[Bug tree-optimization/119096] Loop with conditional, cast and reduction vectorized incorrectly with AVX-512

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119096

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Gimple level:
  vect__4.8_45 = MEM <vector(16) int> [(int *)A_15(D)];
  vect__10.16_54 = [vec_unpack_lo_expr] vect__4.8_45;
  vect__10.16_55 = [vec_unpack_hi_expr] vect__4.8_45;
  mask__5.9_46 = vect__4.8_45 > { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0 };
  vect__7.12_49 = MEM <vector(16) int> [(int *)B_16(D)];
  mask__8.13_50 = vect__7.12_49 > { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0 };
  mask__9.14_51 = mask__5.9_46 & mask__8.13_50;
  mask_patt_37.15_53 = [vec_unpack_hi_expr] mask__9.14_51; // Only use the
upper half
  vect_patt_36.18_58 = .COND_ADD (mask_patt_37.15_53, vect__10.16_54,
vect__10.16_55, vect__10.16_55);
  _60 = .REDUC_PLUS (vect_patt_36.18_58); [tail call]