[Bug target/97286] GCC sometimes uses an extra xmm register for the destination of _mm_blend_ps

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97286

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
simplified to remove intrsinics:
```
typedef int v4si __attribute__((vector_size(4*sizeof(int))));
typedef float v4sf __attribute__((vector_size(4*sizeof(int))));

void f(v4sf *a, v4sf *b, int l)
{
  v4si c = (v4si)*a;
  for(int i = 0;i < l; i++)
  {
    v4sf d = (v4sf)c;
    d+=b[i];
    c = (v4si)d;
  }
  *a = (v4sf)c;
}
```