PFISTER, G. F., AND NORTON, V. A. Hot Spot Contention and Combining Multistage Interconnection Networks. IEEE Transactions on Computers C-34, 10 (1985), 943--8.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....This may lead to the output being loaded more than 100 . Such a condition may exist for a transient period of time. Traffic engineering, traffic control and routing should prevent such situations from occurring for extended periods of time. The hot spot model we use here is the same as that in [16], 15] a high rate of traffic is destined to one hot spot and all other traffic is distributed to other outputs uniformly. Define # as the fraction of cells destined to the hot spot. Then the cell arrival rate # on one input port can be expressed as: # # ## #### ### (7) For one input, a cell ....

G. F. Pfister, "`Hot Spot' contention and combining in multistage interconnection networks", IEEE Trans. on Computers, vol. C-34, No. 10, pp. 943-948, Oct. 1985.

....improves hot spot throughput as described below. C. Hot Spot Throughput In many situations traffic is not uniform, but rather is concentrated into hot spots. A hot spot is a pair of nodes that accounts for a disproportionately large portion of the total network traffic. As described by Pfister [18] for a shared memory computer, hot spot traffic can degrade performance of the entire network by causing congestion. The hot spot throughput of a network is the maximum rate at which messages can be sent from one specific node Pi to another specific node Pj. For a k ary n cube with deterministic ....

G.F. Pfister and V. A. Norton, "Hot spot contention and combining in multistage interconnection networks," 1EEE Trans. Cornput., vol C-34, no. 10, pp. 943-948, Oct. 1985.

....an arbitrary number of threads; and robustness the time it takes to perform operations should minimize sensitivity to load fluctuations. Finally, the method should be widely applicable to avoid the need to invent a new synchronization protocol for every application. It is well documented [1, 7, 12, 21] that concurrent access to a single object by many threads can lead to a degrada tion in performance due to contention. A relatively well established method which has been used to alleviate this hot spot contention is combining. Originally designed by Gottlieb et as. to be used in switches of a ....

G.H. Pfister and A. Norton. 'Hot Spot' contention and combining in multistage interconnection networks. IEEE 2ansactions on Computers, C-34(11):933-938, November 1985.

....or write into the same memory block. This experiment models the behavior of a single I O server being accessed by multiple clients and provides basic results for better understanding the network. This localized communication pattern can lead to a severe form of congestion known as tree saturation [18], which can seriously degrade the overall performance of the interconnect. The network traffic generated by a parallel job that is performing input output can be modeled with a collection of hotspots, where each hot spot is a node that acts as an I O server 5 and is the target of multiple ....

G.F. Pfister and V.A. Norton, Hot-spot contention and combining in multistage interconnection networks, IEEE Transactions on Computers C-34(10) (1985) 943-948.

....packet, as long as the multicast set is physically contiguous. For a multicast packet to be successfully delivered, a positive acknowledgment must be received from all the recipients of the multicast group. The Elite switches combine the acknowledgments, as pioneered by the NYU Ultracomputer [4] [24], returning a single one to the source. Acknowledgments are combined in a way that the worst ack wins (a network error wins over an unsuccessful transaction, which on its turn wins over a successful one) returning a positive ack only when all the partners in the collective communication ....

G. F. Pfister and V. A. Norton. Hot-spot Contention and Combining in Multistage Interconnection Networks. IEEE Transactions on Computers, C-34(10):943--948, October 1985.

....destinations may be more favored by the sources. To analyze the network performance in a more practical situation, the normal distribution is also used to generate destination addresses or hot spots. Hot spots cause a disproportionate number of packets to be routed to a subset of the destinations [DeC89, PfN85]. This is done by generating packets over a range of destination addresses with a preferential concentration around a mean destination address. For the simulations with hot spots, packet destinations are chosen using a normal distribution with a specified mean and standard deviation. The mean is ....

G. F. Pfister and V. A. Norton, "Hot spot contention and combining in multistage interconnection networks," IEEE International Conference on Parallel Processing, pp. 790-797, 1985.

