On Mon, Sep 8, 2025 at 7:14 PM Hongtao Liu <crazy...@gmail.com> wrote:
>
> On Tue, Sep 9, 2025 at 4:26 AM H.J. Lu <hjl.to...@gmail.com> wrote:
> >
> > Enable SSE4.1 ceil/floor/trunc for -Os to replace a function call with
> > roundss or roundsd.
> >
> > gcc/
> >
> >         PR target/121861
> >         * config/i386/i386.cc (ix86_optab_supported_p): Return true for
> >         SSE4.1 with SSE math for floor_optab, ceil_optab and
> >         btrunc_optab.
> >
> > gcc/testsuite/
> >
> >         PR target/121861
> >         * gcc.target/i386/pr121861-1a.c: New file.
> >         * gcc.target/i386/pr121861-1b.c: Likewise.
> >
> > Signed-off-by: H.J. Lu <hjl.to...@gmail.com>
> > ---
> >  gcc/config/i386/i386.cc                     |  9 ++---
> >  gcc/testsuite/gcc.target/i386/pr121861-1a.c | 43 +++++++++++++++++++++
> >  gcc/testsuite/gcc.target/i386/pr121861-1b.c |  7 ++++
> >  3 files changed, 54 insertions(+), 5 deletions(-)
> >  create mode 100644 gcc/testsuite/gcc.target/i386/pr121861-1a.c
> >  create mode 100644 gcc/testsuite/gcc.target/i386/pr121861-1b.c
> >
> > diff --git a/gcc/config/i386/i386.cc b/gcc/config/i386/i386.cc
> > index 471be3e8615..c8626f51973 100644
> > --- a/gcc/config/i386/i386.cc
> > +++ b/gcc/config/i386/i386.cc
> > @@ -27189,11 +27189,10 @@ ix86_optab_supported_p (int op, machine_mode 
> > mode1, machine_mode,
> >      case floor_optab:
> >      case ceil_optab:
> >      case btrunc_optab:
> > -      if (((SSE_FLOAT_MODE_P (mode1)
> > -           && TARGET_SSE_MATH
> > -           && TARGET_SSE4_1)
> > -          || mode1 == HFmode)
> > -         && !flag_trapping_math)
> > +      if ((SSE_FLOAT_MODE_P (mode1)
> > +          && TARGET_SSE_MATH
> > +          && TARGET_SSE4_1)
> > +         || (mode1 == HFmode && !flag_trapping_math))
>
> (mode1 == HFmode && TARGET_AVX512FP16)?
> <rounding_insn>hf2 is defined under TARGET_AVX512FP16 w/o !flag_trapping_math.

Dropped the !flag_trapping_math check in the v2 patch.

> >         return true;
> >        return opt_type == OPTIMIZE_FOR_SPEED;
> >


-- 
H.J.

Reply via email to