[Bug rtl-optimization/70873] [7 Regressio] 20% performance regression at 482.sphinx3 after r235442 with -O2 -m32 on Haswell.

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70873

--- Comment #14 from H.J. Lu <hjl.tools at gmail dot com> ---
(In reply to Uroš Bizjak from comment #13)
> (In reply to H.J. Lu from comment #12)
> 
> > I still see:
> > 
> >     vcvtss2sd       (%ecx,%eax,4), %xmm5, %xmm5
> > 
> > without vxorpd.
> 
> Maybe vxorpd gets scheduled away from the insn? What is the name of the
> pattern?

We need to disable

define_split
  [(set (match_operand 0 "any_fp_register_operand")
        (float_extend (match_operand 1 "memory_operand")))]
  "reload_completed
   && (GET_MODE (operands[0]) == TFmode
       || GET_MODE (operands[0]) == XFmode
       || GET_MODE (operands[0]) == DFmode)"
  [(set (match_dup 0) (match_dup 2))]
{

for SF->DF.