S. Fu and N. Tzeng. A circular list-based mutual exclusion scheme for large shared-memory multiprocessors. In IEEE Transactions on Parallel and Distributed Systems, pages 628--639, 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....list. In [6] Anderson presents a spinlock based locking mechanism where processors spin on a remote flag which increases network traffic. Johnson and Harathi [7] present an algorithm where all requesting processes spin on a local flag in a queue while waiting for the lock to be released. In [8], Fu and Tzeng implement a tree of locks to remedy hot spot contention. Only the holder of the root lock has access to the data. Prior work on reader writer synchronization is largely for shared memory environment where the number of users trying to access data are typically much smaller than in ....

S. S. Fu, N. Tzeng, "A Circular List-Based Mutual Exclusion Scheme for Large Shared-Memory Multiprocessors," In Proc. IEEE Transactions on Parallel and Distributed Systems, Vol. 8, No. 6, June 1997.

....of reader writer locks [20] In this case, either readers have priority over writers or vice versa. Mellor Crummey and Scott have presented extensions of Algorithm MCS that support reader writer synchronization [48] Other authors have investigated algorithms that use circular waiting lists [27, 29, 30]. Note that, in Algorithm MCS, each process p nds its predecessor in the list by performing the fetch and store at line 3. However, p cannot identify its successor (if any) until that successor rst updates p s next eld. By using a circular list, this problem can be eliminated, because all nodes ....

....p cannot identify its successor (if any) until that successor rst updates p s next eld. By using a circular list, this problem can be eliminated, because all nodes can be reached by chaining through the list in the same direction. Despite this potential advantage, we found the algorithms in [27, 29, 30] to be quite dicult to understand, so we cannot assert that they really are an improvement over previous algorithms. Indeed, as shown in [29] by chaining through the list in only one direction, the FIFO nature of Algorithm MCS is lost (in that algorithm, processes enter their critical sections in ....

S. Fu and N.-F. Tzeng. A circular list-based mutual exclusion scheme for large shared-memory multiprocessors. IEEE Transactions on Parallel and Distributed Systems, 8(6):628-639, June 1997.

.... studies with different benchmark suites and programming models [8, 19] exemplify that the overhead of synchronization remains a dominant bottleneck in parallel programs for cache coherent shared memory multiprocessors, despite the enormousness of related work that appears in the literature [1, 4, 6, 7, 9, 11, 14, 16]. 1 To appear in the IEEE 2000 International Parallel and Distributed Processing Symposium, Cancun, Mexico, May 1 5 2000. Synchronization operations account frequently for significant fractions of execution time and may limit the scalability of parallel programs on large processor scales and ....

....executed by the processors during a synchronization period. Research on scalable synchronization for shared memory multiprocessors has concentrated on devising algorithms that minimize the number of remote references to synchronization flags during the acquire, waiting and release intervals [1, 4, 5, 7, 9, 14]. These algorithms have proven to be efficient on large scale shared memory systems 1 We define as scalability of a synchronization primitive the ability of the primitive to exhibit nearly constant latency as the number of processors that synchronize increases linearly [14] without hardware ....

[Article contains additional citation context not shown here]

S. Fu and N. Tzeng. A Circular List-Based Mutual Exclusion Scheme for Large Shared-Memory Multiprocessors. IEEE Trans. on Parallel and Distributed Systems, 8(6), pp. 628-- 639, Jun. 1997.

....required per synchronization operation, or the number method and show how it eliminates this source of synchronization of cycles between when one processor acquires a lock until the inefficiency. We have developed a model in synthesizable VHDL next processor can acquire that lock [LiA94] [FuT97] [KBG97] of the functionality required to implement our proposed method; When a processor is waiting to acquire a synchronization variable, space limitations preclude a detailed description. it performs no useful work. As each processor is running only one thread of the overall computation, from ....

S.S. Fu and N.F. Tzeng, "A circular list-based mutual exclusion scheme for large shared-memory multiprocessors", IEEE Transactions on Parallel and Distributed Systems, 8(6), pp 628-639, June 1997

No context found.

S. Fu and N. Tzeng. A circular list-based mutual exclusion scheme for large shared-memory multiprocessors. In IEEE Transactions on Parallel and Distributed Systems, pages 628--639, 1997.

No context found.

Shiwa S. Fu, Nian-Feng Tzeng, "A Circular List-Based Mutual Exclusion Scheme for Large Shared-Memory Multiprocessors", IEEE Transactions on Parallel and Distributed Systems, Vol. 8, No. 6, June 1997.

No context found.

S. Fu and N.-F. Tzeng. A circular list-based mutual exclusion scheme for large shared-memory multiprocessors. IEEE Transactions on Parallel and Distributed Systems, 8(6):628-639, June 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC