Re: Graph colouring and local register allocation

parthaspanda22@gmail.com
Sun, 04 Nov 2007 03:30:36 -0800

          From comp.compilers

Related articles
Graph colouring and local register allocation shafitvm@gmail.com (Mohamed Shafi) (2007-10-31)
Re: Graph colouring and local register allocation preston.briggs@gmail.com (preston.briggs@gmail.com) (2007-10-31)
Re: Graph colouring and local register allocation jle@ural.owlnet.rice.edu (2007-11-03)
Re: Graph colouring and local register allocation parthaspanda22@gmail.com (2007-11-04)
Re: Graph colouring and local register allocation SidTouati@inria.fr (Sid Touati) (2007-11-05)
Re: Graph colouring and local register allocation miles.bader@necel.com (Miles Bader) (2007-11-22)
Re: Graph colouring and local register allocation Sid.Touati@uvsq.fr (Sid Touati) (2007-11-30)
Re: Graph colouring and local register allocation gneuner2@comcast.net (George Neuner) (2007-12-01)
Re: Graph colouring and local register allocation sjdutoit@gmail.com (sdt) (2007-12-01)
Re: Graph colouring and local register allocation preston.briggs@gmail.com (preston.briggs@gmail.com) (2007-12-01)
[8 later articles]
| List of all articles for this month |
From: parthaspanda22@gmail.com
Newsgroups: comp.compilers
Date: Sun, 04 Nov 2007 03:30:36 -0800
Organization: Compilers Central
References: 07-10-10307-11-003
Keywords: registers
Posted-Date: 05 Nov 2007 00:29:13 EST

> Many other global allocators are also able to solve local allocation.
> After all, local allocation is an easier problem. I expect the reason
> some people use a separate local allocator is that they have a bright
> idea about how to do a better job on the simpler problem, so they try
> it out.


One bright spot is how you approach register allocation on a stack
architecture(e.g. x87). That architecture allows you to think only in
local allocation terms. A useful idea to exploit in local register
allocation is placement of spill code. If you build a trace(like the
one that is created by an instruction scheduler) then you can try
moving code across basic block boundaries(either up or down) to bring
live ranges "closer"(i.e. make them shorter). The more closer a live
range becomes, the less likely will it approach k.


Sincerely.



Post a followup to this message

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