david@ztivax.UUCP (07/10/86)
>... > Chris Tork asks: `are virtual machines a good thing'? > >Anybody ever work on a Nanodata? Yup. Sort of like writing an assembler emulator using front panel lights for debugging. Not easy Took about as long to develop an emulation as to build the hardware. Maybe longer. Not very fast at execution time, at least by today's standards - no, even by mid 70's standards. It never worked, either. As you said, software emulations of systems execute slowly. If they are written in a high level language (well, C seems pretty good), then they are developed quickly, but slow. At least you have SOMETHING, and you can have a very good debug environment for embedded applications. For efficiency, probably something like a RISC machine with a very orthagonal instruction set, and FAST general purpose instructions would be very good. What needs to be thought about is a good debug environment. We had to develop a special purpose environment for each emulator on the QM-BA (Boat Anchor). We just had a language (SIMPL-Q, pascal-like) which could read data anywhere in the target address space, and display it in "any" format. Floating point still had to be treated specially. David Smyth seismo!unido!ztivax!david