----- [email protected] a écrit : > On Tue, 22 Mar 2011 12:55:05 +0100 (CET), Tristan Gingold > <[email protected]> wrote: > >> What kind of high-level optimizations do you have in mind ? > > > > Being able to merge processes, being able to statically schedule > > processes... > > > > For example, if a process G is the only driver to a signal S and > > process R is sensitive to S, > > R can be directly scheduled after G. > > Wonderful ! I'm eager to see it :-)
No problems, once it is implemented... > However I imagine that it is an elaboration-time optimisation, > not compile-time. So imagine that the elaboration code has to : > * prepare the stack and frame pointer (with a "leave" sequence) > to adjust to depth R-G at the end of process G > * jump directly to R, past the stack/frame pointer entry point > > What else needs to be done ? Elaboration must be statically done to be able to do global scheduling. Tristan. _______________________________________________ Ghdl-discuss mailing list [email protected] https://mail.gna.org/listinfo/ghdl-discuss