....is described in [15 I. It is shown that certain nonuniform traffic patterns can have a crucial influence on the performance of the MIN. Kim 16 presented an analytical model to evaluate the performance of a single input buffered Banyan switch under nonuniform traffic patterns. Pfister and Norton [17] reported on a quantitative investigation of the performance impact of memory contention in highly parallel shared memory multiprocessors. They first investigated the effect of a nonuniform traffic pattern consisting of a single hot spot of higher access rate superimposed on a background of ....

....SE. 10. The uniform traffic pattern is defined to be the traffic pattern in which every source port has the same rate of incoming packets and they are destined to every destination port with the same probability l N, where N is the number of output port. 11. For a hot spot traffic pattern [17], a large proportion of the traffic is directed to a particular output called the hot output. The packets destined to the hot output are called hot packets. In Fig. 2, 04 is the hot output and requests for O4 are called hot requests. The internal links of the switch that carry hot packets are ....

[Article contains additional citation context not shown here]

G.F. Pfister & V.A. Norton, Hot spot contention and combining in multistage interconnection networks, IEEE Trans. Cornput., C-3J(10), 1985, 943-948.

....92 Chapter I The Hot Spot Problem 1. 1 Introduction In high speed routing networks, slight imbalances in the traffic distribution can have dra matic effects on network performance [18] One particularly important type of imbalance a hot spot was discovered recently by Pfister and Norton [17]. Variations of this type of imbalance were further studied by Lee [13] When a hot spot occurs, a par ticular communication link experiences a much greater number of requests than the rest of the links many more than it can service. In a remarkably short period of time, the entire network ....

....network links (in bold) are shared by several paths. The same is true for all of the network s internal buffers. This shaxing causes little contention in the networks if the traffic load is balanced [15] However, if the traffic distribution is uneven, radically lower performance can result [13,17,18]. The performance degradation is direct result of the sharing of links and buffers within the communication network. A consequence of this widespread sharing is that when a hot spot occurs, all of the network inputs may experience throughput degradation and increased delay. When a hot spot ....

[Article contains additional citation context not shown here]

G. F. Pfister and V. A. Norton. Hot spot contention and combining in multistage inter- connection networks. IEEE Transactions on Computers, C-34(10):943-948, October 1985.

....4.3 Hotspot A further test of the network and the network interface is the hotspot. Under hotspot traffic, a set of communication partners try to read from or write into the same memory block. This localized communication pattern can lead to a severe form of congestion known as tree saturation [10], which can seriously degrade the performance of the overall network. In our experiments we will consider both read and write hotspots. 5 Experimental Results 5.1 Unidirectional Ping Figure 4 a) compares the performance of the unidirectional ping. The asymptotic bandwidth for all communication ....

G. F. Pfister and V. A. Norton. Hot-spot Contention and Combining in Multistage Interconnection Networks. IEEE Transactions on Computers, C-34(10):943--948, October 1985.

....avoiding the spread from the congestion area. The experiments reported are designed to examine the effects of the congestion control in detail. A. INTRODUCTION. An important issue in traffic management of multistage interconnects is the handling of tree saturation (caused by hot spot traffic[6]) and the impact that tree saturation can have on unrelated flows. In a production multi user parallel machine such as the AC3 Velocity cluster, this is particularly important as traffic patterns are not predictable, and hot spots cannot be avoided through application level structuring. There ....

Pfister, G.F. and V.A. Norton, Hot Spot Contention and Combining Multistage Interconnection Networks, IEEE Transactions on Computers, C-34(10). 1985.

....4.5 Hotspot A further test of the network and the network interface is the hotspot. Under hotspot traffic, a set of communication partners try to read from or write into the same memory block. This localized communication pattern can lead to a severe form of congestion known as tree saturation [17], which can seriously degrade the overall performance of the network. 4.6 Multiple Hotspots Experiments with multiple hotspots can provide important information on the behavior of the network under heavy load. The network traffic generated by a parallel job that is performing input output (I O) ....

