Den Tue, 16 Apr 2019 16:37:43 +0200 skrev Michał Pleban <lists_at_michau.name>: > Mia Magnusson wrote: > > > Something that Michal and you might want conteplate: Would there be > > any benefit for the new 8088 support software if it were possible > > to hide the $0/$1 registers in '816 native mode? > > I am wondering about this. That would necessarily mean that the IPC > library would run in '816 mode, which would mean that there could be > no KERNAL calls unless the KERNAL is rewritten for the '816? > > > (For example, when emulating an MDA display adapter, the first > > char and its corresponding attribute can > > afaik not be read by the 65xx processor so any transfer from the > > emulated MDA display to the CBM-II native display would need at > > least two bytes copied to some other location by the 8088 > > processor. > > Yes, this is how it's currently done. > > > Also disk operations that read/write memory feely can't access the > > first two bytes in every 64k block. Maybe the benefit of being able > > to hide those two registers is less than the cost (in work and chip > > usage) to actually implement this though. Sorry if I'm bringing up > > stuff that we've already discussed. > > But disk reads call the KERNAL IEEE routines extensively, so it would > be very nontrivial to use them in '816 mode? How "expensive" is it to switch to/from '816 mode? If it's just one instruction it might not be much of an issue, but if it's more complicated it might not be worth the effort. On a system with plenty of ram parts of the ieee routines could be duplicated in ram. Btw, recently the blog Pagetable.com got some really nice articles about the Commodore peripheral buses. -- (\_/) Copy the bunny to your mails to help (O.o) him achieve world domination. (> <) Come join the dark side. /_|_\ We have cookies.Received on 2020-05-29 21:33:16
Archive generated by hypermail 2.3.0.