https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125305
--- Comment #7 from Drea Pinski <pinskia at gcc dot gnu.org> --- (In reply to Mingcong Bai from comment #6) > For my part (as maintainer of LoongArch port for AOSC OS), I do not think it > is a good idea to forgo instruction scheduling, even at build-time cost. Note the talk about disabling instruction scheduling before Register allocation; not the one afterwards. I suspect the instruction scheduler before the register allocation tries to cleanup some of the stuff that the gimple layer does. The gimple layer does not do much instruction scheduling at all except when it comes to TER. There are bigger issues along the lines about instruction scheduling in general. Even things like IV-OPTS can change due to statements moving around slightly; which I just noticed recently. Though this is all bigger GCC infastructure issues rather than a target issue :).
