Shau-Ping Lo and Virgil D. Gligor. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In Proceedings of the Seventh International Conference on Distributed Computing Systems, pages 356--363, September 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....who observed that blocking should be avoided if wait times are short, and suggested pausing a waiting process for some fixed time before blocking. His Medusa system implemented two phase waiting with a user settable LpozZ. In a later study of multiprocessor scheduling algorithms, Lo and Gligor [22] found that use of two phase waiting (with LpozZ in between B and 2B) improved the performance of group scheduling. In a theoretical study of competitive algorithms, Karlin et al. presented a dynamic, randomized algorithm that achieves a competitive factor of e (e 1) They also proved that an ....

S. Lo and V. Gligor. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In 7th International Conference on Distributed Computing Systems, pages 356 363. IEEE, Sept. 1987.

....on a system, processes belonging to different jobs will compete for resources with each other and then some processes have to be blocked when communicating or synchronising with non scheduled processes on other processors. This effect can lead to a great degradation in overall system performance [4, 6, 9, 11, 13]. One method to alleviate this problem is to use two phase blocking [8, 22] which is also called implicit coscheduling in [8] In this method a process waiting for communication spins for some time in the hope that the process to be communicated with on the other processor is also scheduled, and ....

S.-P. Lo and V. D. Gligor, A comparative analysis of multiprocessor scheduling algorithms, Proceedings of the 7th International Conference on Distributed Computing Systems, Sept. 1987, pp.205-222.

....Systems For efficient computing, simultaneously running frequently communicating threads is a relatively good scheduling policy. Gang scheduling runs all threads of a parallel process simultaneously. Many studies have showed the superiority of gang scheduling to other scheduling policies [20, 19, 16, 6]. Gang scheduling has been implemented on various systems. Cedar[17] is a multiprocessor with Alliant FX 8 systems connected to a shared memory. Xylem operating system on Cedar[9] uses global and local queues to implement 2 level scheduling. An Alliant FX 8 in Cedar is called a cluster. First ....

S-P. Lo and V. D. Gligor. A comparative analysis of multiprocessor scheduling algorithms. In 7th Intl. Conf. Distributed Comput. Syst., pp. 356--363, September 1987.

.... the operating system leads to improved and predictable performance [417] 63 allows threads to be associated with data structures in local memory [106] Indeed, studies that compared gang scheduling with other scheduling schemes have concluded that gang scheduling is a relatively good policy [376, 511, 361, 249, 133, 106]. It should be noted, however, that other schemes most notably variable partitioning provide the same support. The advantage of gang scheduling (and the second desirable property) is that it provides interactive response times for short jobs, by virtue of using preemption [203] Just as ....

.... wait for threads that trail behind the others, and response time will be shortened [425] 6 Implications of the Sharing Scheme A number of studies have compared the effectiveness of different scheduling schemes, typically using their impact on the performance of specific applications as a metric [376, 384, 361, 249, 133, 198, 421, 106]. This section summarizes these results and adds many additional considerations, such as the support for different programming models and the influence on user satisfaction. These considerations are divided into four areas: the interaction between scheduling and the application, the impact on ....

S-P. Lo and V. D. Gligor, "A comparative analysis of multiprocessor scheduling algorithms". In 7th Intl. Conf. Distributed Comput. Syst., pp. 356--363, Sep 1987.

.... Synchronizing these interruptions in effect, gang scheduling among the application and the operating system leads to improved and predictable performance [255] 40 that compared gang scheduling with other scheduling schemes have concluded that gang scheduling is a relatively good policy [225, 306, 217, 152, 71, 57]. It should be noted, however, that other schemes most notably variable partitioning provide the same support. The advantage of gang scheduling (and the second desirable property) is that it provides interactive response times for short jobs, by virtue of using preemption [117] Just as ....

....more detail in Section 7.2.2. A variant based on feedback has also been proposed [311] 6 Implications of the Sharing Scheme A number of studies have compared the effectiveness of different scheduling schemes, typically using their impact on the performance of specific applications as a metric [225, 230, 217, 152, 71, 122, 259, 57]. This section summarizes these results and adds many additional considerations, such as the support for different programming models and the influence on user satisfaction. These considerations are divided into four areas: the interaction between scheduling and the application, the impact on ....

S-P. Lo and V. D. Gligor, "A comparative analysis of multiprocessor scheduling algorithms". In 7th Intl. Conf. Distributed Comput. Syst., pp. 356--363, Sep 1987.

....who observed that blocking should be avoided if wait times are short, and suggested pausing a waiting process for some fixed time before blocking. His Medusa system implemented two phase waiting with a user settable L poll . In a later study of multiprocessor scheduling algorithms, Lo and Gligor [22] found that use of two phase waiting (with L poll in between B and 2B) improved the performance of group scheduling. In a theoretical study of competitive algorithms, Karlin et al. presented a dynamic, randomized algorithm that achieves a competitive factor of e= e Gamma 1) They also proved that ....

S. Lo and V. Gligor. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In 7th International Conference on Distributed Computing Systems, pages 356--363. IEEE, Sept. 1987.

....blocked when communicating or synchronising with non scheduled processes on rpb169: In Proceedings of the 20th Australasian Computer Science Conference, Sydney, Australia, February 5 7 1997, pp 336 345. other processors. This effect can lead to a great degradation in overall system performance [1, 2, 5, 6, 10]. One method to alleviate this problem is to use two phase blocking [14] or adaptive two phase blocking algorithms [4] In this method a process waiting for communication spins for some time, and then blocks if the response is still not received. The reported experimental results show that for ....

S.-P. Lo and V. D. Gligor, A comparative analysis of multiprocessor scheduling algorithms, Proceedings of the 7th International Conference on Distributed Computing Systems, Sept. 1987, pp.205-222.

