In the context of Common Lisp, what is Symmetric-multiprocessing?
An SMP Lisp is one in which performance can scale with processor count as well as with processor speed. It makes “just add more cpus” an option for dealing with increased load. It’s pretty well known that a problem has to factor in an unusual way for this to work, whatever language and library is used. Since support functions (file access, paging, network communications, etc) are already using extra processors, every application that uses these functions is getting some benefit from any extra CPU’s on the system. The iconic application for SMP is transaction processing. Network speed and current cycle rates (factored by cpu count) can feed a huge number of transactions to an application. Allowing parallel execution of several transactions could boost transaction throughput, as long as a serializing resource (data base access for example) doesn’t become the real bottleneck.
