Rick DeNatale wrote:
>  * Previous experience with compiling dynamic OO languages has shown
> that the much smaller code representation of byte codes compared to
> machine code can actually lead to better performance on machines with
> virtual memory (almost all machines these days) due to the smaller
> working set.

Absolutely correct... but there's even more to this
than meets the eye. The working set that matters most
is the cache, not the RAM. When you have an instruction
cycle time that's more than 50x the RAM cycle time, you
can do a lot of work on something that's already in the
cache while you're waiting for the next cache line to
fill.

Reducing the working set on boxes with GB of RAM typically
has more effect through decreasing cache spills than via
reductions in page faults. The byte-codes also go in
d-cache while the interpreter itself is in I-cache.

Clifford Heath.