Seems Apple has some optimization extra features they've added to GCC  
and fiddling they've done to turn them on.

So if you specify
-fast -mcpu=G4 -mtune=G4

You get a VM that has about a 9.50% better send/sec rate and 2.5%  
better bytecode rate on a G4 machine. For a G3 the bytecodes stay about  
the same, but send/sec improve by 2.5%

Originally this is exposed as a G5 tuning parm, but careful reading of  
the release notes say it's usable for G4s too, and not all features it  
turns on are available for generic GCC, althought they've been  
submitted for 3.4.

I'll toss out a 3.7.0b3 in a day or so.

--
======================================================================== 
===
John M. McIntosh <johnmci at smalltalkconsulting.com> 1-800-477-2659
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
======================================================================== 
===