On Sat, Sep 18, 2010 at 4:36 PM, Alexander J. Vondrak
<[email protected]> wrote:
> On the other hand, Slava discussed writing a GVN pass, so I was looking over 
> some GVN stuff -- mostly skimming the VanDrunen GVN-PRE thesis.  I found that 
> really interesting; I think implementing some sort of GVN would make for a 
> good project (thesis or otherwise, though I think my advisor will like it).  
> If you're okay with me working on it:

Great!

> - We'd need the GC map refactoring first.  Would it be part of the project, 
> or is it more of a "todo"?  Maybe I should do it to get more familiar with 
> the code?

The changes I was describing to compiler.cfg construction are mostly
independent of a GVN pass. They will make GVN more effective though,
by exposing more values.

> - On that note, is there a recommended reading list of compiler.cfg vocabs to 
> get up to speed?  I mean, I'd read all I could, but maybe there's handful of 
> important ones to make for a shorter adjustment period.

I'd say the four most important passes are:

compiler.cfg.value-numbering
compiler.cfg.representations
compiler.cfg.ssa.destruction
compiler.cfg.linear-scan

Slava

------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances
and start using them to simplify application deployment and
accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Factor-talk mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/factor-talk

Reply via email to