Following the packed data treatment, Hyde moves into a discussion of floating point arithmetic and the shortcomings of trying to perform infinite precision arithmetic, that is to say, math where there are an infinite number of numbers between say zero and one.

I read this bit a few times. It’s fairly dense and I’ll probably read it a few more times and there’s apparently a deeper treatment of it later in the book, which is good. Turns out floating point math on binary systems is not straightforward, precision can be easily lost, especially for addition and subtraction operations, less so for multiplication and division.

There’s a discussion of IEEE floating point format standards for 32-bit, single precision, 64-bit, double precision and 80-bit, extended precision formats.