Michael Latta writes:
 > Now if we take the bytecode speed up from the recent JIT project and 
 > the new message semantics, we are getting there.  Making ObjectMemory 
 > concurrent will be the biggest part of the effort.

I'm interested in exploring instruction level parallelism in Squeak.
The reason is the machinery needed to fully optimise at: and at:put:
bytecodes (to C speeds) is very similar to that needed to vectorise.
But first I need to fully optimise at: ...

An ObjectMemory that supported multiple interpreter threads would
be nice. My guess is that would require a separate eden for each
thread. Requiring synchronisation for each object allocation would
be expensive.

Oh, I've had problems with code that was limited by write bandwidth.
On my CPU I'm only allowed one write every 7 clock cycles. Extra cores
on the same die wouldn't help with that code, even with perfect
parallelisation.

Bryce