I’m getting very close to the first real BMOW bootup. I’ve got two data registers wired up, along with enough of the data bus to use them. Now I just need to double-check the current wiring, add the rest of the registers, and see what kind of test program I devise.

My instruction set is copied from the 6502, with some minor additions and changes. In the course of thinking about a good example program for first bootup, a couple of oddities occurred to me for the first time. The biggest one is that there’s no way to add values from two registers. Instead, the ADC instruction always adds the value in the accumulator register to a value in RAM. That’s a problem for me, since I haven’t yet wired up the RAM, but it also seems like a deeper problem. Why wouldn’t you want something like an ADX instruction, to add A + X? Wouldn’t it be faster than a memory access, if the values you want to add are already in registers?

The second oddity is the very non-symmetric nature of the 6502 instruction set. It’s something I was certainly aware of before, but never really thought much about. With these instructions, each register has different capabilities. The results of an add are always stored in A, never X or Y. Indexed memory accesses use X or Y, not A. Only the accumulator can be bit shifted. It all seems arbitrary and awkward, although I’m sure there were good reasons for those limitations in the original 6502 hardware.

All this has got me thinking that maybe I ought to pattern the BMOW instruction set from the 68000, or MIPS, or something else that’s a little more rational. There’s nothing specifically tying BMOW to the 6502, and with a reprogramming of the microcode, I can implement any instruction set I want, as long as it can be realized on the BMOW hardware. For the time being I plan to stick with the current instruction set, though, since I’ve already written most of the microcode, and got a working assembler too. Writing a new assembler is something I particularly don’t relish.

Musings on instruction sets aside, the BMOW construction and wiring is proceeding smoothly, and I probably have more than 50% of it finished now. A lot of things like devices and the stack pointer aren’t strictly necessary in order to run simple programs, so I hope to have some good news about my first bootup with a real program very soon.