[MPI3 Fortran] Teleconference Wednesday, Sept. 17: Summary

[Craig Rasmussen]

On Sep 17, 2008, at 12:44 PM, Bill Long wrote:

>
>
> Jeff Squyres wrote:
>> On Sep 17, 2008, at 12:23 PM, Supalov, Alexander wrote:
>>
>>> PS. I would not say that there's resistance to the long counts.  
>>> There's
>>> just a number of ways how they can be introduced, and some  
>>> interesting
>>> implications in the file I/O and other areas that may not meet the  
>>> eye
>>> right away.
>>>
>>> The matter of ILP64 support (i.e., ints as big as longs and  
>>> pointers)
>>> may be rather pertinent. Some guys have old Cray style apps that  
>>> they
>>> want to port. The MPI interface should behave nicely here, I'd  
>>> say, if
>>> only out of respect.
>>>
>>> Another interesting aspect is the autodouble options, as some call  
>>> it. I
>>> believe this makes all reals be real*8. Again, some people use this,
>>> mostly for the same reason, and Fortran MPI interface should  
>>> probably do
>>> something about this now, when we have the chance to do things  
>>> "right",
>>> for a change. :)
>>
>> Agree with all of the above.
>>
>> Suffice it to say, the issue is still open for the MPI-3 forum and  
>> is not expected to be closed/decided for a while (if for no other  
>> reason than MPI-2.2 discussions are likely to take precedence over  
>> MPI-3 discussions).  So it's good to keep this as an "open item"  
>> here in the Fortran WG, but make sure we track what's happening in  
>> the overall forum with regards to this issue.
>>
>
> For what it's worth, Fortran compilers already have to deal with the  
> interaction of modules and default=64 switches.  For example, the  
> intrinsic iso_fortran_env module contains named constants  
> (numeric_storage_size, for example) that have different values  
> depending on these compiler switches.  At least in our case, we have  
> separate modules that are stored in library subdirectories with  
> names like 'default64'. The compiler driver knows which place to  
> look depending on the command line options.  This works for "known"  
> modules that are supplied along with the compiler (as opposed to  
> user created modules).  Clearly it is possible to make the MPI  
> module(s) known in this sense.  (In fact, we do this for MPI already  
> on our systems.)
>

Bill,

Would this work for a third person library like OpenMPI?

If two interfaces are supplied, one for 4 byte integers and one for 8  
byte integers, what happens when the compiler switch to go to longs is  
chosen.  Does the compiler call the 8 byte kind version?  I believe if  
we go this route, we can't have a default integer interface because it  
would collide with the 8 byte integer version if the default64 switch  
is thrown.

Craig