--Ed Post, Real Programmers Don't Use Pascal, 1982
http://www.pbm.com/~lindahl/real.programmers.html
This is a prototype of a base-100 floating point library that I plan to implement in 6502 assembly language. It treats a number as a 12-digit coefficient times 100 raised to a positive or negative power. The coefficient is stored as a sequence of six pairs of digits from 0 to 99.
Known limits as of today:
- No sign bit (as my target application doesn't need negative and positive numbers in the same context)
- No floating-point division or square root (likewise not needed for my app)
- Ill-defined behavior on overflow during addition or multiplication when numbers exceed a quadragintillion (10^123)
Future:
- Have some other smart people play with it to see what I messed up (that's you)
- Rewrite in 6502 assembly, after which I'll have this topic moved to NESdev
- Triangular and/or square roots, which I mentioned in a question on Stack Overflow