KLAT2's Flat Neighborhood Network

نویسندگان

  • Henry G. Dietz
  • Timothy Mattox
چکیده

KLAT2, Kentucky Linux Athlon Testbed 2, is a cluster of 64 (plus two ‘‘hot spare’’) 700MHz AMD Athlon PCs. The raw compute speed of the processors justifies calling the system a supercomputer, but these fast nodes must be mated with a high-performance network in order to achieve the balance needed to obtain speed-up on real applications. Usually, cluster networks are built by combining the fastest available NICs and switching fabric, making the network expensive. Instead, KLAT2 uses a novel ‘‘Flat Neighborhood’’ network topology that was designed by a genetic algorithm (GA). A total of about $8,100 worth of 100Mb/s Fast Ethernet NICs, switches, and Cat5 cable, allows KLAT2’s network to deliver both single-switch latency for any point-to-point communication and up to 25.6Gb/s bisection bandwidth. This paper describes how this new network architecture was derived, how it is used, and how it performs.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Compiler Techniques for Flat Neighborhood Networks

A Flat Neighborhood Network (FNN) is a new interconnection network architecture that can provide very low latency and high bisection bandwidth at a minimal cost for large clusters. However, unlike more traditional designs, FNNs generally are not symmetric. Thus, although an FNN by definition offers a certain base level of performance for random communication patterns, both the network design an...

متن کامل

Analysis of Cluster Interconnection Network Topologies

Cluster computing provides an economical alternative for high performance computing, that in the past could only be provided by expensive parallel supercomputers. Clusters are built with standard components and interconnected by various interconnection topologies. These interconnection topologies provide different approaches for communication between processing nodes within the cluster. A study...

متن کامل

A Proposed Topology for a 192-Processor Symmetric Cluster with a Single-Switch Delay

Cluster computing is becoming an accepted form of supercomputing in the academe, government and industries around the world. The Ateneo High Performance Computing Group (AHPCG) proposes a design for building a 192processor cluster using a symmetric and flat networking neighborhood. A routing schema will also be developed to maximize this network topology. This design aims to demonstrate this ne...

متن کامل

Theorem B . 2 Vector bundles on a neighborhood of a Hirze - bruch surface

Let S be a ruled surface inside a smooth 3-fold W. Choose a neighborhood V of S in W and let Ŝ be the formal completion of S in V. Let E be a vector bundle over Ŝ. We show that (under suitable conditions) the local deformation space of E is finite dimensional and smooth. Moreover, we show that E is a flat limit of a flat family of vector bundles whose general element we describe explicitly.

متن کامل

Vector bundles on a neighborhood of a ruled surface in a threefold

Let S be a ruled surface inside a smooth 3-fold W. Choose a neighborhood V of S in W and let Ŝ be the formal completion of S in V. Let E be a vector bundle over Ŝ. We show that (under suitable conditions) the local deformation space of E is finite dimensional and smooth. Moreover, we show that E is a flat limit of a flat family of vector bundles whose general element we describe explicitly.

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2000