Sarita's Weak Ordering paper is, I believe, the first mention of
sequential consistency for data race free programs. Others had
different criteria for ensuring sequential consistency:
@inproceedings{325100,
author = {Sarita V. Adve and Mark D. Hill},
title = {Weak ordering\—a new definition},
booktitle = {Proceedings of the 17th annual international symposium on
Computer Architecture},
year = {1990},
isbn = {0-89791-366-3},
pages = {2--14},
location = {Seattle, Washington, United States},
}
There are several papers that state the new JMM will have this property.
The first published one that comes to mind is Bill's and my paper in
CSJP last month:
Requirements for a Programming Language Memory Model. Jeremy Manson and
William Pugh. Workshop on Concurrency and Synchronization in Java
Programs, in association with PODC. July, 2004.
You can get that paper here:
http://www.cs.umd.edu/users/jmanson/java/testcases-csjp.pdf
The JSR itself also states this, as will my forthcoming dissertation.
Jeremy
Bart Jacobs wrote:
> I am writing a paper about a locking scheme, and in that paper I
> assume that
> the memory model obeys the rule that if all sequentially consistent
> executions are data-race-free, then all executions are sequentially
> consistent. However, I anticipate that many readers are not familiar
> with
> this material, and I wonder if anyone knows a good published reference
> to
> which I can refer? I would be particularly interested in anything that
> states that this rule will be a guarantee in the next Java Memory
> Model.
-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:08 EDT