Exploiting the capabilities of the interconnection network on Dawning-1000

On Dawning-1000,the two-dimension mesh interconnection network enables low-latency,high-bandwidth communication,however,these capabilities have not been realized because of the high processing overhead imposed by existing communication software.Active Messages provide an efficient communication mechanism with small overhead,which may expose the raw capabilities of the underlying hardware.In addition,one of the most promising techniques,use-level communication,is often used to improve the performance of the traditional protocols such as TCP and UDP,and is also adopted in implementing the novel abstractions like Active Messages.Thus a user-level Active Messages model is designed and implemented on Dawning-1000.Preliminary experiments show that the combination of Active Messages mechanism and user-level communication technique is quite efficient in reducing software overhead associated with sending and receiving messages,and in exploiting the capabilities of the interconnection network.     

A unified formulation of Kautz network and generalized hypercube

Hypercube and Kautz network each possess certain desirable properties. However, some of the attractive features of one network are not found in the other. A novel class of network topologies proposed in this paper has the generalized hypercube and the Kautz network as its two extremes. The proposed network inherits the topological properties of both the Kautz network and the generalized hypercube to a varying degree. This allows us to trade-off cost and performance effectively and construct networks which are most suitable for a particular purpose. In the present paper, we investigate the connectivity, wide-diameter, fault-tolerance, Hamiltonicity.     

Adaptive routing protocols for hypercube interconnection networks

A taxonomy for characterizing adaptive routing protocols for hypercube interconnection networks (HINs) is presented. The taxonomy is based on classes of routing decisions common to any HIN. This taxonomy is used to discuss existing and proposed protocols. Rather than an exhaustive enumeration of related research, the protocols selected for discussion are intended to be representative of the classes defined by the taxonomy. These protocols are candidates for use in massively parallel architectures configured with HINs. To provide some insight into their behavior in very large HINs, results of simulation studies of representative protocols are presented     

A spanning multichannel linked hypercube: a gradually scalableoptical interconnection network for massively parallel computing

A new, scalable interconnection topology called the Spanning Multichannel Linked Hypercube (SMLH) is proposed. This proposed network is very suitable to massively parallel systems and is highly amenable to optical implementation. The SMLH uses the hypercube topology as a basic building block and connects such building blocks using two-dimensional multichannel links (similar to spanning buses). In doing so, the SMLH combines positive features of both the hypercube (small diameter, high connectivity, symmetry, simple routing, and fault tolerance) and the spanning bus hypercube (SBH) (constant node degree, scalability, and ease of physical implementation), while at the same time circumventing their disadvantages. The SMLH topology supports many communication patterns found in different classes of computation, such as bus-based, mesh-based, and tree-based problems, as well as hypercube-based problems. A very attractive feature of the SMLH network is its ability to support a large number of processors with the possibility of maintaining a constant degree and a constant diameter. Other positive features include symmetry, incremental scalability, and fault tolerance. It is shown that the SMLH network provides better average message distance, average traffic density, and queuing delay than many similar networks, including the binary hypercube, the SBH, etc. Additionally, the SMLH has comparable performance to other high-performance hypercubic networks, including the Generalized Hypercube and the Hypermesh. An optical implementation methodology is proposed for SMLH. The implementation methodology combines both the advantages of free space optics with those of wavelength division multiplexing techniques. A detailed analysis of the feasibility of the proposed network is also presented     

Parallel routing algorithms for incomplete hypercube interconnection networks

Hypercube interconnection networks have been receiving considerable attention in the supercomputing environment. However, the number of processors must be exactly 2^r for an r-cube complete hypercube. This restriction severely limits its applicability. In this paper, we address three variant hypercube topologies with more flexibility in system sizes, the labelled hypercubes I_m^r, I_M^r, and I_A^r. Incomplete hypercube I_m^r consists of an r-cube and an m-cube complete hypercubes; I_m^r is composed of 2^r and Σ_{m ε M} 2^m nodes; I_Ar comes from an r-cube complete hypercube which operates in a degraded manner and allows that the missing nodes to be arbitrarily distributed. Specifically, we focus on the parallel paths routing algorithms for these three classes of incomplete hypercubes. Parallel paths between any given two nodes mean that these paths have the same source and destination nodes but with different intermediate nodes. Parallel communication is important as it will allow us to use the full bandwidth of the multiprocessors for the data transfer operation between any two nodes, and3these redundant paths can increase system fault-tolerance and communication reliability. With these parallel routing algorithms, one can use them as a criterion to design multiprocessor systems.     

A new hierarchy of hypercube interconnection schemes for parallel computers

This paper introduces a new hierarchy of cube-based interconnection schemes, called the base-b cube (which properly contains the well-known binary cube), for the design of parallel computers. This hierarchy admits a recursive definition and allows many more reconfigurations than are possible with the binary cube. Our analysis addresses the inherent cost-delay trade-off for this hierarchy along with a number of related topological properties such as sparsity, diameter, existence of node disjoint paths, and odd and even cycles. Embeddings of standard interconnection schemes including linear and two-dimensional arrays, rings, and complete binary trees in a base-b cube are illustrated.     

