|What ideas are better for assigning registers to terminals? email@example.com (Bill A.) (1999-09-16)|
|Re: What ideas are better for assigning registers to terminals? firstname.lastname@example.org (Peter Bergner) (1999-10-06)|
|Re: What ideas are better for assigning registers to terminals? email@example.com (1999-10-06)|
|Re: What ideas are better for assigning registers to terminals? firstname.lastname@example.org (Ben Franchuk) (1999-10-11)|
|Re: What ideas are better for assigning registers to terminals? email@example.com (Max Hailperin) (1999-10-11)|
|From:||Max Hailperin <firstname.lastname@example.org>|
|Date:||11 Oct 1999 02:37:39 -0400|
|Organization:||Gustavus Adolphus College|
|References:||99-09-060 99-10-032 99-10-037|
email@example.com (Andreas Krall) writes:
> Peter Bergner <firstname.lastname@example.org> writes:
> .... Read about conservative coalescing in Preston [Briggs]'s
> thesis ...
> Conservative coalescing gives worse results than Chaitins original
> agressive coalescing. ...
Just to be absolutely clear: Briggs did not suggest using conservative
coalescing in the normal case, only in the very limited context of
un-splitting live-range splits. His allocator used agressive
coalescing normally. Krall may well have known this, and Moon and
Park (whom he cites) certainly did, and say as much. But I have run
into a number of people who do not know this, and the comments quoted
above might well reinforce their mis-understanding.
So far as I know, it is Appel and George who first applied
conservative coalescing as the normal coalescing scheme. Appel's book
has been spreading this idea rather widely. But it is not Briggs's.
Associate Professor of Computer Science
Gustavus Adolphus College
800 W. College Ave.
St. Peter, MN 56082
Return to the
Search the comp.compilers archives again.