Hmm, difficult to spot a pattern, but then with these large 24 bit values you will be getting overflow when dividing by such a small number and try to fit the result into a 12 bit register. I would try with a much larger number as the memory operand, say 0x100. Another thing you can try is using the 12 order to do a multiply and then using the 13 order to do a divide by the same memory operand and see if you get anything like the numbers you started with. Ah that's better. Except I still can't see any pattern except that the Q register is counting up every other time. I wonder if it is doing one step in the division each time and you are supposed to loop and stop when Q reaches a certain value. This would improve maximum interrupt latency. But you say this order takes a long time. Maybe it is just broken as you thought. I'll see what I can come up with. Actually I was employed a junior programmer and eventually became responsible for the Coral 66 compiler and utility software for the 920. Later on I became a 'Senior Systems Engineer' and made some suggestions to the engineers here and there. I really learned about hardware as a hobby with my ICT1301 mainframe. After I left I designed interface cards for micros and my most complex hardware design was a NuBus card for the Macintosh which could transfer up to 64k of data to wide format Versatec Plotters at a million BYTES a second without any intervention from the CPU (i.e. by Direct Memory Access). While that was going on the CPU was imaging the next chunk of data to be sent. It had to be real time because if the plotter ran out of data it would stop and the liquid toner would soak into the paper and spoil the image. Good. Yes the CPU and some memory was in one box and extra memory in others, and lots of peripherals too. Not commercial things of course, things like compasses, gyros, special digital radio receivers and transmitters, custom keyboards and vector and raster monitors, chart recorders and much more. A chap who used to be a service engineer on my computer many years ago contacted a while back and kindly agreed to help me bring my computer back into a useable state. He has a web site describing the project, and there are some pictures of him and my machine at http:// ict1301.co.uk/13010510.htm . The rotary switches can be used to put data into registers, including the three instruction registers, so you can enter programs from the keys if all else fails. The green squares have four indicators in each one, so can display one decimal digit (it is primarily a decimal machine, 12 digits * 4 bits) The machine can be run at full speed, or at three instructions per second, or at one instruction per keypress, or at three clock cycles per second or at one clock cycle per keypress. There are 6 tape decks see http://web.onetel.com/~rodritab/mtani.htm The bigger picture of a complete machine is here: http:// ict1301.co.uk/13012006.htm though it is actually a model and has the earlier one inch tape decks. An actual machine is the lower picture at : http:// www.milsom100.freeserve.co.uk/1301/1301f.htm Oh for all that space! I have two machines in less space they use for one. Mind you one of mine is not assembled. It makes taking pictures very difficult. Roger.