AM and ADM destinations

[...]

So, as far as I reached, the AM and ADM instructions store data in the A register and in the memory, and in case of ADM also in the D register. But when trying to recode this, should I change the A register before or after commiting changes to the RAM?

[WBahn]

... wrote

So, as far as I reached, the AM and ADM instructions store data in the A register and in the memory, and in case of ADM also in the D register. But when trying to recode this, should I change the A register before or after commiting changes to the RAM?

After. All register/memory writes reflect the values that were present at the end of the prior instruction.

Be aware that one of the author's emulators violates this behavior. I think it is the A register. I don't recall the details or if the current version of that tool has been fixed. It's a bug that only rears it's head in very unusual circumstances, none of which are relevant to any of the N2T projects, which is why it went undiscovered for so long.

[dolomiti7]

The register updates should happen after all other changes. The original emulator was correct for instructions like AM=M+1, with the semantics that M refers to [A_before]. E.g. AM=M+1 => A_after=RAM[A_before]+1; RAM[A_before]=RAM[A_before]+1

What the Java based emulator didn't implement correctly are corner cases like:

A=0;JMP, where PC should be A_before, but the emulator would jump to address 0 instead. This has been fixed in the web-based emulator though.