J.-P. Li and M. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proc. Intl. Parallel Processing Symposium, pages 433--438, May 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....performance requires matching application communication characteristics with a suitable network design. Parallel applications impose a wide range of communication patterns on the underlying interconnection network. Scientific computations [1, 2] parallel databases, and real time applications [3, 4] generate distinct distributions for message lengths, interarrival times, and target destination nodes. Finding a suitable network design to support such diverse communication requirements is difficult since there are a myriad of design factors which can potentially impact performance. Network ....

J.-P. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in Proc. International Parallel Processing Symposium, pp. 433--438, April 1994.

....teardown. Resource deallocation must be done by network management, which periodically discovers allocated, but unused blocking times, and makes them available to new connections. The only assumption we make about the switches in this work is that they be fair. There are current efforts (e.g. [9]) to study the benefits of special support for real time traffic in wormholerouted switches. We are investigating how such additional functionality (e.g. priorities for packets) affects the performance of our scheme. ....

J.-P. Li and M. W. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proceedings of the IEEE IPPS '94, 8th International Parallel Processing Symposium, pp. 433-438, April, 1994.

....performance requires matching application communication characteristics with a suitable network design. Parallel applications impose a wide range of communication patterns on the underlying interconnection network. Scientific computations [1,2] parallel databases, and real time applications [3,4] generate distinct distributions for message lengths, interarrival times, and target destination nodes. Finding a suitable network design to support such diverse communication requirements is difficult since there are a myriad of design factors which can potentially impact performance. Network ....

J.-P. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in Proc. hternational Parallel Processing Symposium, pp. 433 438, April 1994.

....A. Wormhole Routing Networks Wormhole routing networks with backpressure flow control require bidirectional links, thereby being a natural context for the use of the bidirectional shufflenet. Wormhole routing is often the scheme of choice in low latency high speed networks [14] 21] 3] [16]. In wormhole routing, the (variable size) unit of information transfer is a worm, which can range in length from a few bytes to several thousand bytes. Low latency is achieved by using cut through or wormhole routing instead of the conventional store and forward approach: each intermediate node ....

J.-P. Li and M. Mutka, "Priority based real-time communication for large scale wormhole networks," in Proc. Int. Parallel Processing Symp., May 1994, pp. 433--438.

....requirements with a suitable network design. However, applications employ a wide variety of communication paradigms that affect the quantity and frequency of communication between nodes. Parallel applications such as scientific computations [1, 2] parallel databases, and real time applications [3] generate distinct distributions The work reported in this paper was supported in part by the National Science Foundation under grant MIP 9203895. Any opinions, findings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views ....

J.-P. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in Proc. International Parallel Processing Symposium, pp. 433-- 438, April 1994.

....without waiting for the entire worm to be assembled. Thus, the worm can stretch across several nodes and links at any one time. Wormhole routing with backpressure flow control was originally proposed for multiprocessor computer communications [17] and later adapted to local area networking [10, 4, 7, 13, 14, 22, 23]. An example of a wormhole routing network is Myrinet [4] Myrinet LANs use source routing and 8bit wide data transmission to achieve data rates up to 640 Mbit s. However, their geographical reach is limited since the maximum link length is only about 25 meters. The ARPA sponsored Supercomputer ....

.... are topology dependent (i.e. the number of virtual channels required to allow deadlock free shortest path routing depends upon the topology and the routing function) Previous work in the area of virtual channels has in part focused on throughput advantages and on Quality of Service support [7, 13, 14]. In these studies, unrestricted shortest path routing is assumed which may result in deadlocks. Other works [8, 20] address the use of virtual channels for deadlock prevention. Several protocols have been proposed for the realization of virtual channels with deadlockfree routing [20] These ....

J-P. Li and M. Mutka. Priority Based real-Time Communication for Large Scale Wormhole Networks. In Proceedings of International Parallel Processing Symposium, pages 433--438, May 1994.

....processes to meet their real time deadlines. Multiprocessor scheduling problems are well known to be NP Complete [8] for most precedence constrained tasks. Because of this, heuristic methods have been employed to reach these bounds. These heuristic methods attempt to either dynamically [9] have the messages meet their deadlines or to determine a priori some way of meeting the deadlines. The determination of a schedule is the key to both approaches and many techniques have been implemented [10, 11] What is often neglected in work done on multiprocessor communication is a discussion ....

