There is another problem with automatic parallelisation and Erlang. As Erlang DOES have side-effects, message passing basically, the order in which things are evaluated is significant. It is therefore not safe to automatically parallelise code. It is up to the programmer/designer to decide how the application should be made parallel.
<br><br>This is, of course, not always easy. Especially for most who come from a sequential world where you usually work out how to make things sequential. Yes I have been there. :-)<br><br>Robert<br><br><div><span class="gmail_quote">
On 17/09/2007, <b class="gmail_sendername">Jouni Rynö</b> <<a href="mailto:Jouni.Ryno@fmi.fi">Jouni.Ryno@fmi.fi</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
On Mon, 2007-09-17 at 12:33 +0200, Kenneth Lundin wrote:<br><br>> If the system design already allows thousands of parallell tasks it is<br>> far from obvious that also the internal execution of each such task<br>> should benefit from
<br>> automatic parallellisation. I actually believe that a system can be<br>> over parallellized and that this will degrade the performance.<br>><br>There used to be a beautiful processor architecture called Transputer
<br>and the language called Occam to program it. Making things going<br>parallel was even more easier and natural than in Erlang.<br><br>And there were many papers published, that going from extreme parallel<br>systems to more serialised one was the way optimise the performance.
<br>Extreme parallel meaning systems, where you have decades more processes<br>than real executing hardware. It all depends on the process switching<br>time and communication overhead.<br><br>regards<br> Jouni<br>--
<br><br> Jouni Rynö <a href="http://mailto://Jouni.Ryno@fmi.fi/">mailto://Jouni.Ryno@fmi.fi/</a><br> <a href="http://space.fmi.fi/~ryno/">http://space.fmi.fi/~ryno/
</a><br> Finnish Meteorological Institute <a href="http://www.fmi.fi/">http://www.fmi.fi/</a><br> Space Research <a href="http://space.fmi.fi/">http://space.fmi.fi/</a><br> P.O.BOX 503 Tel (+358)-9-19294656
<br> FIN-00101 Helsinki FAX (+358)-9-19294603<br> Finland priv-GSM (+358)-50-5302903<br><br> "It's just zeros and ones, it cannot be hard"<br><br>_______________________________________________
<br>erlang-questions mailing list<br><a href="mailto:firstname.lastname@example.org">email@example.com</a><br><a href="http://www.erlang.org/mailman/listinfo/erlang-questions">http://www.erlang.org/mailman/listinfo/erlang-questions