Nobody would deny that DSM is an easier programming model
to use for a large class of applications. Most of the overhead/
loss of performance in a DSM system has to do with the cost
of the underlying communication mechanism; the fact that DSM
systems can almost never come to par with an efficiently written
message-passing application in terms of the number/size of messages
generated, hurts the performance almost exponentially corresponding
to the cost of the underlying messaging - making the DSM system look
extremely poor.
Having said that, I'd venture that DSM technology can be
considered to be waiting for the NUMA communication technology
(interconnect + the protocols) to become commodity.
:a
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Mar 07 2000 - 21:00:20 EST