http://sean.chittenden.org/ Sean's Digital Dumping Ground For Public Information en Serendipity - http://www.s9y.org/ Mon, 21 Jan 2008 23:43:00 GMT http://sean.chittenden.org/templates/default/img/s9y_banner_small.png http://sean.chittenden.org/ 100 21 http://sean.chittenden.org/archives/2008/01/21/268/sumSTM-mutex-CAS-unstupid-things.html C/C++ http://sean.chittenden.org/archives/2008/01/21/268/sumSTM-mutex-CAS-unstupid-things.html#comments http://sean.chittenden.org/wfwcomment.php?cid=268 0 http://sean.chittenden.org/rss.php?version=2.0&type=comments&cid=268 nospam@example.com (Sean Chittenden) <p>Robert Ennals's <a href="http://sean.chittenden.org/exit.php?url_id=46&amp;entry_id=268" title="http://berkeley.intel-research.net/rennals/pubs/052RobEnnals.pdf" onmouseover="window.status='http://berkeley.intel-research.net/rennals/pubs/052RobEnnals.pdf';return true;" onmouseout="window.status='';return true;">Software Transactional Memory Should Not Be Obstruction-Free</a> is definitely unstupid and worth reading. Same with Dave Dice, Ori Shalev, and Nir Shavit's paper on <a href="http://sean.chittenden.org/exit.php?url_id=47&amp;entry_id=268" title="http://research.sun.com/scalable/pubs/DISC2006.pdf" onmouseover="window.status='http://research.sun.com/scalable/pubs/DISC2006.pdf';return true;" onmouseout="window.status='';return true;">Transactional Locking II</a>. Survey says? Holy damn. Use &quot;per object transaction locking&quot; models where possible and avoid the use of pthread_mutex_lock*, pthread_rwlock_rdlock*, and pthread_rwlock_wrlock* where possible. <a href="http://sean.chittenden.org/exit.php?url_id=48&amp;entry_id=268" title="http://en.wikipedia.org/wiki/Software_transactional_memory" onmouseover="window.status='http://en.wikipedia.org/wiki/Software_transactional_memory';return true;" onmouseout="window.status='';return true;">STM</a> makes pthreads* the <a href="http://sean.chittenden.org/exit.php?url_id=49&amp;entry_id=268" title="http://www.youfail.org/" onmouseover="window.status='http://www.youfail.org/';return true;" onmouseout="window.status='';return true;">new fail</a>. For some idiotic reason, I was thinking/hoping that <a href="http://sean.chittenden.org/exit.php?url_id=50&amp;entry_id=268" title="http://fxr.watson.org/fxr/source/sys/rwlock.h#L58" onmouseover="window.status='http://fxr.watson.org/fxr/source/sys/rwlock.h#L58';return true;" onmouseout="window.status='';return true;">rwlock</a>'s would suck less than they apparently do. <a href="http://sean.chittenden.org/exit.php?url_id=51&amp;entry_id=268" title="http://en.wikipedia.org/wiki/Compare-and-swap" onmouseover="window.status='http://en.wikipedia.org/wiki/Compare-and-swap';return true;" onmouseout="window.status='';return true;">CAS</a>? Eh, not quite worth the hassle yet.</p> Mon, 21 Jan 2008 15:43:00 -0800 http://sean.chittenden.org/archives/2008/01/21/268/guid.html cas concurrency mutex programming pthreads stm