....of messages is dealt with by each node trying to send its messages through the interconnect as soon as the messages are ready. This can be considered a first come first served, of FCFS approach. Then, once the message has started, techniques are used to either delay it or to let it pass through [9]. Priorities are often assigned to messages to let the more time critical ones process first. In this paper it is shown that it is beneficial to consider the scheduling of these messages to obtain better performance. Consider as an example a filter section represented by the task graph of Figure ....

J. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks, " in International Conference on Parallel Processing Symposium, pp. 433--438, April 1994.

....higher priority transactions have a greater probability of being met than those of lower priority transactions. While this is also the case for MSHN and BADD AICE, the research presented here includes other attributes that are important in evaluating the overall value of the tasks completed. In [17], priority is used for the mapping, adjustment, and dropping of messages. The priority of a task is adjusted by the length of time it was blocked by a higher priority message and tardy messages that already missed their deadlines are dropped. This Least Laxity First priority mapping gives an ....

....of a task is adjusted by the length of time it was blocked by a higher priority message and tardy messages that already missed their deadlines are dropped. This Least Laxity First priority mapping gives an improved missed deadline ratio, which is the rate of messages missing their deadlines. In [17], priority and deadline is used as measures of performance but the research effort does not consider other QoS attributes used in heterogeneous distributed networks. An algorithm that allows transmission of messages belonging to several classes of situational mode is presented in [20] The ....

J. P. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," 8th Int'l Parallel Processing Symp., Apr. 1994, pp. 433-438.

....in a multi access shared bus network. Some scheduling policies such as Virtual Clock [23] Stop and Go [6] Rotating Combined Queuing [9] do not explicitly use message deadlines, but just attempt to keep the worst case message delay small and bounded. Other relevant experimental work includes [16, 22, 14, 21, 5]. Of course, routing without timing constraints has been the subject of a large number of works; see [11, 12] for a survey. These works usually attack the problem under a best effort model, and thus focus on optimizing global performance measures such as the overall completion time of a routing ....

J.-P. Li and M.W. Mutka (1994): Priority based realtime communication for large scale wormhole networks. Intl. Parallel Proc. Symp., 433-438.

....and can adapt to changes in the system. Because parallel machines are increasingly being used in real time applications, researchers are very interested in providing some type of time boundary on message transfers. Many examples of heuristically determined dynamic schedules have been proposed [14 17]. Priorities are often assigned to messages to let the more time critical ones process first. However, when considering the ordering of the message transmissions, a dynamic approach has the disadvantage of not providing the designer with a guarantee that the time constraints can be satisfied. ....

J. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in International Conference on Parallel Processing Symposium, pp. 433--438, April 1994.

....This periodically discovers allocated, but unused resources ( due to connections torn down ) and makes them available to new connections. The only assumption we make about the switches in this work is that they serve waiting packets in a roundrobin manner. There are current efforts (e.g. [10]) to study the benefits of special support for real time traffic in wormhole routed switches. We are currently investigating how such additional functionality (e.g. priorities for packets) affects the performance of our scheme. Substantial work is currently being done on host and switch supported ....

J.-P. Li and M. W. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proceedings of the IEEE IPPS '94, 8th International Parallel Processing Symposium, pp. 433-438, April, 1994.

....attributes of both static and dynamic approaches. This new approach will use known information that can be ascertained in advance to determine priorities to be assigned to the individual messages. This priority is different from time based priorities determined for each message as discussed in [26]. Rather, the priorities are based on producing the best overall processing time for the entire message list. At run time a node will act dynamically as its first requirement by trying to transmit its messages which have the earliest departure time when the channel is available. However, where ....

J. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in International Conference on Parallel Processing Symposium, pp. 433--438, April 1994.

....if Ti N 1 A message can request to be allocated any virtual channel with lower number than its priority (then, a high priority message can request a larger number of virtual channels) increasing the probability to meet deadlines of messages with tight deadlines. 8.3. Least laxity first (LLF) In [Li94b], the laxity up to the message deadline is used as a priority. The laxity is the remaining time up to message delivery, and it can be estimated as Lx i = d i latency i . Let us call N to the number of virtual channels, W to the estimated time of the message blocking before the priority is ....

....limiting the number of virtual channels a message can request, LLF increase the chance that a message with smaller laxity can meet its deadline. However, the messages with low priorities can be blocked by higher priority messages for an indefinite period. 8.4. Priority Climbing (PC) In this scheme [Li94b], the priority of a message is adjusted according to its current state, increasing the priority of blocked messages. Hence, high priority messages cannot block low priority ones for an indefinite period. If each physical link has 2 n virtual channels, n bits are used to represent all the priority ....

