Questions about implementation of a global register allocator for the tiny c compiler

Question

upcoming summer i will hopefully start writing my masters thesis and i have been quite busy looking for a thesis subject. I now have a pool of subjects that i am interested in and the one that struck me most is the implementation of a global register allocator for the tiny C compiler (graph coloring or linear scan).

So i wanted to drop by and ask if any of you has ever done this and if it is a feasible subject for a master thesis, or if it is too hard. I would also be very happy if you could direct me to any good literature about this subject(i already have the dragonbook).

Answer 1

You should ask your tutor, not the internet in general!

Whatever your thesis, you need:

1. a good tutor
2. to be best buddies with your tutor
   • and one way to do that is sucking up and asking them their opinion on such a critical decision
3. it must be fun - that's an important motivator - but it must also be something you're going to excel at

Its not fun to mess it up and grade poorly. Its not cheap - or perhaps not even possible - to do it all again.

You have a few months before the thesis starts. I'd say that you can lightly do a global reg allocator if you do a practice run now.

But whatever your thesis subject, you need to be committed.

Answer 2

I have done some register allocation for a much easier language than C (for an undergraduate compilers course). It was pretty hard for me, but I didn't have much time. I would say your project is feasible for a master thesis.

An excellent book to read for compiler design and especially optimizations (including register allocation) is Advanced Compiler Design & Implementation.