[PATCH v3 2/2] x86: Add vec_duplicate<mode> expander

[H.J. Lu]

On Tue, Jun 8, 2021 at 11:41 PM Uros Bizjak <ubizjak@gmail.com> wrote:
>
> On Tue, Jun 8, 2021 at 7:59 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> >
> > 1. Update vec_duplicate to allow to fail so that backend can only allow
> > broadcasting an integer constant to a vector when broadcast instruction
> > is available.  This can be used by memset expander to avoid vec_duplicate
> > when loading from constant pool is more efficient.
> > 2. Add vec_duplicate<mode> expander and enable vec_duplicate from a
> > non-standard SSE constant integer only if vector broadcast is available.
> >
> >         * config/i386/i386-expand.c (ix86_expand_integer_vec_duplicate):
> >         New function.
> >         * config/i386/i386-protos.h (ix86_expand_integer_vec_duplicat):
> >         New prototype.
> >         * config/i386/sse.md (INT_BROADCAST_MODE): New mode iterator.
> >         (vec_duplicate<mode>): New expander.
> >         * doc/md.texi: Update vec_duplicate.
> > ---
> >  gcc/config/i386/i386-expand.c | 21 +++++++++++++++++++++
> >  gcc/config/i386/i386-protos.h |  1 +
> >  gcc/config/i386/sse.md        | 20 ++++++++++++++++++++
> >  gcc/doc/md.texi               |  2 --
> >  4 files changed, 42 insertions(+), 2 deletions(-)
> >
> > diff --git a/gcc/config/i386/i386-expand.c b/gcc/config/i386/i386-expand.c
> > index 29d96805d9d..145e028353c 100644
> > --- a/gcc/config/i386/i386-expand.c
> > +++ b/gcc/config/i386/i386-expand.c
> > @@ -15669,6 +15669,27 @@ ix86_expand_vector_extract (bool mmx_ok, rtx target, rtx vec, int elt)
> >      }
> >  }
> >
> > +/* Expand integer vec_duplicate.  Return true if successful.  */
> > +
> > +bool
> > +ix86_expand_integer_vec_duplicate (rtx *operands)
> > +{
> > +  /* Enable VEC_DUPLICATE from a non-standard SSE constant integer only
> > +     if vector broadcast is available.  */
> > +  if (CONST_INT_P (operands[1])
> > +      && (!TARGET_AVX2
> > +         || standard_sse_constant_p (operands[1],
> > +                                     GET_MODE (operands[0]))))
> > +    return false;
> > +
> > +  if (!ix86_expand_vector_init_duplicate (false,
> > +                                         GET_MODE (operands[0]),
> > +                                         operands[0], operands[1]))
> > +    gcc_unreachable ();
> > +
> > +  return true;
> > +}
> > +
> >  /* Generate code to copy vector bits i / 2 ... i - 1 from vector SRC
> >     to bits 0 ... i / 2 - 1 of vector DEST, which has the same mode.
> >     The upper bits of DEST are undefined, though they shouldn't cause
> > diff --git a/gcc/config/i386/i386-protos.h b/gcc/config/i386/i386-protos.h
> > index 578750a2532..dc191dc18ec 100644
> > --- a/gcc/config/i386/i386-protos.h
> > +++ b/gcc/config/i386/i386-protos.h
> > @@ -260,6 +260,7 @@ extern void ix86_expand_mul_widen_hilo (rtx, rtx, rtx, bool, bool);
> >  extern void ix86_expand_sse2_mulv4si3 (rtx, rtx, rtx);
> >  extern void ix86_expand_sse2_mulvxdi3 (rtx, rtx, rtx);
> >  extern void ix86_expand_sse2_abs (rtx, rtx);
> > +extern bool ix86_expand_integer_vec_duplicate (rtx *);
> >
> >  /* In i386-c.c  */
> >  extern void ix86_target_macros (void);
> > diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
> > index 2a34756be2a..a227295cc1d 100644
> > --- a/gcc/config/i386/sse.md
> > +++ b/gcc/config/i386/sse.md
> > @@ -24570,3 +24570,23 @@
> >    "TARGET_WIDEKL"
> >    "aes<aeswideklvariant>\t{%0}"
> >    [(set_attr "type" "other")])
> > +
> > +;; Modes handled by broadcast patterns.
> > +(define_mode_iterator INT_BROADCAST_MODE
> > +  [(V64QI "TARGET_AVX512F") (V32QI "TARGET_AVX") V16QI
> > +   (V32HI "TARGET_AVX512F") (V16HI "TARGET_AVX") V8HI
> > +   (V16SI "TARGET_AVX512F") (V8SI "TARGET_AVX") V4SI
> > +   (V8DI "TARGET_AVX512F") (V4DI "TARGET_64BIT") V2DI])
>
> In ix86_convert_const_wide_int_to_broadcast, there is:
>
> +  else if (TARGET_64BIT
> +   && ix86_broadcast (val, GET_MODE_BITSIZE (DImode),
> +      val_broadcast))
>
> So, shouldn't all modes with DImode inner mode have TARGET_64BIT predicate?

Fixed in the v4 patch.

Thanks.

> Uros.
>
> > +;; Broadcast from an integer.  NB: Enable broadcast only if we can move
> > +;; from GPR to SSE register directly.
> > +(define_expand "vec_duplicate<mode>"
> > +  [(set (match_operand:INT_BROADCAST_MODE 0 "register_operand")
> > +       (vec_duplicate:INT_BROADCAST_MODE
> > +         (match_operand:<ssescalarmode> 1 "general_operand")))]
> > +  "TARGET_SSE2 && TARGET_INTER_UNIT_MOVES_TO_VEC"
> > +{
> > +  if (!ix86_expand_integer_vec_duplicate (operands))
> > +    FAIL;
> > +  DONE;
> > +})
> > diff --git a/gcc/doc/md.texi b/gcc/doc/md.texi
> > index 00caf3844cc..e66c41c4779 100644
> > --- a/gcc/doc/md.texi
> > +++ b/gcc/doc/md.texi
> > @@ -5077,8 +5077,6 @@ the mode appropriate for one element of @var{m}.
> >  This pattern only handles duplicates of non-constant inputs.  Constant
> >  vectors go through the @code{mov@var{m}} pattern instead.
> >
> > -This pattern is not allowed to @code{FAIL}.
> > -
> >  @cindex @code{vec_series@var{m}} instruction pattern
> >  @item @samp{vec_series@var{m}}
> >  Initialize vector output operand 0 so that element @var{i} is equal to
> > --
> > 2.31.1
> >



-- 
H.J.