Getting locks and sharing: was RE: Getting concurrency
Vlad Dumitrescu XX (LN/EAB)
Fri Jun 17 10:05:51 CEST 2005
The point about RPC times is very important, yes.
I would be very surprised if the times for RPC4 will be different than for RPC2 - at least when the multi-core CPU design matures. But you are right not to rely on that.
One could also consider using a different distribution bearer for such multi-core clusters of nodes, maybe using shared memory instead of TCP/IP. Hopefully that could provide better times.
>In any case the layers must be *very* clear and the consequences of
> using a library must be clear to the programmer.
True, I didn't think about that. Was looking from the other end, about not having to change old code to be able to take advantage of the multiple CPUs.
>Then we should add a few new bifs that are "multi-CPU aware".
>Then we should write some OTP behaviours to abstract the multi-CPU
Well, that's a supporting framework, if anything :-)
More information about the erlang-questions