https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83253
--- Comment #12 from Bill Schmidt <wschmidt at gcc dot gnu.org> --- By contrast, on powerpc64le, we see: - mult_by_coeff_cost (3, E_SImode, true) returns a cost of 8 - mult_by_coeff_cost (4, E_SImode, true) returns a cost of 4 These are the sort of costs one would expect, and as a result we don't make the poor replacement. So each of the bad targets needs to look into their cost models to see what is going wrong.
