP->INT.

    * gcc.dg/tree-ssa/forwprop-18.c: New test.

Added:
    trunk/gcc/testsuite/gcc.dg/tree-ssa/forwprop-18.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree-ssa-forwprop.c

[Bug middle-end/27139] Optimize double INT->FP->INT conversions with -ffast-math

[marc.glisse at normalesup dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27139

--- Comment #4 from Marc Glisse <marc.glisse at normalesup dot org> 2012-05-01 09:32:25 UTC ---
Hello Uros,
is there any other case you think should be handled, or should we close the
bug?

[Bug middle-end/27139] Optimize double INT->FP->INT conversions with -ffast-math

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27139

glisse at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |glisse at gcc dot gnu.org
         Resolution|                            |FIXED
   Target Milestone|---                         |4.8.0

--- Comment #5 from glisse at gcc dot gnu.org 2012-05-05 00:20:08 UTC ---
Closing then.

[Bug middle-end/27139] Optimize double INT->FP->INT conversions with -ffast-math

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-04-12 23:59 -------
Confirmed, I doubt this shows up that much anyways.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
 GCC target triplet|x86_64-unknown-linux-gnu    |
           Keywords|                            |missed-optimization
   Last reconfirmed|0000-00-00 00:00:00         |2006-04-12 23:59:25
               date|                            |
            Summary|Optimize double INT->FP->INT|Optimize double INT->FP->INT
                   |conversions                 |conversions with -ffast-math


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27139