Fault-tolerant routing methodology for hypercube and cube-connected cycles interconnection networks

This paper presents a fault-tolerant routing methodology for both injured hypercube and cube-connected cycles interconnection topologies. The proposed routing methodology efficiently tolerates any pattern of faulty regions with any number of faulty nodes in the network which is based on the best-first search and backtracking strategy. Deadlock freedom of the proposed routing methodology is obtained by only one virtual channel per physical channel. In order to evaluate the proposed routing methodology, a 7-dimensional hypercube network is simulated in various conditions, i.e., different traffic rates, different number of faulty nodes and different message lengths. Simulation results confirm that the proposed routing methodology in comparison with the previous methods provides acceptable performance while it significantly increases the reliability of the network. It also guarantees delivery of messages between any pair of source and destination while the network is connected.     

Computer network interconnection

This report examines the current situation regarding the interconnection of computer networks, especially packet switched networks (PSNs). Four major types of interconnections are surveyed:

• 1.1. Circuit Switched Network to PSN
• 2.2. Star Network to PSN
• 3.3. Simple Terminal to PSN
• 4.4. PSN to PSN

The emphasis is on identifying the barriers to interconnection and on surveying approaches to a solution, rather than recommending any single course of action.     

Routing algorithms on the bus-based hypercube network

In this paper, we study the properties of the bus-based hypercube, denoted as U(n,b), which is a kind of multiple-bus networks (MBN). U(n,b) consists of 2/sup n/ processors and 2/sup b/ buses, where 0 /spl les/ b /spl les/ n - 1, and each processor is connected to either /spl lceil/(b+2)/2/spl rceil/ or /spl lceil/(b+1)/2/spl rceil/ buses. We show that the diameter of U(n,b) is /spl lceil/(b-1)/2/spl rceil/ if b /spl ges/ 2. We also present an algorithm to select the best neighbor processor via which we can obtain one shortest routing path. In U(n,b), we show that if there exist some faults, the fault diameter DF(n,b,f) /spl les/ b+1, where f is the sum of bus faults and processor faults and 0 /spl les/ f /spl les/ /spl lceil/(b+3)/2/spl rceil/. Furthermore, we also show that the bus fault diameter DB(n,b,f) /spl les/ b/-2/spl rfloor/ - 3, where 0 /spl les/ f /spl les/ /spl lceil/(b-1)/2/spl rceil/ and f is the number of bus faults. These results improve significantly the previous result that DB(n,b,f) /spl les/ b - 2f + 1, where f is the number of bus faults.     

FBI’s communications surveillance capabilities widen

The US Government has established yet another national surveillance centre in northern Virginia’s burgeoning ‘Spy Tech Corridor’, a swath of hush-hush agencies extending from Central Intelligence Agency headquarters in Langley, to a hodgepodge of office complexes near Dulles Airport. Located in the pristine Westfields Office Park and a stone’s throw away from the super secret National Reconnaissance Office is the FBI’s ‘CALEA Implementation Section’. CALEA stands for the Communications Assistance to Law Enforcement Act.     

Reliable multistage interconnection network design

High-performance supercomputers generally comprise millions of CPUs in which interconnection networks play an important role to achieve high performance. New design paradigms of dynamic on-chip interconnection network involve a) topology b) synthesis, modeling and evaluation c) quality of service, fault tolerance and reliability d) routing procedures. To construct a dynamic highly fault tolerant interconnection networks requires more disjoint paths from each source-destination node pair at each stage and dynamic rerouting capability to use the various available paths effectively. Fast routing and rerouting strategy is needed to provide reliable performance on switch/link failures. This paper proposes two new architecture designs of fault tolerant interconnection networks named as reliable interconnection networks (RIN-1 and RIN-2). The proposed layouts are multipath multi-stage interconnection networks providing four disjoint paths for all the source-destination node pairs with dynamic rerouting capability. The designs can withstand switch failures in all the stages (including input and output stages) and provide more reliability. Reliability analysis of various MIN architectures is evaluated. On comparing the results with some existing MINs it is evident that the proposed designs provides higher reliability values and fault tolerance.     

The pancyclicity and the Hamiltonian-connectivity of the generalized base-b hypercube

The interconnection network considered in this paper is the generalized base-b hypercube that is an attractive variance of the well-known hypercube. The generalized base-b hypercube is superior to the hypercube in many criteria, such as diameter, connectivity and fault diameter. In this paper, we study the Hamiltonian-connectivity and pancyclicity of the generalized base-b hypercube by the algorithmic approach. We show that a generalized base-b hypercube is Hamiltonian-connected for b ? 3. That is, there exists a Hamiltonian path joining each pair of vertices in a generalized base-b hypercube for b ? 3. We also show that a generalized base-b hypercube is pancyclic for b ? 3. That is, it embeds cycles of all lengths ranging from 3 to the order of the graph for b ? 3.     

