Results 1  10
of
7,466
On kary ncubes: theory and applications
, 2003
"... Many parallel processing applications have communication patterns that can be viewed as graphs called kary ncubes, whose special cases include rings, hypercubes and tori. In this paper, combinatorial properties of kary ncubes are examined. In particular, the problem of characterizing the subgrap ..."
Abstract

Cited by 3 (0 self)
 Add to MetaCart
Many parallel processing applications have communication patterns that can be viewed as graphs called kary ncubes, whose special cases include rings, hypercubes and tori. In this paper, combinatorial properties of kary ncubes are examined. In particular, the problem of characterizing
Immucube: Scalable Fault Tolerant Routing for kary ncube Networks
 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
, 2007
"... This work presents Immucube, a scalable and efficient mechanism to improve dependability of interconnection networks for parallel and distributed computers. Immucube achieves better flexibility and scalability than any other previous faulttolerant mechanism in kary ncubes. The proposal inherits ..."
Abstract

Cited by 10 (1 self)
 Add to MetaCart
This work presents Immucube, a scalable and efficient mechanism to improve dependability of interconnection networks for parallel and distributed computers. Immucube achieves better flexibility and scalability than any other previous faulttolerant mechanism in kary ncubes. The proposal inherits
Contiguous And NonContiguous Processor Allocation Algorithms For KAry nCubes
 IEEE Transactions on Parallel and Distributed Systems
, 1995
"... Efficient utilization of processing resources in a large, multiuser parallel computer depends on processor allocation algorithms that minimize system fragmentation. We propose three processor allocation algorithms for the kary ncube class of parallel architectures, which includes the hypercube and ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
Efficient utilization of processing resources in a large, multiuser parallel computer depends on processor allocation algorithms that minimize system fragmentation. We propose three processor allocation algorithms for the kary ncube class of parallel architectures, which includes the hypercube
PANCYCLICITY IN FAULTY KARY 2CUBES
"... We prove that a kary 2cube Qk2 with 3 faulty edges but where every vertex is incident with at least 2 healthy edges is bipancyclic, if k ≥ 3, and kpancyclic, if k ≥ 5 is odd (these results are optimal). ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
We prove that a kary 2cube Qk2 with 3 faulty edges but where every vertex is incident with at least 2 healthy edges is bipancyclic, if k ≥ 3, and kpancyclic, if k ≥ 5 is odd (these results are optimal).
A Partial IrregularNetwork Routing on Faulty kary ncubes
"... Interconnection networks have been studied to connect a number of processing elements on parallel computers. Their design increasingly includes a challenge to high faulttolerance, as entire systems become complicated. This paper presents a partial irregularnetwork routing in order to provide a high ..."
Abstract
 Add to MetaCart
high faulttolerance in kary ncube networks. Since an irregularnetwork routing usually performs poorly in kary ncube networks, it is only used for progressive deadlockrecovery, and avoiding hard failures. The network is logically divided into the fault and regular regions. In the regular region
On the Performance Merits of Bypass Channels in Hypermeshes and kAry nCubes
, 1999
"... This paper evaluates the relative performance merits of total and partial bypassing provided by the DCSH and the express cube respectively, taking into account channel bandwidth constraints imposed by implementation technology. While Dally [10] has considered constraints for VLSI technology only, th ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
This paper evaluates the relative performance merits of total and partial bypassing provided by the DCSH and the express cube respectively, taking into account channel bandwidth constraints imposed by implementation technology. While Dally [10] has considered constraints for VLSI technology only, this study considers both VLSI and multiplechip technologies. Furthermore, whereas Dally has used the average message distance as a performance measure, which does not truly reflect network behaviour in the presence of message blocking, the present analysis develops queueing models to compute message latency in the DCSH and express cube under message blocking. The models have two important features: first, they make realistic assumptions, ignored in previous studies [1, 7, 10], such as including effects due to switching delays and the use of pipelinedbit transmission to lower the effects of long wires; second, they take into account the multiplexing effects of multiple virtual channels [18] onto individual physical channels, yielding more accurate results. The comparative analysis reveals that in most cases considered here the total bypassing strategy inherent in the DCSH topology provides better performance than the partial bypassing strategy used in express cubes. Furthermore, the DCSH can employ slower SEs than the express cube and still maintains superior performance levels
Keywords: kary ntree topology Traffic characterization Simulation
"... mance of these thintrees, and compare it with that of regular trees. Evaluation is carried sed o dwid e low ulers function this way: there are a few that are topologyaware and schedule applications in consecutive partitions of the network, thus allowing for an effective exploitation of locality. I ..."
Abstract
 Add to MetaCart
mance of these thintrees, and compare it with that of regular trees. Evaluation is carried sed o dwid e low ulers function this way: there are a few that are topologyaware and schedule applications in consecutive partitions of the network, thus allowing for an effective exploitation of locality
Embedding of Rings in 2D Meshes and Tori with Faulty Nodes
 Journal of Systems Architecture
, 1997
"... To increase the reliability of networks embedded in meshes or tori and to allow their use in the presence of faulty nodes, efficient faulttolerant schemes in meshes and tori are necessary. In this paper, we present algorithms for embedding a ring in a meshtype and torustype multiprocessor network ..."
Abstract

Cited by 3 (0 self)
 Add to MetaCart
To increase the reliability of networks embedded in meshes or tori and to allow their use in the presence of faulty nodes, efficient faulttolerant schemes in meshes and tori are necessary. In this paper, we present algorithms for embedding a ring in a meshtype and torustype multiprocessor
A Meshing Technique for de Bruijn Tori
 Contemp. Math
, 1994
"... An (R; S; m;n) k de Bruijn torus is a kary RS toroidal array with the property that every kary m n matrix appears exactly once contiguously on the torus. The torus is a generalization of de Bruijn cycles and has been extended to higher dimensions by many authors. The central question, asked by C ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
An (R; S; m;n) k de Bruijn torus is a kary RS toroidal array with the property that every kary m n matrix appears exactly once contiguously on the torus. The torus is a generalization of de Bruijn cycles and has been extended to higher dimensions by many authors. The central question, asked
Minimal Adaptive Routing with Limited Injection on Toroidal kary ncubes
 In Supercomputing 96
, 1996
"... Virtual channels can be used to implement deadlock free adaptive routing algorithms and increase network throughput. Unfortunately, they introduce asymmetries in the use of buffers of symmetric networks as the toroidal kary ncubes. ..."
Abstract

Cited by 8 (5 self)
 Add to MetaCart
Virtual channels can be used to implement deadlock free adaptive routing algorithms and increase network throughput. Unfortunately, they introduce asymmetries in the use of buffers of symmetric networks as the toroidal kary ncubes.
Results 1  10
of
7,466