> I wouldn't just put the -fschedule-insns or -fschedule-insns2 in without > a pretty careful test. In the past the non-default optimization flags > would sometimes cause slowdowns and sometimes cause bugs. I'll run the regressions. This only really matters for -native false, and the PolySpace guys have the options in and are the main (only?) user of the C backend, so I figure they're ok.