Re: MOSIX

[Dariush Pietrzak <eyck@forumakad.pl>]

> There is a nice (non-free) SGI parallelizing FORTRAN compiler which unrolls loops and
I only have 3SGI's here at the moment and they are becoming outdated very
quickly. I won't get any new SGI hardware :(

> distributes them across processors, but that's for IRIX and only works well on shared
hmm, loops are the easiest part. I can do it myself quite easily.

> memory machines- like their Origin 2000s.  That's about as easy as it gets, but you
> need a large SMP box to make it worth while.
only one two processor SGI here. not to fast. sparcs are faster. but due
to hardware aging new students-labs based on PCs will become faster that
any hardware I have here.

> separate and may be unrolled, C assumes they happen in series.  So for example, the C
there are very funny compilers that allow one to mark parts of code that
may be parallellized..(pardon my english)


> would generate the Fibonacci series, but if you did it in FORTRAN, a good optimizing
> compiler would not necessarily do the loop sequentially and would probably give you
> garbage.
I've seen parallel fibonacci algorithm :) not that I understand it fully
;) 



> I don't know of anyone using it to do HPC.  The trouble is, even with MOSIX, you have
> to write your app in multiple threads/processes, and MOSIX will distribute the
> threads/processes over machines, the way Linux SMP distributes them over processors on
> one machine.  Still non-trivial.
But that's easier to do. and programmers do it all the time due to popular
dual-processor bords. 

> delays.  I don't know how MOSIX does this, but it's bound to be slower than if you use
> MPI to control communication yourself.
Yes, I assume some there will be problems, I'm just wondering how big they
are.


regards, Eyck