Tree-ssa status with POOMA

[Daniel Berlin]

> With -ftree-points-to=andersen I get:
>
> 		890MB	5m14s	41s		1725584
>
> a nice 10% speedup (but that memory usage!).
>

The memory usage of PTA (which i assume is why it also takes a minute 
longer to compile) is something i'll take a look at after i'm done with 
my current work.
I'm sure it's something stupid if someone feels like tracking it down 
(We are surely keeping alias variables around longer than we need to 
be, etc).

The reference implementation I have using libbanshee that does whole 
program PTA can analyze an entire 2.3 million LOC program in roughly 3 
seconds, with 40 meg of memory.
So there is hope.
--Dan