On Mon, Apr 23, 2012 at 11:34 AM, R. Diez <[email protected]> wrote: > > >> Whether or not you have delay slots needs to be known at compile time... > > Not necessarily. If you emit an l.nop after every jump, the code will run on > CPUs with and without the delay slot. > > Removing the l.nop on CPUs which don't have it is just an optimisation. >
It's not really feasible to take advantage of it at run time without having a second copy of the code, though. -Pete _______________________________________________ OpenRISC mailing list [email protected] http://lists.openrisc.net/listinfo/openrisc
