<memory mapping would be a mite silly, wouldn't it, given that the HERC use <64K of refresh RAM. I'd say the way to do it is to use a strategy put fort No, you partition in to 16k quads and your off. <by the guys who designed the 6545, the '65xx' version of the 6845. It had <an update register which essentially allowed you to write an address to th <chip and then send a stream of characters, kind of like the cursor <addressing on a terminal. Run a kaypro some time... slower than sludge video. <Some scheme like that would work even in the limited I/O space of a Z-80. <Another option might be to bank the BIOS such that when it's writing to or <reading from the video page, it has a space equal to the disk buffers, whic <works as a window in the display RAM. The refresh of the display would kee <the entire video memory refreshed, while the banking scheme could minimize <the memory consumption of the video device. Full circle. Hence the need for a MMU. Me I'd never use the 6845/herc in graphic modes. For text the Herc cards only need 4k. <The PC certainly has made for cheaper serial/parallel interfaces than we <ever saw for the S-100. It's too bad there were never any truly general Volume will do that. <purpose parallel ports use commonly enough on the PC to make them cheap. <It's only since the adoption of the 1284 standard that bidirectional I/O vi <the printer port has been practical. It's too bad there aren't <i8255-equivalents which can drive something. The problem with my IMSAI <PIO-6 is that it won't even drive an LED directly. That's why things are buffered. the 8255 will drive a few feet of ribbon but generally thngs like ground loops and other anomolies make a buffered interface more robust. At the time of the 8255 heyday the output current was limited by the Nmos process of the time to only a few mA. Now, the CMOS processes can easily do 30-70 mA but... you can still fry them. Allison