23 Dec 1997 23:01:27 -0500

Related articles |
---|

algorithm for computing the interference graph in SSA form? haahr@netcom.com (Paul Haahr) (1997-12-17) |

Re: algorithm for computing the interference graph in SSA form? preston@cs.rice.edu (1997-12-19) |

Re: algorithm for computing the interference graph in SSA form? cliff.click@Eng.Sun.COM (cliffc) (1997-12-23) |

Re: algorithm for computing the interference graph in SSA form? cliff.click@Eng.Sun.COM (cliffc) (1998-01-06) |

Re: algorithm for computing the interference graph in SSA form? mwolfe@pgroup.com (1998-01-08) |

Re: algorithm for computing the interference graph in SSA form? cliff.click@Eng.Sun.COM (cliffc) (1998-01-13) |

From: | cliffc <cliff.click@Eng.Sun.COM> |

Newsgroups: | comp.compilers |

Date: | 23 Dec 1997 23:01:27 -0500 |

Organization: | Sun Microsystems |

References: | 97-12-141 97-12-143 |

Keywords: | analysis |

Paul Haahr <haahr@netcom.com> wrote:

>I'm looking for an algorithm to quickly compute the interference graph

>(suitable for use in register allocation) for a program in SSA form.

Preston Briggs wrote:

*> If you want to build interference graphs, or register allocators*

*> in general, then you should be reading my thesis.*

*>*

*> author="Preston Briggs",*

*> title="Register Allocation via Graph Coloring",*

*> school="Rice University",*

*> year=1992,*

*> month=apr*

*>*

*> especially the chapter on engineering. At the time, I thought I*

*> pretty much beat the subject to death, though there have been a couple*

*> of small improvements since, along with several "improvements" that*

*> are best avoided.*

*>*

*> You can grab a copy via anonymous ftp from ftp.cs.rice.edu,*

*> in pub/preston/thesis.ps.gz*

I'm finishing up my 3rd Briggs-Chaitin allocator, and have been quite

pleased with the results. This last one is used inside a Java system

where compilation speed is paramount.

Cliff

--

Cliff Click Compiler Designer and Researcher

cliffc at acm.org JavaSoft

(408) 863-3266 MS UCUP02-302

--

Post a followup to this message

Return to the
comp.compilers page.

Search the
comp.compilers archives again.