http:^^www.cs.rochester.edu^u^scott^synchronization.html
来自「This data set contains WWW-pages collect」· HTML 代码 · 共 159 行
HTML
159 行
Date: Wednesday, 20-Nov-96 20:04:46 GMTServer: NCSA/1.3MIME-version: 1.0Content-type: text/htmlLast-modified: Friday, 11-Oct-96 15:30:22 GMTContent-length: 6317<HTML><HEAD><TITLE>High-Performance Synchronization</TITLE></HEAD><BODY BACKGROUND="patterns/yellow_paper.gif"><!WA0><IMG ALIGN=MIDDLE SRC="http://images/urcslogo.gif"><B>NSF grant CCR-9319445</B><H1>High-Performance Synchronization for Shared-Memory Parallel Programs</H1><BLOCKQUOTE><!WA1><A HREF="http://www.cs.rochester.edu/urcs.html">Computer Science Department</A><BR>University of Rochester<BR>Rochester, NY 14627-0226<BR>4-1-94 through 9-30-97<BR></BLOCKQUOTE><P>With increases in the size and availability of parallel processors withshared-memory programming models, high-performance synchronization isbecoming increasingly important.Several groups, including ours, have demonstrated in recent years thatsoftware synchronization algorithms can scale well to very largenumbers of processors, and that they can avoid certain negativeinteractions with high-performance scheduling algorithms.We are continuing this research in several directions, including(1) mechanisms for cooperative synchronization and scheduling,which minimize unnecessary spinning, maximize processor locality, andavoid contention for both lock and non-lock data;(2) comparative evaluation of alternative mechanisms for atomic update ofshared data structures, including locks, non-blocking synchronization,and function shipping;(3) implementation of atomic hardware primitives on scalablearchitectures;(4) evaluation of the interaction of synchronization with coherence; and(5) new synchronization algorithms.<H2>Principal Investigator</H2><BLOCKQUOTE><!WA2><A HREF="http://www.cs.rochester.edu/u/scott/home.html">Michael L. Scott</A><BR>Associate Professor and Department Chair<BR><tt>scott@cs.rochester.edu</tt><BR>716-275-7745</BLOCKQUOTE><H2>Recent Graduates</H2><UL><LI><!WA3><A HREF="http://www.cs.rochester.edu/users/grads/kthanasi/">Leonidas Kontothanassis</A><LI><!WA4><A HREF="http://www.cs.rochester.edu/users/grads/bob/">Bob Wisniewski</A></UL><H2>Graduate Students</H2><UL><LI><!WA5><A HREF="http://www.cs.rochester.edu/u/michael/home.html">Maged Michael</A><LI><!WA6><A HREF="http://www.cs.rochester.edu/u/gchunt/home.html">Galen Hunt</A><LI><!WA7><A HREF="http://www.cs.rochester.edu/u/srini/home.html">Srinivasan Parthasarathy</A></UL><H2>Publications</H2><UL><LI><!WA8><A HREF="http://www.cs.rochester.edu/u/scott/synchronization/pubs.html">Project-specific papers</A><LI><!WA9><A HREF="http://www.cs.rochester.edu/trs/systems-trs.html">Systems Technical Report Archive</A></UL><H2>Pseudocode</H2><UL><LI><!WA10><A HREF="http://www.cs.rochester.edu/u/scott/synchronization/pseudocode/ss.html">Scalable spinlocks and barriers.</A>Includes test-and-set and ticket locks; queue locks; andcentralized, tree-based, and fft-style (``butterfly'') barriers.From the 1991 <i>TOCS</i> paper.<LI><!WA11><A HREF="http://www.cs.rochester.edu/u/scott/synchronization/pseudocode/rw.html">Scalable busy-wait reader-writer locks.</A>Includes reader-preference, writer-preference, and fair locks.From the 1991 <i>PPoPP</i> paper.<LI><!WA12><A HREF="http://www.cs.rochester.edu/u/scott/synchronization/pseudocode/adaptive.html">Scalable adaptive combining tree barriers.</A>Combine local-only spinning, logarithmic critical paths, amortization ofoverhead for skewed arrival, and ``fuzziness''.From the 1994 <i>IJPP</i> paper.<LI>Variations on Lamport's fast mutual exclusion lock. Use no atomicinstructions other than read and write.From UR TR 460 (submitted for publication).<LI><!WA13><A HREF="http://www.cs.rochester.edu/u/scott/synchronization/pseudocode/ps_and_sc.html">Preemption-safe and scheduler-conscious synchronization algorithms.</A>Includes two queue-based mutual exclusion locks; test-and-set and ticketlocks; a fair, scalable, queue-based reader-writer lock; competitive andoptimal-time small-scale barriers; and a scalable barrier.All algorithms avoid busy-waiting for action by preempted processes,including those waiting in line for a FIFO queue or ticket lock. Mostemploy a widened kernel-user interface.Revised from UR TR 550; to appear in <i>ACM TOCS</i>.<LI><!WA14><A HREF="http://www.cs.rochester.edu/u/scott/synchronization/pseudocode/heap.html">A highly-concurrent multi-lock concurrent priority queue.</a>Uses bottom-up insertions and ``bit-reversal'' choice among fringe nodes.<LI><!WA15><A HREF="http://www.cs.rochester.edu/u/scott/synchronization/pseudocode/queues.html">Fast concurrent queue algorithms.</A>We believe these algorithms to be the best concurrent queues available,for almost any application.</UL><H2>Executable Code</H2><UL><LI>Basic and scalable spinlocks and barriers.Code to run on the<!WA16><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/locks_and_barriers/Symmetry.tar.Z">Sequent Symmetry</A>,<!WA17><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/locks_and_barriers/Bfly1.tar.Z">BBN Butterfly 1</A>, and<!WA18><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/locks_and_barriers/TC2000.tar.Z">BBN TC2000</A>.<LI>Scalable busy-wait reader-writer locks.Code to run on the<!WA19><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/read_write.tar.Z">BBN TC2000</A>.<LI>Scalable adaptive combining tree barriers.Code to run on the<!WA20><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/adaptive/Bfly1.tar.Z">BBN Butterfly 1</A>,<!WA21><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/adaptive/TC2000.tar.Z">BBN TC2000</A>, and<!WA22><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/adaptive/KSR1.tar.Z">Kendall Square KSR 1</A>,<LI>Variations on Lamport's fast mutual exclusion lock. Code to run on the<!WA23><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/scalable_synch/fast/locks_sgi.c">Silicon Graphics Iris</A>.<LI>Preemption-safe and scheduler-conscious synchronization algorithms.Code to run on the<!WA24><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/sched_conscious_synch/multi.KSR.tar.Z">Kendall Square KSR 1</A>and<!WA25><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/sched_conscious_synch/multi.SGI.tar.Z">Silicon Graphics Challenge</A>.<LI><!WA26><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/concurrent_heap">A highly-concurrent multi-lock concurrent priority queue.</a>Code to run on the SGI Challenge.<LI><!WA27><A HREF="ftp://ftp.cs.rochester.edu/pub/packages/sched_conscious_synch/concurrent_queues/concurrent_queues.tar.gz"> Fast concurrent queue algorithms</A>.Includes SGI Challenge code for our two-lock and non-blocking queues, andfor previous algorithms by other researchers.</UL><HR><ADDRESS>Last Change: 23 August 1996 /<!WA28><A HREF="mailto:scott@cs.rochester.edu">scott@cs.rochester.edu</A></ADDRESS></BODY></HTML>⌨️ 快捷键说明
复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?