Re: Type Alignment

[Jeff Squyres <jsquyres at open-mpi dot org>]

On Feb 21, 2006, at 6:13 PM, Tobias Schlüter wrote:

> > 1. What is a reliable / correct test to determine the alignment of a
> > Fortran type?  (I don't care a whit about the previous tests -- if
> > they're wrong, we'll throw them out)
>
> Portably, there isn't.  Fortran doesn't make any guarantees.  I  
> believe for fortran they will always be the same as for gcc, but a  
> middleend or backend person would have to verify this.

Yikes.  Let me re-phrase my question...

Fortran standard aside, are there tests that will work with most/all  
Fortran compilers?  I.e., even though the Fortran standard doesn't  
make any guarantees, do compilers obey the same/similar conventions  
such that one test (or a small set of tests) would result in the  
correct answer?

> > 3. Am I reading these replies correctly that the *alignment* of a
> > type may change depending on whether a derived type is sequenced or
> > not?  I can understand how members might be re-ordered in memory by
> > the compiler (and that's not a problem for us), but can it really
> > change the datatype's alignment?  If so, this could be a Big Problem
> > for us.
>
> In principle yes.  Again, the Fortran standard doesn't make any  
> guarantees about this.

Similar to above: the standard notwithstanding, are there actually  
compilers out there that change the alignment of a type based on  
whether the derived type is sequenced or not?  Any idea what an MPI  
implementation should do in such cases?

--
{+} Jeff Squyres
{+} The Open MPI Project
{+} http://www.open-mpi.org/