Author: elliot (recyledsteel.demon.co.uk)
Date: 12-18-2002 23:36
Why not ask the makers of Steem or other STE emulator, they should have C code to do the routines (i.e. Blits).
Iether way sounds good.
I have never used the PMMU so I do not know how hard it is to re-map. The above code should be able to cope with register alterations too (mid blit). You would only need to check the active register on the interupt. The problem would be the constant checking of the active flag, it would use a lot of CPU time - not sure how the TT-MFP works.
My idea would mean the blit would happen as soon as it was requested (the interupt is the trap). Also there would be no lag when the blitter is not in use, however, the problem here would be with the slow down when trapping ALL changes to blitter registers.
Better still, combine the idea. Re-map the low registers with the PMMU, leave the status reg out and then wait for the trap! No interupts, no lag with trap catching (well apart from one, but you want that).
One problem, the Blitter would (for ease) have to run in HOG mode, unless someone wants to write massive interupt stuff. The TT should be able to Blit way faster than the STE Blitter can anyway (I mean the STE can at some things).
Elliot...
|
