Re: Suggestion for next EGCS (cluster,fe?)

[Edward Jason Riedy <ejr at cs dot berkeley dot edu>]

Oh well.  And "Christopher W. Curtis" writes:
 - 
 - Why not build intrinsic support for these libraries directly into the
 - compiler?

You might be interested in Cilk.  Earlier versions worked on distributed-
memory systems.  See http://supertech.lcs.mit.edu/cilk/.  Also poke around
for Sisal. although I think it's mostly dead.

 - I would think that #pragma directives could be used to indicate shared 
 - memory, code granularity, etc. 

See http://www.openmp.org/, at least for shared-memory systems.  There's 
now an official C/C++ specification.  You'll note that it gets a bit
hairy...

 - Now, I haven't done any code like this and really don't know any issues
 - surrounding it, but does this seem unreasonable?

Yup.  Not all algorithms can be parallelized for distributed memory 
systems through simple means.  Most need a fundamental re-thinking to 
get cost-justifying performance increases with high-latency 
communications.

Jason