PERFORMANCE ANALYSIS OF WORMHOLE ROUTED K-Ary N-TREES

Abstract

The past few years have seen a rise in popularity of massively parallel architectures that use fat-trees as their interconnection networks. In this paper we formalize a parametric family of fat-trees, the fc-ary n-trees, built with constant arity switches interconnected in a regular topology. A simple adaptive routing algorithm for fc-ary n-trees sends each message to one of the nearest common ancestors of both source and destination, choosing the less loaded physical channels, and then reaches the destination following the unique available path. Through simulation on a 4-ary 4-tree with 256 nodes, we analyze some variants of the adaptive algorithm that utilize wormhole routing with 1, 2 and 4 virtual channels. The experimental results show that the uniform, bit reversal and transpose traffic patterns are very sensitive to the flow control strategy. In, all these cases, the saturation points are between 35 -40% of the network capacity with 1 virtual channel, 55-60% with 2 virtual channels and around 75% with 4 virtual channels. The complement traffic, a representative of the class of the congestion-free communication patterns, reaches an optimal performance, with a saturation point at 97% of the capacity for all flow control strategies. In this case virtual channels are of little help and the average network latency experiences an increase proportional to the number of virtual channels, due to the multiplexing of several packets onto the same physical links.

Dynamic power saving in fat-tree interconnection networks using on/off links

“…This represents 72% of the computers in the list (one year before, only 58.8% of the systems in the list were clusters). A significant fraction of these clusters use fat-tree interconnection networks due to its high bisection bandwidth and ease of application mapping for arbitrary communication topologies [9]. But most applications have communication topology requirements that are far less than the total connectivity provided by fat-trees.…”

An ROBDD-Based Combinatorial Method for the Evaluation of Yield of Defect-Tolerant Systems-on-Chip

“…2 for the case , . The system includes IPs IPA and IPs IPB interconnected by a -ary, -tree fat-tree [30]. The system is func- tioning if unfailed IPAs can communicate through the fat-tree with unfailed IPBs.…”

The Quadrics network: high-performance clustering technology

“…9 A quaternary fat tree of dimension n is composed of 4 n processing nodes and n × 4 n−1 switches interconnected as a delta network; it can be recursively built by connecting four quaternary fat trees of dimension n − 1. Figure 2 shows quaternary fat trees of dimensions 1, 2, and 3.…”