This file lists known differences and probably differences
between the real KDJ11-A (that is, the KDJ11-A I have at home) 
and the emulator. Only processor differences are listed, 
there are much more in the peripherals.

Known
-----
- Instruction timing, of course.

- KDJ11-A has no display register. The emulator has one which can be disabled.

- There are only mode 1 and 2 power-up options. These are controlled
  by a command line switch rather than a jumper :-)

- A memory trap during vector fetch halts the emulator but not the KDJ11-A.

- The emulator has no cache memory (wouldn't it be nice??). All related
  registers read constant values. The maintenance bits are not implemented.

- No parity errors (Do you need them?).

- The BEVNT line can not be re-wired to an external interrupt.

- No HALT option jumper.

- No trap to ODT if a break is seen on the console.

Probably
--------
- There may be differences in the priority of interrupts. There are a lot
  of hard-to-check situations.

- rtt'ing to an rtt with T-bit set.

- register contents (MMR1/2 and FPU) if a operand fetch MMU-aborts in
  the middle of an operand.

- floating point accuracy. There may be differences in the last bit.

- I'm not sure about the contents of MMR1. KDJ11-A documentation states that
  'explicite references trough PC' are recorded. I was not able to observe this.
  (See Tests/mmr.s and check it on your processor).

- The register contents for auto-increment/-decrement operations may
  not the same as in the real processor when the MMU aborts the instruction.
  If you use a backup technique that tries to guess (like the one used
  in V7 for machines without MMR1) you may lose. Using the implemented MMR1
  should give the right results.
