Andreas Färber wrote: > Yes that's one for sure, and there were also some others, don't have > the URLs at hand right now, one other being from Prague I believe. The > point is those are mainly P/Invoke wrappers, requiring MPICH2 or > LAM/MPI to be installed on each node as I understand, which is more > suited for a cluster than for a grid. I'll give it a try though. > > For Java there are a lot more options, for example the Globus Toolkit > (http://globus.org/toolkit/), based on the OGSA (Web Services). > > The PMPI from São Paulo ("PMPI: A multi-platform, multi-programming > *language MPI using .NET") was the best equivalent I could find to > date but their paper does not give a source location and Google didn't > find any. Their implementation is based on .NET Remoting.* > > OGSI.NET (http://www.cs.virginia.edu/~gsw2c/ogsi.net.html > <http://www.cs.virginia.edu/%7Egsw2c/ogsi.net.html>) and WSRF.NET > (http://www.ws-rf.net/) on the other hand, being sponsored by > Microsoft, appear to rely on Microsoft's WSE, which is not available > for Mono. And don't forget JavaSpaces. JavaSpaces is an awesome distributed computing system that is applicable not just for HPC, but also for HA grid and cluster type solutions. I saw a demonstration of GigaSpaces, a commercial JavaSpaces implementation, and I was very impressed. The tools which GigaSpaces add are very awesome. Unfortunately, JavaSpaces is pretty java centric. There is an OpenSource project called Blitz (http://www.dancres.org/blitz/) which implements javaspaces. It may be possible to do some tricks with ikvm and/or jacil to run mono IL on blitz. That would be awesome, or a MonoSpaces project would be awesome, so would a Mono alternative to Jini. -- Jay _______________________________________________ Mono-list maillist - [email protected] http://lists.ximian.com/mailman/listinfo/mono-list
