Graph coloring when targeting IA32 processors?

david.boyle@ed.tadpole.com (Dave Boyle)
3 Jun 2003 01:13:39 -0400

          From comp.compilers

Related articles
Graph coloring when targeting IA32 processors? david.boyle@ed.tadpole.com (2003-06-03)
Re: Graph coloring when targeting IA32 processors? nicolas_capens@hotmail.com (2003-06-05)
Re: Graph coloring when targeting IA32 processors? vmakarov@redhat.com (Vladimir Makarov) (2003-06-05)
| List of all articles for this month |

From: david.boyle@ed.tadpole.com (Dave Boyle)
Newsgroups: comp.compilers
Date: 3 Jun 2003 01:13:39 -0400
Organization: http://groups.google.com/
Keywords: 386, optimize, question
Posted-Date: 03 Jun 2003 01:13:39 EDT

Hi Folks,


On page 123 of the third edition of 'Computer Architecture: A
Quantitative Approach', the authors state:


"Graph coloring works best when there are at least 16 (and preferably
more) general-purpose registers available for global allocation for
integer variables and additional registers for floating point.
Unfortunately, graph coloring does not work very well when the number
of registers is small because the heuristic algorithms for coloring
the graph are likely to fail."


Does this mean that graph coloring is rarely (or possibly never) used
by compilers targeting the IA32 family of processors, with their eight
integer GPRs?


Thanks in advance for your thoughts.


Dave
[It's true. It doesn't help that most of the x86 registers have magic
properties so they're not really interchangable with each other. -John]



Post a followup to this message

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