Kurt J. Windisch

Learn More
Current processor allocation techniques for highly parallel systems are typically restricted to contiguous allocation strategies for which performance suuers signiicantly due to the inherent problem of fragmentation. As a result, message passing systems have yet to achieve the high utilization levels exhibited by traditional vector supercomputers. We are(More)
Current processor allocation techniques for highly parallel systems are typically restricted to contiguous allocation strategies for which performance suffers significantly due to the inherent problem of fragmentation. As a result, message-passing systems have yet to achieve the high utilization levels exhibited by traditional vector supercomputers. We are(More)
Two basic approaches are taken when modeling workloads in simulation-based performance evaluation of parallel job scheduling algorithms: (1) a carefully reconstructed trace from a real supercomputer can provide a very realistic job stream, or (2) a exible synthetic model that attempts to capture the behavior of observed workloads can be devised. Both(More)
Current processor allocation techniques for highly parallel systems have thus far been restricted to contiguous allocation strategies for which performance suffers significantly due to the inherent problem of fragmentation. We are investigating processor allocation algorithms which lift the restriction on contiguity of processors in order to address the(More)
ProcSimity is a software tool that supports research in processor allocation and scheduling for highly parallel systems. ProcSimity's multicomputer simulator supports experimentation with selected allocation and scheduling algorithms on architectures with a range of network topologies and for several current routing and ow control mechanisms.(More)
We investigate the applicability of spanners as shared structures that o er both low cost and low delay for broadcast and multicast. A k-spanner has the potential to o er lower delay than shared trees because it limits the distance between any two nodes in the network to a multiplicative factor k of the shortest-path distance. Using simulation over random(More)
EEcient 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 k-ary n-cube class of parallel architectures, which includes the hypercube and multidimensional torus. The k-ary Partner strategy is a(More)
The analysis of workload traces from real production parallel machines can aid a wide variety of parallel processing research, providing a realistic basis for experimentation in the management of resources over an entire workload. We analyze a ve-month workload trace of an Intel Paragon machine supporting a production parallel workload at the San Diego(More)
  • 1