[Bug tree-optimization/17687] sincos can be folded at the tree level

[Bug tree-optimization/17687] sincos can be folded at the tree level

[bonzini at gnu dot org]

------- Comment #7 from bonzini at gnu dot org  2006-08-10 06:28 -------
Was this implemented?  There is an expand_builtin_sincos, but I think it cannot
remove TREE_ADDRESSABLE on the arguments, making most of the performance
advantages go away...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos can be folded at the tree level

[rguenth at gcc dot gnu dot org]

------- Comment #8 from rguenth at gcc dot gnu dot org  2006-08-10 07:41 -------
I implemented the expander to allow the x87 backend using fsincos for it.  The
problems (TREE_ADDRESSABLE, no real folding, no CSE with previous sin/cos)
still exist.  I had/have (I don't remember if I submitted them) patches to
canonicalize sin/cos/sincos to cexp and expand cexp to sin/cos/sincos dependent
on arguments.

If this PR was only about x87 using fsincos for sincos this is fixed now.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos can be folded at the tree level

[paolo dot bonzini at lu dot unisi dot ch]

------- Comment #9 from paolo dot bonzini at lu dot unisi dot ch  2006-08-10 08:04 -------
Subject: Re:  sincos can be folded at the tree
 level


> If this PR was only about x87 using fsincos for sincos this is fixed now.
Well, it was mostly about getting rid of TREE_ADDRESSABLE, which you can 
really do efficiently only on x87, using fsincos.

But maybe it's time to change the subject.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos expansion is suboptimal

[bonzini at gnu dot org]

------- Comment #10 from bonzini at gnu dot org  2006-08-10 08:08 -------
see patch at http://gcc.gnu.org/ml/gcc-patches/2005-12/msg01151.html


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|sincos can be folded at the |sincos expansion is
                   |tree level                  |suboptimal


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos expansion is suboptimal

[bonzini at gnu dot org]

------- Comment #11 from bonzini at gnu dot org  2006-08-10 08:11 -------
If all that we care about is TREE_ADDRESSABLE, and not folding together with a
previous sin/cos call, we may just change

  sincos (x, &s, &c);

to

  sincos (x, &t1, &t2);
  s = t1;
  c = t2;

maybe?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos expansion is suboptimal

[rguenth at gcc dot gnu dot org]

------- Comment #12 from rguenth at gcc dot gnu dot org  2006-08-10 08:16 -------
Created an attachment (id=12055)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=12055&action=view)
patch using cexp

Well, then you make t1 and t2 addressable.  We could introduce a
__builtin_sane_sincos which returns in a vector, but then at expand time we'd
have the same problems if we ever want to go back to a sincos libcall.  With
libgcc-math we could provide one with a sane interface though.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos expansion is suboptimal

[bonzini at gnu dot org]

------- Comment #13 from bonzini at gnu dot org  2006-08-10 08:32 -------
sure, but t1 and t2 die the moment they are assigned back.  it would be just a
trick to return in memory, but not make s and c addressable all the way down to
RA.

though i don't remember how big a penalty is if your variable is addressable
just for the sake of one or two uses.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos tree representation causes extra addressable vars

[rguenth at gcc dot gnu dot org]

------- Comment #14 from rguenth at gcc dot gnu dot org  2006-10-24 12:55 -------
This PR is only about a non-optimal tree-representation for __builtin_sincos.
See also http://gcc.gnu.org/ml/gcc-patches/2005-12/msg01151.html for an
alternative and some discussion.

(The other patch was rejected, I believe further discussion is necessary here.
For reference:
http://gcc.gnu.org/ml/gcc-patches/2005-03/msg01571.html
)


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|http://gcc.gnu.org/ml/gcc-  |
                   |patches/2005-               |
                   |03/msg01571.html            |
           Keywords|patch                       |
   Last reconfirmed|2005-12-09 04:25:17         |2006-10-24 12:55:11
               date|                            |
            Summary|sincos expansion is         |sincos tree representation
                   |suboptimal                  |causes extra addressable
                   |                            |vars


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos tree representation causes extra addressable vars

[rguenth at gcc dot gnu dot org]

------- Comment #15 from rguenth at gcc dot gnu dot org  2006-12-06 09:54 -------
Paolo, are you working on this?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos tree representation causes extra addressable vars

[paolo dot bonzini at lu dot unisi dot ch]

------- Comment #16 from paolo dot bonzini at lu dot unisi dot ch  2006-12-06 09:58 -------
Subject: Re:  sincos tree representation causes
 extra addressable vars


> Paolo, are you working on this?

No. :-(


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos tree representation causes extra addressable vars

[rguenth at gcc dot gnu dot org]

------- Comment #17 from rguenth at gcc dot gnu dot org  2006-12-06 10:04 -------
So I hope you don't mind...


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|bonzini at gnu dot org      |rguenth at gcc dot gnu dot
                   |                            |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos tree representation causes extra addressable vars

[patchapp at dberlin dot org]

------- Comment #18 from patchapp at dberlin dot org  2006-12-07 12:50 -------
Subject: Bug number PR17687

A patch for this bug has been added to the patch tracker.
The mailing list url for the patch is
http://gcc.gnu.org/ml/gcc-patches/2006-12/msg00476.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos tree representation causes extra addressable vars

[patchapp at dberlin dot org]

------- Comment #19 from patchapp at dberlin dot org  2006-12-08 13:10 -------
Subject: Bug number PR17687

A patch for this bug has been added to the patch tracker.
The mailing list url for the patch is
http://gcc.gnu.org/ml/gcc-patches/2006-12/msg00536.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687

[Bug tree-optimization/17687] sincos tree representation causes extra addressable vars

[rguenth at gcc dot gnu dot org]

------- Comment #20 from rguenth at gcc dot gnu dot org  2006-12-13 10:27 -------
Fixed.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687