(j3.2006) [MPI3 Fortran] Feedback from Fortran J3 meeting

[Bill Long]

Keith Bierman wrote:
> On May 28, 2008, at 10:56 AM   5/28/, Craig Rasmussen wrote:
>
>   
>> I wonder how C does this?  What are the rules for code motion in C?
>> I'll ask around.
>>
>>     
> Highly prescribed by their Standard. Fortran's advantage has been  
> much more latitude to the optimizer.
>
> While the Committee has always preferred to design more generic  
> solutions, MPI is an important special case ... it might just be a  
> lot easier to have an optional part of the Standard which puts MPI in  
> as special modules ...

I see this as very backward looking.  We've done a lot of work to 
produce a native parallel programming model within Fortran that lacks 
these problems that MPI has.  Producing a whole optional part of the 
standard just to accommodate a legacy library that one would expect, 
over time, to fall by the wayside, seems like a bad idea on multiple fronts.


>  then it's up to the implementors to "do it  
> right" 

If vendors are currently getting it "wrong", how is it possible that so 
many people are using MPI today?  I'm yet to be convinced there is a 
real problem.

Cheers,
Bill




> rather than trying to describe all the right behaviors in a  
> processor and OS agnostic manner consistent with the "abstract  
> Fortran machine" (I know, we don't have those words, but we have the  
> concept ;>).
>
>
>   

-- 
Bill Long                                   longb at cray.com
Fortran Technical Support    &              voice: 651-605-9024
Bioinformatics Software Development         fax:   651-605-9142
Cray Inc., 1340 Mendota Heights Rd., Mendota Heights, MN, 55120