[MLton] Multicore CPU's and MLton
Matthew Fluet
fluet@cs.cornell.edu
Sun, 3 Jul 2005 14:49:40 -0400 (EDT)
> > hmm, how much work do you think that would take?
>
> I think the C codegn hack would be easy (hours). The native codegen
> one I'm not so sure, but probably a few days. Hopefully Matthew will
> comment.
The native codegen just sends each MachineIL through the codegen passes
one at a time, so it ought to be parallelizable. Of course, once you have
multiple threads in flight, they are competing for the same memory (and,
more importantly with multi-core, the same cache), so I'm not sure how the
trade off would be resolved.