....were particularly surprising, suggesting that sequential portions typically run for only short periods of time. 6.3 Idle Period Time Distribution One approach to dealing with short idle periods is to filter them by waiting a short time before context switching. Ousterhout [25] Lo and Gligor [18], and Karlin et al. 15] take this approach in the context of implementing locks for mutual exclusion, where such filtering is called two phase blocking or spin then block. McCann et al. 20] have proposed a delayed reallocation scheme as part of a dynamic scheduling policy. 5 0 5 10 15 20 ....

S.-P. Lo and V. Gligor. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In Proceedings of the 7th International Conferenceon Distributed Computing Systems, pages 356--63, Sept. 1987.

....remains largely unexplored, but should receive increased attention in operating systems for large scale parallel machines like the Intel Paragon multiprocessor. 3.1. 3 Scheduling Policies A scheduling policy allocates available time and processors to a job or a process statically or dynamically [153]. Processor load balancing 2 is considered to be a part of a scheduling policy [232] Basic theoretical results on static process scheduling on parallel machines show that the scheduling problem is NP hard; static algorithms minimizing average response time include those described in [164] and ....

S. Lo and V. Gilgor. A comparative analysis of multiprocessor scheduling algorithms. In Proceedings of the 7th International Conference on Distributed Computing Systems, September 1987.

....run time choices between spinning and blocking. Ousterhout first proposed the two phase waiting algorithm in his Medusa operating system [47] The operating system implements two phase waiting with a user settable L poll . In a study of multiprocessor scheduling algorithms, Lo and Gligor [40] found that use of two phase waiting improves the performance of group scheduling when L poll is set in between B and 2B, where B is the cost of blocking. This thesis shows that the effectiveness of two phase waiting depends on both the distribution of waiting times and the setting of L poll . ....

S. Lo and V. Gligor. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In 7th International Conference on Distributed Computing Systems, pages 356--363. IEEE, Sept. 1987.

....on a system, processes belonging to different jobs will compete for resources with each other and then some processes have to be blocked when communicating or synchronising with non scheduled processes on other processors. This effect can lead to a great degradation in overall system performance [3, 4, 6, 7, 10]. One method to alleviate this problem is to use two phase blocking [12] or adaptive two phase blocking algorithms [5] In this method a process waiting for communication spins for some time, and then blocks if the response is still not received. The reported experimental results show that for ....

S.-P. Lo and V. D. Gligor, A comparative analysis of multiprocessor scheduling algorithms, Proceedings of the 7th International Conference on Distributed Computing Systems, Sept. 1987, pp.205-222. P1-S-7

.... the problem of splitting a computation specified as a task graph across two processors [Ni89] A study of processor allocation and scheduling issues in multiprocessors composed of non identical processors has been carried out by Menasce, et al. MDT92] From the studies discussed above and others [Li90, LG87, ST91], some conclusions that emerge by general consensus are: 1) space sharing is preferable to time sharing, yielding benefits analogous to those of round robin in single processor scheduling, while avoiding the overhead of frequent preemptions; 2) in UMA systems, the benefits of the adaptability ....

S.-P. Lo and V.D. Gligor, A Comparative Analysis of Multiprocessor Scheduling Algorithms, Proc. 7th Int. Conf. on Dist. Comp. Systems (Sept. 1987).

....application s processes will run at the same time, processes may be blocked while waiting for a preempted process or may be required to context switch after every synchronization operation. Several studies have shown that this effect can lead to severe performance degradation [Leutenegger, 1990; Lo and Gligor, 1987a; Lo and Gligor, 1987b; Tucker and Gupta, 1989] Many operating systems support a single centralized ready queue from which processes are dispatched according to their priority. This approach is very popular on small scale UMA machines. However, Squillante and Lazowska [1990] have shown that by ....

S.-P. Lo and V.D. Gligor, "A Comparative Analysis of Multiprocessor Scheduling Algorithms," In Proceedings 7th International Conference on Distributed Computing Systems, pages 205--222, September 1987.

....the tradeoff between spinning and blocking creates a second order of tradeoff between loss of cache affinity and interconnection network contention. A small spin duration implies a higher degree of locality violation, yet a longer one leads to a greater interconnection contention. Lo and Gligor [10] have compared critical path scheduling and coscheduling, they found that coscheduling achieved better performance. They credit this gain in performance mainly to the use of two phase waiting which involves spinning for a limited time, then blocking if the awaited event has still not occured. ....

S. Lo and V. Gligor. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In 7th International Conference on Distributed Computing Systems, pages 356-363, Sept. 1987.

No context found.

Shau-Ping Lo and Virgil D. Gligor. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In Proceedings of the Seventh International Conference on Distributed Computing Systems, pages 356--363, September 1987.

No context found.

Lo, S.-P. and Gligor, V. A Comparative Analysis of Multiprocessor Scheduling Algorithms. In Proceedings of the 7th International Conference on Distributed Computing Systems, pages 356-363, September 1987.

No context found.

S.P. Lo, V. Gligor, "A Comparative Analysis of Multiprocessor Scheduling Algorithms", 7th International Conference on Distributed Computing Systems, September 1987.

No context found.

LO, S.-P., AND GLIGOR, V. A comparative analysis of multiprocessor scheduling algorithms. In Proceedings of the 7th International Conference on Distributed Computing Systems (Sept. 1987), pp. 356--363.

No context found.

Lo, S.-P., AND GLIGOR, V. A comparative analysis of multiprocessor scheduling algorithms. In Proceedings of the 7th International Conference on Distributed Computing Systems (Sept. pp 356-363.

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