[Bug tree-optimization/56873] New: vector shift lowered to scalars

[Bug tree-optimization/56873] New: vector shift lowered to scalars

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56873

             Bug #: 56873
           Summary: vector shift lowered to scalars
    Classification: Unclassified
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: glisse@gcc.gnu.org
            Target: x86_64-linux-gnu


#define SIZE 32
typedef long long veci __attribute__((vector_size(SIZE)));

veci f(veci a, veci b){
  return a>>b;
}

Compiling this with -O3 -mxop, the vector lowering pass gives scalar
operations. However, if SIZE is either 16 or 64, I get vector shifts of size
16.

[Bug middle-end/56873] vector shift lowered to scalars

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56873

Marc Glisse <glisse at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|tree-optimization           |middle-end

--- Comment #1 from Marc Glisse <glisse at gcc dot gnu.org> ---
In the vector lowering pass, we only try to find a smaller vector that supports
the operation if the larger vector is not a supported mode (never mind whether
it supports the operation or not). The fact that the x86 backend says V4DI is
supported seems strange to me, but let's continue. There is no reason why a
target couldn't support some operations on V4DI and more on V2DI, so lowering
still makes sense even from a supported mode. I quickly tried a hack in
tree-vect-generic.c:

  /* For very wide vectors, try using a smaller vector mode.  */
  compute_type = type;
  if (op && (!VECTOR_MODE_P (TYPE_MODE (type))
             || optab_handler (op, TYPE_MODE (type)) == CODE_FOR_nothing))

But we then ICE in expr.c:

                /* Vector CONSTRUCTORs should only be built from smaller
                   vectors in the case of BLKmode vectors.  */
                gcc_assert (TREE_CODE (TREE_TYPE (value)) != VECTOR_TYPE);

And removing the assertion later fails with an unrecognizable insn:

(insn 18 17 19 2 (set (reg:V2DI 105)
        (vec_merge:V2DI (vec_duplicate:V2DI (reg:V2DI 102 [ D.2163 ]))
            (reg:V2DI 105)
            (const_int 1 [0x1]))) a.c:5 -1
     (nil))
a.c:6:1: internal compiler error: in extract_insn, at recog.c:2161

(looks like it got confused and lost the V4DI mode)

So this is not a simple tree-vect-generic issue :-(

I don't think we have any front-end or optimization producing a constructor
v4di{v2di,v2di}, but that's more luck than anything else...

[Bug middle-end/56873] vector shift lowered to scalars

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56873

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
This looks to be fixed in GCC 5.1.

[Bug middle-end/56873] vector shift lowered to scalars

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56873

Marc Glisse <glisse at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #3 from Marc Glisse <glisse at gcc dot gnu.org> ---
Indeed, I now get sensible code with -mxop. Not so with -mavx2, but that seems
independent and I filed it as PR 101611.

[Bug middle-end/56873] vector shift lowered to scalars for -mxop

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56873

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |5.0
            Summary|vector shift lowered to     |vector shift lowered to
                   |scalars                     |scalars for -mxop