Subject: Re: publishing
From: Craig Chambers (chambers@cs.washington.edu)
Date: Fri Sep 07 2001 - 21:56:12 PDT
Actually, these are improvements to Whirlwind, not Vortex. Oops.
-- Craig
chambers@cs.washington.edu wrote:
>
> I'm publishing some more small improvements to Vortex, in my quest to
> get the towers benchmark to compile and start to run w/ and w/o
> optimization before launching into getting dispatching implemented.
> Now it does!
>
> *) I added a pass to insert explicit assignments along incoming
> branches to a merge, to implement phi nodes. I also added the
> necessary code to generate C declarations for phi nodes.
>
> *) I fixed some glitches with TopDefNode needing, so that there are
> multiple occurrences now, each with its own possibly distinct
> representation.
>
> *) I fixed some problems with how ptr comparison operators were being
> codegened. Now the operations on pointers use different C macros that
> operations on numbers.
>
> *) I fixed some bugs with how operands to C prim stmt blocks were
> being generated.
>
> *) I improved the optimization reporting messages for phi-,
> unused-value-, and unreachable-code-elimination, along the lines I did
> for constant-prop.
>
> *) I extended the Wil runtime to include the conservative GC library.
>
> -- Craig
> _______________________________________________
> Cecil mailing list
> Cecil@cs.washington.edu
> http://majordomo.cs.washington.edu/mailman/listinfo/cecil
_______________________________________________
Cecil mailing list
Cecil@cs.washington.edu
http://majordomo.cs.washington.edu/mailman/listinfo/cecil
This archive was generated by hypermail 2b25 : Fri Sep 07 2001 - 21:53:03 PDT