the venerable 80186 (and 80188) (development system with 188)

Chris M wrote: We had a pair of cards at Peer Protocols, the 5000 and 7000 development systems. They were used to run SCSI test diagnostics and had ISA interfaces on the cards. They had 188's which were controllable from the host machine via a couple of gates. They were held reset at power on. There was 512k of memory that the 188 had, and a 4 k shared window to the host's isa I/O range which was strappable. We would load most of the 512k when starting the development environment, and then hit reset and release the 188. Also had 2 serial lines for initial development environment work. I don't recall the exact debugger, but it is still around. It had a 256 byte or so footprint for grabbing the debug, which we included in the support environment. Original work was thru a serial port to a second PC. Later after the code stabilized, I added a virtual com port that could run on the same PC that talked to a virtual port on the 188's shared memory (4 bytes with flag / data each way). Main challenge up till windows 95 and 98 came out was memory on the DOS side to run the debugger + the host side stack. The host side could be run as either a library linked into applications that run under dos, or could be loaded as a TSR with a TSR call. The DOS TSR was about 300K so a full system of memory on the AT compatible was required to run the thing. Later memory options for expansion were able to be configured to get the AT DOS memory up to the 768K range and still not step on the 4k memory window, and that made things lot easier. Windows 95 and 98 allowed the debugger to run in an 8086 box and still see the shared memory window, so we could debug, and run the full system all on one box. Also had a couple of Toshiba 3100's with ISA boxes which were used for demos / portable development environments. Those were really nice early portable plasma screen "laptops" in the day. Our main code was compiled with Microsoft C6 which was preferred by more developers at the time than Borland, though we maintained Borland compatability. We would have used the 186, but the 188 had a simpler bus and our hardware guys preferred it.