G. F. Pfister and V. A. Norton. Hot-spot Contention and Combining in Multistage Interconnection Networks. IEEE Transactions on Computers, C34 (10):943--948, October 1985.

.... 7 Related Work Nearly all of the past work on reduction parallelization has been based on software only transformations [8, 27] The most related architectural work that we are aware of is the work of Larus et al. 20] Zhang et al. 28] and the work on advanced synchronization mechanisms [3, 9, 10, 16, 17, 18, 23, 24, 25, 29]. Larus et al. briefly mention an idea similar to PCLR as one application of their Reconcilable Shared Memory (RSM) 20] RSM is a family of memory systems whose behavior can be controlled by the compiler. They use RSM to support programming language constructs. The paper only mentions the ....

....to the existing body of work on hardware support for synchronization. Such work includes the Full Empty bit of the HEP multiprocessor [25] the atomic Fetch Add primitive of the NYU Ultracomputer [10] the Fetch Op synchronization primitives of the IBM RP3 [3, 23] support for combining trees [16, 24], the memory based synchronization primitives in Cedar [17, 18, 29] and the set of synchronization primitives proposed by Goodman et al. [9] 8 Summary In this paper, we have proposed new architectural support to speed up parallel reductions in scalable sharedmemory multiprocessors. The support ....

G. Pfister and A. Norton. 'Hot Spot' Contention and Combining in Multistage Interconnection Networks. In Proc.

No context found.

PFISTER, G. F., AND NORTON, V. A. Hot Spot Contention and Combining Multistage Interconnection Networks. IEEE Transactions on Computers C-34, 10 (1985), 943--8.

No context found.

G. F. Pfister and V. A. Norton. Hot-spot Contention and Combining in Multistage Interconnection Networks. IEEE Transactions on Computers, C-34 (10), pages 943--048, 1985.

No context found.

PFISTER, G. F., AND NORTON, V. A. Hot Spot Contention and Combining Multistage Interconnection Networks. IEEE Transactions on Computers C-34, 10 (1985), 943--8.

No context found.

G. F. Pfister and V. A. Norton. "Hot Spot contention and combining in multistage interconnection networks," IEEE Trans. Comput., vol. C-34, no. 10, pp. 934-948, Oct. 1985.

No context found.

G. F. Pfister and V. A. Norton, "Hot spot contention and combining in multistage interconnection networks," IEEE Transactions on Computers, vol. 34, no. 10, pp. 943--948, October 1985.

No context found.

G. F. Pfister and V. A. Norton, "Hot Spot Contention and Combining in Multistage Interconnection Networks," IEEE Transactions on Computers, Vol.34, No.10, October 1985, pp.943--948.

No context found.

Pfister, G.F., Norton, V.A.: Hot spot contention and combining in multistage interconnect networks. IEEE Transactions on Computers C-34 No. 10 (1985) 943--948

No context found.

G.F. Pster, V.A. Norton, Hot spot contention and combining in multistage interconnect networks, IEEE Trans. Comp. C-34 (10) (1985) 943948.

No context found.

G. F. Pfister and A. Norton. Hot spot contention and combining in multistage interconnection networks. IEEE Trans. on Computers, 34(10):943--948, Octuber 1985.

No context found.

Pfister, G. F. and Norton, V. A., "Hot spot contention and combining in multistage interconnection networks," IEEE Transactions on Computers, Vol. 34, pp. 943--948, October 1985.

No context found.

G. F. Pfister and V. A. Norton. Hot-Spot Contention and Combining in Multistage Interconnection Networks. IEEE Transactions on Computers, C-34(10):943--948, October 1985.

No context found.

G. Pfister and V. Norton. Hot Spot Contention and Combining in Multistage Interconnection Networks. In Proc. of the 1985.

No context found.

G. F. Pfister and V. A. Norton, "Hot-spot Contention and Combining in Multistage Interconnection Networks," IEEE Transactions on Computers, C-34(10):943--948, October 1985.

First 50 documents  Next 50

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