IEEE floating point pitfalls introductory manual

Question

Is there a good introductory level documentation on floating point related practical pitfalls? I need to give some reading material on this subject to a fairly bright CS student.

I remember seeing a good comprehensive manual somewhere couple of years ago, but now, when I need it, I can't find one.

Answer 1

I'd recommend starting with Floating point numbers are a leaky abstraction then reading Anatomy of a floating point number.

Perhaps the manual you're thinking of is What Every Computer Scientist Should Know About Floating-Point Arithmetic.

Answer 2

See http://stackoverflow.com/questions/1089018/why-cant-decimal-numbers-be-represented-exactly-in-binary

http://stackoverflow.com/questions/963873/1-265-10000-126499-99999999999

These answers have numerous good references.

For more information, see http://stackoverflow.com/questions/tagged/floating-point

Answer 3

http://arxiv.org/pdf/cs/0701192v5