The hierarchical cliques interconnection network

The fully connected network possesses extremely good topological, fault-tolerant, and embedding properties. However, due to its high degree, the fully connected network has not been an attractive candidate for building parallel computers. On the other hand, tree-based networks are popular as parallel computer networks, even though they suffer from poor fault-tolerance and embedding properties. The hierarchical cliques interconnection network described in this paper incorporates positive features of the fully connected network and the tree network. In other words, the hierarchical cliques possess such desirable properties as low diameter, low degree, self routing, versatile embedding, good fault-tolerance and strong resilience. Hierarchical cliques can efficiently embed most important networks and possess a scalable, modular structure. Further, by combining hierarchical cliques with fat trees congestion in the upper levels can be alleviated.     

Limits on interconnection network performance

The latency of direct networks is modeled, taking into account both switch and wire delays. A simple closed-form expression for contention in buffered, direct networks is derived and found to agree closely with simulations. The model includes the effects of packet size and communication locality. Network analysis under various constraints and under different workload parameters reveals that performance is highly sensitive to these constraints and workloads. A two-dimensional network is shown to have the lowest latency only when switch delays and network contention are ignored; three- or four-dimensional networks are favored otherwise. If communication locality exists, two-dimensional networks regain their advantage. Communication locality decreases both the base network latency and the network bandwidth requirements of applications. It is shown that a much larger fraction of the resulting performance improvement arises from the reduction in bandwidth requirements than from the decrease in latency     

Topological properties of the Extended OTIS-n-Cube interconnection network

We have recently introduced the Extended OTIS-n-Cube to overcome the weakness of some limitations found in the well-known OTIS-n-Cube such as the degree and the diameter. This paper investigates the topological properties of the new interconnection network by proposing an extensive study on some attractive topological properties of the extended OTIS-n-Cube interconnection network. Inspired by the attractive features of the new network, such as regular degree, small diameter, and semantic structure, we present a theoretical study on some topological properties of the Extended OTIS-n-Cube including routing paths and embedded cycles. Furthermore, the paper presents a performance evaluation on the topology by comparing it with the OTIS-n-Cube. Results prove the superiority of the new topology especially in minimizing routing distances.     

两类重要网络的传输延迟分析

提出了网络平均距离参数概念,用以度量网络的整体传输性能。与平均距离μ不同,网络平均距离μ′具有较强的网络应用背景。针对叉立方体网络的结构特性,给出了在交叉立方体网络中确定任意两个顶点之间最短路的长度和最短路条数的算法。从最短路、直径、平均距离、网络平均距离方面综合分析比较了超立方体网络和交叉立方体网络的信息传输延迟性能。     

Communication aspects of the star graph interconnection network

Basic communication algorithms for star graph interconnection networks are developed by using the hierarchical properties of the star graph, with the assumption that one input channel can drive only one output communication channel at a time. With this constraint, communication algorithms for each node can be expressed only as sequences of generators corresponding to the communication channels. Sequences that are identical exploit the symmetry and hierarchical properties of the star graph and can be easily integrated in communication hardware. Their time complexities are evaluated and compared with the corresponding results for the hypercube     

A fast pessimistic diagnosis algorithm for generalized hypercube multicomputer systems

The reliability of processors is an important issue for designing a massively parallel processing system for which fault-tolerant computing is crucial. In order to achieve high system reliability and availability, a faulty processor (node) when found should be replaced by a fault-free processor. Within a multiprocessor system, the technique of identifying faulty nodes by constructing tests on the nodes and interpreting the test outcomes is known as system-level diagnosis. The topological structure of a multicomputer system can be modeled by a graph of which the vertices and edges correspond to nodes and links of the system, respectively. This work presents a system-level diagnosis algorithm for a generalized hypercube which is an attractive variance of a hypercube. The proposed algorithm is based on the PMC model and can isolate all faulty nodes to within a set which contains at most one fault-free node. If the total number of nodes to be diagnosed in a generalized hypercube is N, the proposed algorithm can run in O(Nlog?N) time, and being superior to Yang??s algorithm proposed in 2004, it can diagnose not only a hypercube but also a generalized hypercube.     

A Reliable and Fault-Tolerant Interconnection Network

An interconnection network with multistage redundant paths is introduced for using in high-performance multiprocessor systems.The routing algorithm of the proposed network is simple and dynamically reroutable.The analysis of the fault-tolerance and performance of the network are given.It is shown that the probability of acceptance and the performance-to-cost ration of the network are better than those of F and Gamma Networks.Another advantages of the proposed network is the smaller amount of interstage links compared with F network.     

Optimal hypercube simulation on the partitioned optical passive stars network

Partitioned Optimal Passive Stars network, POPS(d,g), is an optical interconnection network of N processors (N=dg) with g ² optical passive star couplers. In this network, there are g groups of d processors each and the g ² couplers are used for connecting each group with each of the groups, including itself. In this paper, we present a technique for optimally simulating a frequently arising hypercube communication pattern on this network for all combinations of values of d and g. Specifically, we show that one-hop movements on the hypercube along the same dimension can be simulated on the POPS(d,g) network in slots for d≠g and in 2 slots for d=g.