(j3.2006) (SC22WG5.3905) [MPI3 Fortran] [ukfortran] [MPI3 Fortran] MPI non-blocking transfers

Jim Xia jimxia
Thu Jan 22 17:19:13 EST 2009


From:
Van Snyder <Van.Snyder at jpl.nasa.gov>
To:
"longb at cray.com" <longb at cray.com>, fortran standards email list for J3 
<j3 at j3-fortran.org>
Cc:
WG5 <sc22wg5 at open-std.org>, "mpi3-fortran at lists.mpi-forum.org" 
<mpi3-fortran at lists.mpi-forum.org>
Date:
01/22/2009 04:45 PM
Subject:
(j3.2006) (SC22WG5.3903) [MPI3 Fortran] [ukfortran] [MPI3       Fortran] 
MPI non-blocking transfers




On Thu, 2009-01-22 at 10:59 -0800, Bill Long wrote:
> When we're talking about a Fortran's asynchronous
> attribute, it seems reasonable to relate it to Fortran's concept of
> I/O.

When a Fortran program invokes an external procedure with an actual
argument that has the ASYNCHRONOUS attribute, and all you have for the
external procedure is a .o file, how does the compiler know whether the
procedure does Fortran I/O, or something else?


Sad to say, but true to some compiler writers: asynchronous alone will 
have no impact on how the procedure call is dealt with (or optimized). 
Asynchronous attribute will only take effect when there is a corresponding 
WAIT statement somewhere due to the "invisible hands" referred to by Bill. 
 I think asynchronous approach to nonblocking transfer may lead to 
nowhere, at least not as easy as it appears.

Cheers


Jim Xia

RL Fortran Compiler Test
IBM Toronto Lab at 8200 Warden Ave, Markham, On, L6G 1C7
Phone (905) 413-3444  Tie-line 313-3444
email: jimxia at ca.ibm.com
D2/YF7/8200 /MKM
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://j3-fortran.org/pipermail/j3/attachments/20090122/4b436a62/attachment.html 



More information about the J3 mailing list