Survey of commercial parallel machines

Ramanathan, Gowri; Oren, Joel

doi:10.1145/152835.152837

Cited by 15 publications

(4 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…N designing interconnection networks, a constant low I degree is generally more preferable than a high degree because of performance advantages, scalability, and various implementation considerations Ill, [91,[171, [201. Within the family of low-degree networks, degree-4 networks such as 2D meshes and 2D tori are among the most popular choices for processor interconnection in today's parallel computers 1141, [16], [20]. One practical problem with these networks, however, is that their diameter and average distance tend to be large as the number of nodes increases, and so there has been a continuous effort in finding degree-4 or lowdegree networks that have smaller diameters and average distances [2], [4], [6], [7], [ll], [12].…”

Section: Introductionmentioning

confidence: 99%

Optimal layouts of midimew networks

Lau¹,

Chen²

1996

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Midimew networks [4] are mesh-connected networks derived from a subset of degree-4 circulant graphs. They have minimum diameter and average distance among all degree-4 circulant graphs, and are better than some of the most common topologies for parallel computers in terms of various cost measures. Among the many midimew networks, the rectangular ones appear to be most suitable for practical implementation. Unfortunately, with the normal way of laying out these networks on a 2D plane, long cross wires that grow with the size of the network exist. In this paper, we propose ways to lay out rectangular midimew networks in a 2D grid so that the length of the longest wire is at most a small constant. We prove that these constants are optimal under the assumption that rows and columns are moved as a whole during the layout process

show abstract

Section: Introductionmentioning

confidence: 99%

Optimal layouts of midimew networks

Lau¹,

Chen²

1996

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

show abstract

“…Take a ring and a torus and strip them of all the end-round connections, we get a chain and a ring respectively. We limit our scope to these structures because they are the most popular choices of topologies in commercial parallel computers 10, 13,16]. Examples include the hypercubestructured Intel iPSC/860 and NCUBE/2, the meshstructured Intel Paragon, Intel Touchstone Delta, iWarp, and Ametek 2010.…”

Section: Introductionmentioning

confidence: 99%

The Generalized Dimension Exchange Method for Load Balancing in k-ary n-Cubes and Variants

Lau

1995

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

“…[1][2][3][4] Many topologies have been explored for parallel computers, including multistage interconnection networks such as omega, baseline, banyan, and crossover, and pointto-point interconnection networks such as hypercube, mesh, ring, bus, and star. 3,5 Currently, two of the most popular point-to-point topologies are the binary n-cube or hypercube [6][7][8][9] and the mesh. [10][11][12][13] The hypercube topology is completely symmetric with a logarithmic diameter 1the diameter of a network is defined to be the largest number of hops in the shortest path between any two nodes.…”

Section: Introductionmentioning

confidence: 99%

Optical binary de Bruijn networks for massively parallel computing: design methodology and feasibility study

Louri

Sung

1995

Appl. Opt.

View full text Add to dashboard Cite

The interconnection network structure can be the deciding and limiting factor in the cost and the performance of parallel computers. One of the most popular point-to-point interconnection networks for parallel computers today is the hypercube. The regularity, logarithmic diameter, symmetry, high connectivity, fault tolerance, simple routing, and reconfigurability (easy embedding of other network topologies) of the hypercube make it a very attractive choice for parallel computers. Unfortunately the hypercube possesses a major drawback, which is the complexity of its node structure: the number of links per node increases as the network grows in size. As an alternative to the hypercube, the binary de Bruijn (BdB) network has recently received much attention. The BdB not only provides a logarithmic diameter, fault tolerance, and simple routing but also requires fewer links than the hypercube for the same network size. Additionally, a major advantage of the BdB network is a constant node degree: the number of edges per node is independent of the network size. This makes it very desirable for large-scale parallel systems. However, because of its asymmetrical nature and global connectivity, it poses a major challenge for VLSI technology. Optics, owing to its three-dimensional and globalconnectivity nature, seems to be very suitable for implementing BdB networks. We present an implementation methodology for optical BdB networks. The distinctive feature of the proposed implementation methodology is partitionability of the network into a few primitive operations that can be implemented efficiently. We further show feasibility of the presented design methodology by proposing an optical implementation of the BdB network.

show abstract

Survey of commercial parallel machines

Cited by 15 publications

References 2 publications

Optimal layouts of midimew networks

Optimal layouts of midimew networks

The Generalized Dimension Exchange Method for Load Balancing in k-ary n-Cubes and Variants

Optical binary de Bruijn networks for massively parallel computing: design methodology and feasibility study

Contact Info

Product

Resources

About