Swedish

Every other autumn semester

The purpose of the course is that the students should understand the possibilities and limitations of modern optimising compilers, and thereby understand what they must optimise manually and what the compiler can do automatically.

Control-flow analysis, data-flow analysis, dependence analysis, alias analysis, redundancy elimination, loop optimisations, procedure optimisations, register allocation, code scheduling, optimisation for object oriented languages, optimisation of the memory hierarchy and vectorisation for processors with SIMD instructions.

understand which algorithms and data structures are used when modern optimising compilers are implemented, and
understand which language constucts limit the compiler's ability to optimise code.

be able to analyse benchmark programs in order to understand which optimisation algorithms are suitable to implement, and
be able to implement optimisation algorithms on SSA form, validate that they are implemented correctly and to measure their performance effects.

without assistance be able to read and understand a scientific article on optimising compilers and to present it to the other students,
be able to make a rough estimation of how long it may take to implement the ideas of the article, and
make a judgement of whether that amount of work is worthwhile,
and finally to determine whether the optimisation ideas can be suitable to perform by hand, if the compiler cannot do it automatically.

Lectures

Laboratory exercises

Written exam

Failed, pass

EDAF05 Algorithms, data structures and complexity or EDA027 Algorithms and Data Structures

Skeppstedt, J.: An Introduction to the Theory of Optimizing Compilers. Skeppberg, 2016. ISBN 9789197794039.

EDAN75F

 -03-22

Professor Thomas Johansson