International Journal of High Speed Computing

ATM networks are high speed networks and the volume of information carried by them at any instant of time is enormous. Since this information is likely to be of high commercial value, the effect of a transmission link or node failure would be catastrophic. Therefore ATM networks need to be equipped with mechanisms by which the failure of a network entity can be detected and service restored by changing over to alternate entities. Restoration mechanisms based on the self healing concept have been successful in achieving restoration times well below the two seconds limit to prevent any interruption of service to the users. This paper studies two self healing protocols for ATM networks. One is the Backup Virtual Path Self Healing protocol, and the second is the Broadcast Self Healing protocol, which is an extension of the protocol proposed for STM networks. Simulation studies were carried out to compare their restoration speeds, spare resource requirements and maximum restoration ratio. The results suggest that the backup virtual path protocol has an edge over the broadcast protocol.

In [6], we distributed training instances over a single-channel broadcast communication model to speed up execution of the back-propagation learning algorithm for classification problems. In this paper, we extend this concept to control problems, where the output is not necessarily 0 or 1, but ranges over an interval. We first propose a modified back-propagation learning algorithm that incrementally decreases the error threshold by half in order to process training instances with large weight changes as quickly as possible. This modified back-propagation learning algorithm is then parallelized using the single-channel broadcast communication model to n processors, where n is the number of training instances. Finally, the parallel back-propagation learning algorithm is modified for execution on a bounded number of processors to cope with real-world conditions.

The importance of data compression has been increasing of late due to the increasing demand put on the networks by applications consuming high amount of network bandwidth. Thus compressing data to at least a particular level becomes mandatory for applications to avoid congestion in the network. Recently there has been an interest in increasing the capacity of storage systems, which employ lossless data compression.

In this paper, we have proposed two improved algorithms for data compression. These algorithms are tested and compared with best known existing algorithms of data compression. It has been observed that compression rate of our proposed algorithms reaches up to 99% for huge files like swap files.

In this paper we consider five types of parallel preconditioners for solving large sparse nonsymmetric linear systems on the CRAY-T3E. They are ILU(0) in the wavefront ordering, ILU(0) in the multi-coloring ordering, SSOR in the wavefront ordering, the SPAI(SParse Approximate Inverse) preconditioner, and finally Multi-color Block SOR preconditioner. The ILU(0) is known to be robust and the wavefront ordering naturally exploits the parallelism but has a limited speedup due to the nonuniform lengths of the wavefronts. Multi-coloring is an efficient way of introducing the parallelism of order(N), where N is the order of the matrix but the convergence rate often deteriorates. The SPAI type preconditioner is inherently parallel and is gaining popularity. Finally, for the 5-point Laplacian matrix SOR method is known to have a nondeteriorating rate of convergence when the multi-coloring order is adopted. Also, Block SOR is expected to incur less communication overheads in a message-passing machine. Hence, Multi-Color Block SOR method is expected to have a good performance. Experiments were conducted for the Finite Difference discretizations of two problems with various meshsizes varying up to 1024×1024. MPI library was used for interprocess communications. The results show that ILU(0) in the multi-coloring ordering gives the best performance.

In this paper, we evaluate a variety of combinations of a load value predictor and a load address predictor. We consider a dynamic hybrid predictor using a predictor selection counter, a static hybrid predictor utilizing execution profiles, and a cooperative predictor. The cooperative predictor is a load value predictor supported by a load address predictor when it is unable to predict a load value. The static hybrid and the cooperative predictors have a benefit that the hardware cost of the selection counter is removed. On the other hand, the dynamic hybrid and the cooperative predictors are free from tedious process of profiling. Based on cycle-by-cycle simulations, we have evaluated the variations and found that the cooperative predictor exploits instruction level parallelism most effectively.

The Incrementally Extensible Hypercube (IEH) is a novel interconnection network derived from the hypercube. Unlike the hypercube, the IEH graph is incrementally extensible, that is, it can be constructed for any number of nodes. In addition, it has optimal fault tolerance and its diameter is logarithmic in the number of nodes and the difference of the maximum and the minimum degree of a node in the graph is (i.e., the graph is almost regular). In this paper, we show that almost the entire IEH graph, except for those with N =2ⁿ-1 nodes for all , has a Hamiltonian cycle; if an IEH graph has N=2ⁿ-1 nodes then it has only a Hamiltonian path, not cycle. These results enable us to obtain the good embedding of rings and linear arrays into the IEH graph.