>John,
>
>I ran a quick check on the changes and the results were roughly what I
>expected:
>	CCodeGenEnh:  +10% bps,  +5% sps
>	GlobalStruct:  -4% bps, -12% sps
>[bps - byte codes per second, sps - sends per second]
>
>BTW, several years ago I had a similar CS to that of Anthony. It was
>rejected back then because CW9 (or what was used then) generated
>slightly slower code although the life-time analysis of the compiler
>should improve performance. Glad to see that MacOS finally caught on ;-)
>
>Cheers,
>   - Andreas

Ok, are you saying the CCodeGenEnh makes bytecodes run 10% faster? 
If so is that a worthwhile addition to the regular update stream?

Yes under os-x we now use gcc so it does do better local 
optimizations, versus CW which just gives up when faced with lots of 
locals.

If GlobalStruct makes bytecodes 4% slower then it just shows that the 
issue is powerpc specific, versus platform independent. For ppc it's 
faster.

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