mmm, compiler options, VM options, compiler versions, register usage,  
gnu-header differences etc all come to mind.
Likely need to beat the compiler into producing the same instructions  
as the carbon vm has for the same interp.c

On 26-May-06, at 3:12 PM, Avi Bryant wrote:

>
> On May 26, 2006, at 1:35 PM, John M McIntosh wrote:
>
>> Todd, did we ever solve the performance issues with this VM as  
>> compared to the carbon mac intel VM? 3.8.12b4U
>
> 0 tinyBenchmarks  '224956063 bytecodes/sec; 7593723 sends/sec'
>
> vs. the Carbon VM:
>
>  '348536419 bytecodes/sec; 7042112 sends/sec'
>
> So the sends are comparable but it's running at 2/3 the bytecode  
> speed.
>
> Avi
>

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