Newsgroups: comp.parallel.mpi
From: marco luchini <luchini@ic.ac.uk>
Subject: Extended precision using MPICH
Organization: i. c.
Date: Mon, 10 Aug 1998 18:45:10 +0100
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-ID: <35CF31A6.6D8BB445@ic.ac.uk>

Hi,

I need to pass REAL*16 data around using MPICH on a DEC alpha.  The
compiler supports extended precision so the single node code works OK. 
But there doesn't seem to be an intrinsic type corresponding to it.  I
can't see how I can use derived datatypes to achieve what I need.  Or am
I missing something in the contiguous types? 

For straightforward passing I guess I can cheat and use an EQUIVALENCE
statement to a REAL*8 array twice the size, but collective operations
won't work without a lot more fiddling.  

Just in case I'm missing something that should be obvious.


Thanks,							Marco