[Article contains additional citation context not shown here]

LI, P.; MUTKA, M. "Priority based real-time communication for large scale wormhole networks". Proceedings. of the IEEE International Parallel Symposium. 1994.

....to Processors processors. By using a new model presented by Surma and Sha in [2] known as a Collision Graph, a new scheduling framework able to deal with real life problems is developed. The problem of static scheduling for computational tasks has been studied by researchers for many years [3, 4]. Because parallel machines are increasingly being used in real time applications, researchers are very interested in providing some type of time boundary on message transfers. Multi processor scheduling problems are well known to be NP Complete for most precedence constrained tasks. Because of ....

J. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in International Conference on Parallel Processing Symposium, pp. 433--438, April 1994.

....This periodically discovers allocated, but unused resources ( due to connections torn down ) and makes them available to new connections. The only assumption we make about the switches in this work is that they serve waiting packets in a round robin manner. There are current efforts (e.g. [10]) to study the benefits of special support for real time traffic in wormhole routed switches. We are currently investigating how such additional functionality (e.g. priorities for packets) affects the performance of our scheme. Substantial work is currently being done on host and switch ....

J.-P. Li and M. W. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proceedings of the IEEE IPPS '94, 8th International Parallel Processing Symposium, pp. 433-438, April, 1994.

.... and sufficient conditions for the same problem [4] Research results on a fault tolerant version of the real time channel [2] and a router architecture designed to support real time channel have been reported recently [7] Real time communication in wormhole switching networks was addressed by Li [8], Mutka [9] and Song [10] recently. In Li s approach, there are as many virtual chan nels as priority levels and each virtual channel is assigned a unique priority level. A packet can request virtual channels which are numbered lower than or equal to its priority. By doing this, the ....

....levels and each virtual channel is assigned a unique priority level. A packet can request virtual channels which are numbered lower than or equal to its priority. By doing this, the probability that a higher priority packet can obtain a virtual channel is higher than that of lower priority packets [8]. Song proposed a new flow control method for real time communications in wormhole networks. With a special router architecture, flit level preemption of physical channels among prioritized packets can be implemented using a smaller number of virtual channels [10] While Song s and Li s methods ....

J. P. Li and M. W. Mutka, "Priority Based Real-Time Communication for Large Scale Wormhole Networks," Proc. of the IPPS, 1994.

....requires matching application communication characteristics with a suitable network design. However, parallel applications impose a wide range of communication patterns on the underlying interconnection network. Scientific computations [1, 2] parallel databases, and realtime applications [3, 4] generate distinct distributions for message lengths, interarrival times, and target destination nodes. Message lengths can vary depending on packetization policies as well as the type of communication (e.g. data requests, data transfers, and acknowledgement messages) while message interarrival ....

J.-P. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in Proc. International Parallel Processing Symposium, pp. 433--438, April 1994.

....Providing such multicast service is an easy task on shared medium LANs with linear topology, but it becomes anything but easy on deadlock prone wormhole networks with mesh topology. 2. Wormhole Routing Networks Wormhole routing is often the scheme of choice in lowlatency, high speed networks [12, 15, 13]. Asynchronous wormhole routing LANs are a promising and cost effective alternative to ATM for high bandwidth, low latency applications in the local area and campus environments. They can also be considered an interesting alternative with respect to the IEEE 802.3z Gigabit Ethernet. In wormhole ....

J.-P. Li and M. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proceedings of International Parallel Processing Symposium, pages 433-- 438, May 1994.

No context found.

J.-P. Li and M. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proc. Intl. Parallel Processing Symposium, pages 433--438, May 1994.

No context found.

J.-P. Li and M. Mutka, "Priority Based Real-Time Communication for Large Scale Wormhole Networks," in Proceedings of International Parallel Processing Symposium, May 1994, pp. 433--438.

No context found.

J.-P. Li and M. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proc. of Intl. Parallel Processing Symp., pages 433--438, May 1994.

No context found.

J.-P. Li and M. Mutka. Priority Based Real-Time Communication for Large Scale Wormhole Networks. In Proc. of Intl. Parallel Processing Symp., pages 433--438, May 1994.

No context found.

J. P. Li and M. W. Mutka, " Priority based real-time communication for large scale wormhole networks," 8th International Parallel Processing Symposium, Apr. 1994, pp. 433-438.

No context found.

J. P. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," 8th International Parallel Processing Symposium, Apr. 1994, pp. 433-438.

No context found.

J.-P. Li and M. W. Mutka, "Priority based real-time communication for large scale wormhole networks," in Proc. International Parallel Processing Symposium, pp. 433--438, April 1994.

First 50 documents

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