question about concurrency implementation

Ketil Z. Malde [email protected]
19 Mar 2002 08:56:40 +0100


Dean Herington <[email protected]> writes:

> From Simon Marlow's reply, I gather that the current implementations of
> Concurrent Haskell provide "concurrency" but not "parallelism", and that
> provision of parallelism is not likely in the near term.

Uh, GPH exists, doesn't it?  So you can get "parallelism" if you're
after performance.  The only implementation appears to be based on
PVM, so I expect some overhead on SMPs, although I haven't gotten
around to actually try it yet.

I don't know if or how you can mix the two paradigms.

There's also an implementation using threads for asynchronous calls to
Win32 libraries, if I understand things correctly.  A full
thread-based parallel implementation might (or might not) be a logical
next step.

-kzm
-- 
If I haven't seen further, it is by standing in the footprints of giants