Floating-Point accuracy

The general idea

Floating-point arithmetic is necessarily approximate. A large part of floating-point hardware design and numeric algorithm design relates to handling errors and approximations.

Some of these approximations are significant to final results. In this project you need to explore conditions under which errors and approximations are significant, and relate these to chosen examples.

The challenge

Error conditions do not occur at random but under specific conditions. Some of these conditions are easy to trigger, others are more subtle. Issues you will need to consider include

creating artificial examples to demonstrate particular inaccuracies or errors
understanding the specific implementation you are using
finding useful examples where the errors or inaccuracies could occur and exploring ways to limit or avoid these problems
there are other properties of caches like associativity which will be more of a challenge to measure

The language chosen is not significant in this project, provided it includes a standard implementation of floating point.