Re: [PATCH v4 12/12] constructor: Check if it is faster to load constant from memory

["H.J. Lu" <hjl.tools@gmail.com>]

On Wed, May 19, 2021 at 2:33 AM Richard Biener
<richard.guenther@gmail.com> wrote:
>
> On Tue, May 18, 2021 at 9:16 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> >
> > When expanding a constant constructor, don't call expand_constructor if
> > it is more efficient to load the data from the memory via move by pieces.
> >
> > gcc/
> >
> >         PR middle-end/90773
> >         * expr.c (expand_expr_real_1): Don't call expand_constructor if
> >         it is more efficient to load the data from the memory.
> >
> > gcc/testsuite/
> >
> >         PR middle-end/90773
> >         * gcc.target/i386/pr90773-24.c: New test.
> >         * gcc.target/i386/pr90773-25.c: Likewise.
> > ---
> >  gcc/expr.c                                 | 10 ++++++++++
> >  gcc/testsuite/gcc.target/i386/pr90773-24.c | 22 ++++++++++++++++++++++
> >  gcc/testsuite/gcc.target/i386/pr90773-25.c | 20 ++++++++++++++++++++
> >  3 files changed, 52 insertions(+)
> >  create mode 100644 gcc/testsuite/gcc.target/i386/pr90773-24.c
> >  create mode 100644 gcc/testsuite/gcc.target/i386/pr90773-25.c
> >
> > diff --git a/gcc/expr.c b/gcc/expr.c
> > index d09ee42e262..80e01ea1cbe 100644
> > --- a/gcc/expr.c
> > +++ b/gcc/expr.c
> > @@ -10886,6 +10886,16 @@ expand_expr_real_1 (tree exp, rtx target, machine_mode tmode,
> >                 unsigned HOST_WIDE_INT ix;
> >                 tree field, value;
> >
> > +               /* Check if it is more efficient to load the data from
> > +                  the memory directly.  FIXME: How many stores do we
> > +                  need here if not moved by pieces?  */
> > +               unsigned HOST_WIDE_INT bytes
> > +                 = tree_to_uhwi (TYPE_SIZE_UNIT (type));
>
> that's prone to fail - it could be a VLA.

What do you mean by fail?  Is it ICE or missed optimization?
Do you have a testcase?

>
> > +               if ((bytes / UNITS_PER_WORD) > 2
> > +                   && MOVE_MAX_PIECES > UNITS_PER_WORD
> > +                   && can_move_by_pieces (bytes, TYPE_ALIGN (type)))
> > +                 goto normal_inner_ref;
> > +
>
> It looks like you're concerned about aggregate copies but this also handles
> non-aggregates (which on GIMPLE might already be optimized of course).

Here I check if we copy more than 2 words and we can move more than
a word in a single instruction.

> Also you say "if it's cheaper" but I see no cost considerations.  How do
> we generally handle immed const vs. load from constant pool costs?

This trades 2 (update to 8) stores with one load plus one store.  Is there
a way to check which one is faster?

> >                 FOR_EACH_CONSTRUCTOR_ELT (CONSTRUCTOR_ELTS (init), ix,
> >                                           field, value)
> >                   if (tree_int_cst_equal (field, index))
> > diff --git a/gcc/testsuite/gcc.target/i386/pr90773-24.c b/gcc/testsuite/gcc.target/i386/pr90773-24.c
> > new file mode 100644
> > index 00000000000..4a4b62533dc
> > --- /dev/null
> > +++ b/gcc/testsuite/gcc.target/i386/pr90773-24.c
> > @@ -0,0 +1,22 @@
> > +/* { dg-do compile } */
> > +/* { dg-options "-O2 -march=x86-64" } */
> > +
> > +struct S
> > +{
> > +  long long s1 __attribute__ ((aligned (8)));
> > +  unsigned s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13, s14;
> > +};
> > +
> > +const struct S array[] = {
> > +  { 0, 60, 640, 2112543726, 39682, 48, 16, 33, 10, 96, 2, 0, 0, 4 }
> > +};
> > +
> > +void
> > +foo (struct S *x)
> > +{
> > +  x[0] = array[0];
> > +}
> > +/* { dg-final { scan-assembler-times "movups\[\\t \]%xmm\[0-9\]+, \\(%\[\^,\]+\\)" 1 } } */
> > +/* { dg-final { scan-assembler-times "movups\[\\t \]%xmm\[0-9\]+, 16\\(%\[\^,\]+\\)" 1 } } */
> > +/* { dg-final { scan-assembler-times "movups\[\\t \]%xmm\[0-9\]+, 32\\(%\[\^,\]+\\)" 1 } } */
> > +/* { dg-final { scan-assembler-times "movups\[\\t \]%xmm\[0-9\]+, 48\\(%\[\^,\]+\\)" 1 } } */
> > diff --git a/gcc/testsuite/gcc.target/i386/pr90773-25.c b/gcc/testsuite/gcc.target/i386/pr90773-25.c
> > new file mode 100644
> > index 00000000000..2520b670989
> > --- /dev/null
> > +++ b/gcc/testsuite/gcc.target/i386/pr90773-25.c
> > @@ -0,0 +1,20 @@
> > +/* { dg-do compile } */
> > +/* { dg-options "-O2 -march=skylake" } */
> > +
> > +struct S
> > +{
> > +  long long s1 __attribute__ ((aligned (8)));
> > +  unsigned s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13, s14;
> > +};
> > +
> > +const struct S array[] = {
> > +  { 0, 60, 640, 2112543726, 39682, 48, 16, 33, 10, 96, 2, 0, 0, 4 }
> > +};
> > +
> > +void
> > +foo (struct S *x)
> > +{
> > +  x[0] = array[0];
> > +}
> > +/* { dg-final { scan-assembler-times "vmovdqu\[\\t \]%ymm\[0-9\]+, \\(%\[\^,\]+\\)" 1 } } */
> > +/* { dg-final { scan-assembler-times "vmovdqu\[\\t \]%ymm\[0-9\]+, 32\\(%\[\^,\]+\\)" 1 } } */
> > --
> > 2.31.1
> >



-- 
H.J.