An increasing fraction of dynamic power consumption can be attributed to switched interconnect capacitances. Non-uniform wire spacing depending on activity had shown promising power reductions for on-chip buses. In this paper, a new and fast routing optimization methodology based on non-uniform spacing is proposed for entire circuits. No area investment is required, since whitespace remaining after detailed routing is exploited. The proposed methodology has been implemented and tapped into an industry-proven design flow. Wire power reductions of up to 9.55% for modern multiprocessor benchmarks with tight area constraints are demonstrated, twice as much as approaches that do not take switching activities into account. Timing is not adversely affected, and the yield limit is slightly improved.

One of the main features of ABR services is the employment of a rate-based flow control mechanism. Feedback from the network switches to the end systems gives users the information necessary to adjust their transmission rates appropriately according to the current network load. This type of control has been investigated to some extent, assuming a constant transmission capacity on a bottleneck link. In this paper we develop a discrete-time analysis to study the effect of bottleneck service rate variations on the performance of the ABR flow control mechanism. Looking at real systems, such variations occur on the one hand due to the establishment and release of ATM connections and on the other hand due to the varying bandwidth demand of VBR connections. To model the rate variations, the bottleneck service rate is assumed to follow a general probability density function.

In ATM networks congestion control plays an essential role in supporting different quality of service guarantees for a large variety of traffic types. While for real-time communications, such as CBR and VBR services, a preventive open-loop congestion control mechanism is applied, a reactive closed-loop mechanism has been suggested for the ABR service class, which was introduced aiming at data communication applications. This paper presents a performance analysis of the rate-based congestion control mechanism developed by the ATM Forum. Using a differential equation approach for the computation of the evolution of the allowed cell rate and the buffer content during the steady state, we derive closed-form expressions to estimate the maximum buffer lengths. These expressions can be formulated for steady-state conditions as well as for transient phases. In our approach, scenarios with and without priority for the resource management cells are considered. From numerical examples we conclude that an appropriate setting of the control parameters is important achieving a reasonable performance of the rate-based control mechanism. In particular, we are interested in limiting the required buffer size while enabling a fast reaction to changes of the available transmission capacity.

In ATM networks the usage parameter control function must be defined very carefully to avoid network congestion due to users violating their negotiated traffic contract intentionally or unintentionally. This control function operates on the basis of connection traffic descriptors and is located at the network access. Connection traffic descriptors, which describe the characteristics of a source, are parameters such as the peak cell rate and the cell delay variation tolerance as well as the sustainable cell rate and the burst tolerance. To monitor these cell rates, the generic cell rate algorithm was proposed. Since a user can specify the peak and the sustainable cell rate together, different structural monitoring schemes exist for the enforcement of these rates. The authors present exact analyses of four monitoring schemes. Numerical results show that the structure of the scheme has only a minor influence on the resulting cell rejection probability. However, some general statements about the performance and the practical applicability of the schemes are given.