Reading advice for Algorithm Implementation

The course is defined by the slides. Some slides contain details which I certainly not will ask on the exam, e.g. various versions of Power instructions for adding numbers or reading from memory and similar details. Principles are more important.

I will also not ask anything about SSA-form or the advanced parts (transformations of loops) on what optimizing compilers can do --- not that it is unimportant for this course but because it would be unfair since there is another course on that subject.

Below are advice on what is important to achieve a certain grade, i.e. if you wish to pass you should study at least those pages marked as 3.

3 = everybody should know this

4 = more advanced

5 = even more advanced

S = skip, e.g. due to it's covered by another course

The basic idea for what to prioritize is to understand principles and not details --- however, what is a detail to some might be a principle to me. Why programs which exhibit locality of references can exploit cache memories better is a principle --- opcodes, the number of rename registers in each register file of the 970MP, and library function parameter types and names are details. Why superscalar processors need branch prediction, rename registers, and a reorder buffer are principles.