Re: compiler optimization project

Karsten Nyblad <148f3wg02@sneakemail.com>
26 Oct 2006 00:31:12 -0400

          From comp.compilers

Related articles
compiler optimization project khubaib92@gmail.com (khubaib) (2006-10-19)
Re: compiler optimization project egor.pasko@gmail.com (2006-10-24)
Re: compiler optimization project 148f3wg02@sneakemail.com (Karsten Nyblad) (2006-10-26)
Re: compiler optimization project ian.rogers@manchester.ac.uk (Ian Rogers) (2006-10-28)
Re: compiler optimization project mwso@earthlink.net (Gary Oblock) (2006-10-28)
| List of all articles for this month |

From: Karsten Nyblad <148f3wg02@sneakemail.com>
Newsgroups: comp.compilers
Date: 26 Oct 2006 00:31:12 -0400
Organization: Compilers Central
References: 06-10-072
Keywords: optimize
Posted-Date: 26 Oct 2006 00:31:12 EDT

khubaib wrote:
> Hello,
>
> I am a graduate student doing major in computer architecture, and I am
> taking a course on optimizing compilers. There is a term project
> involved in this course, and I was looking for ideas. The
> infrastructure that we have used so far in the course is JIKES RVM,
> and I hope to continue using it for the project. I am interested in
> register allocation and scheduling (mostly the backend, near to
> machine). Could any one please suggest an idea? As I am from
> architecture background, I was thinking of something like scheduling
> issues for upcoming multiple-core based machines. In a multiple core
> machine, there might be some simple and some very aggressive
> processors, and studying what optimization should be done for which
> core can be beneficial. (thinking on those lines...).


Ada has some rules on when a task (Ada calls threads tasks) may have a
local copy of a global variable, e.g., in


        A:= 1; B:= 2;


other tasks may see B change to 2 before they see A change to 1. Of
course all global variables must be updated at points of
synchronization. You may want to study how that influence scheduling of
registers holding copies of global variables.


Karsten Nyblad
148f3wg02 at sneakemail dot com



Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.