(j3.2006) co-array variable
Thu May 31 13:01:24 EDT 2007
On Thursday 31 May 2007 09:29, Craig Rasmussen wrote:
> Could you please tell me what you mean by atomic in terms of the ?
It is not in the standard--I used it in order to clarify the reasoning behind
treating volatile scalars of a single numeric unit differently from others.
We allow such a scalar to be referenced "concurrently" with other images
defining it (this is not allowed for any other variable). Implementing this
without complications requires an atomic "load" operation to load the value
from memory. I cannot think a processor that would not have such a load for a
single-word scalar. This is the only "atomic" operation we support in CAF as
far as I can see.
Aleksandar Donev, Ph.D.
Lawrence Postdoctoral Fellow @ LLNL
High Performance Computational Materials Science and Chemistry
E-mail: donev1 at llnl.gov
Phone: (925) 424-6816 Fax: (925) 423-0785
Address: P.O.Box 808, L-367, Livermore, CA 94551-9900
More information about the J3