Malgorzata Marek-Sadowska

Crosstalk reduction for VLSI

The performance of high-speed electronic systems is limited by interconnect-related failure modes such as coupled noise. We propose new techniques for alleviating the problems caused by coupling between signal lines on integrated circuits. We show that models used by previous work on coupled noise-constrained layout synthesis do not allow the use of several important degrees of freedom. These degrees of freedom include the ability to utilize dynamic noise margins rather than static noise margins, the dependence of coupled noise on drive strength, and the possibility of using overlaps to reduce susceptibility to noise. We derive an expression for the coupled noise integral and a bound for the peak coupled noise voltage which shows order of magnitude improvements in both accuracy and fidelity compared to the charge sharing model used in previous work. We use the new bounds to guide a greedy channel router, which manipulates exact adjacency information at every stage, allowing it to introduce jogs or doglegs when necessary for coupled noise reduction. Experimental results indicate that our algorithm compares favorably to previous work. The coupled noise is significantly reduced on benchmark instances.

On designing via-configurable cell blocks for regular fabrics

In this paper we describe the design process of a via-configurable block for regular fabrics. The block consists of via-configurable functional cells, via-decomposable flip-flops, and via-configured sizable repeaters. The fabric has fixed layers up to M2. An M1-M2 via mask is used to define the blocks functionality. The upper-level metals are customized. Compared to other structures based on LUTs or PLAs, and fixed flip-flops, our block has much smaller area, higher performance and lower power consumption.

Via-Configurable Routing Architectures and Fast Design Mappability Estimation for Regular Fabrics

In this paper, we describe a new via-configurable routing architecture which shows much better throughput and performance than the previous structures. We demonstrate how to construct a single-via-mask fabric to reduce further the mask cost, and we analyze the penalties which it incurs. To solve the routability problem commonly existing in fabric-based designs, an efficient white-space allocation scheme is suggested, which provides a fast design convergence and early prediction of the circuit mappability to a given fabric.

An integrated design flow for a via-configurable gate array

In This work we present a complete physical design flow for a via-configurable gate array (VCGA). The VCGA is an array of prefabricated logic blocks and fixed metal masks. The block consists of via-configurable functional cells and a via-decomposable flip-flop. An M1-M2 via mask is used to define the blocks functionality. Interconnects are customized using via masks. We developed a physical design flow for VCGA, which integrates a set of effective techniques. Here, we highlight the packing, cell-binding, and detailed-routing problems. We use our design flow to compare the VCGA-based and standard-cell/FPGA-based designs. Experimental results show the efficiency of our flow.

The magic of a via-configurable regular fabric

In this paper, we provide a comprehensive study of the mappability of a via-configurable gate array (VCGA). Although, the base cell of the VCGA is simple, by customizing only via masks it can implement various combinational logic functions, sequential elements, and SRAM cells. Our VCGA can be efficiently configured into SRAM arrays, adders and multipliers. The strong configurability of our VCGA allows us to minimize the number of fixed parts in a general-purpose VCGA fabric, which greatly improves area utilization.

Crosstalk in VLSI interconnections

We address the problem of crosstalk computation and reduction using circuit and layout techniques in this paper. We provide easily computable expressions for crosstalk amplitude and pulse width in resistive, capacitively coupled lines. The expressions hold for nets with arbitrary number of pins and of arbitrary topology under any specified input excitation. Experimental results show that the average error is about 10% and the maximum error is less than 20%. The expressions are used to motivate circuit techniques, such as transistor sizing, and layout techniques, such as wire ordering and wire width optimization to reduce crosstalk.

Benefits and costs of power-gating technique

Power-gating is a technique for saving leakage power by shutting off the idle blocks. However, without good understanding and careful design, negative effects of power gating may overwhelm the potential gain and may make the technique not worth the effort. In this paper, we report on our study of the benefits and costs of the power-gating technique in terms of power, area, and performance. We model and analyze several strongly related parameters such as sleep-transistor size, decap area, and supply voltage level. We also report on our experiments to demonstrate how the gated area, circuit behavior and power mesh granularity affect the power gating technique at the system level. Experimental results show that, by compromising 4% of the total area and 5% of the dynamic power, we can achieve 47% leakage power saving while maintaining the same performance. With technology scaling down, the saving is significant. We conclude that we can benefit from the power-gating technique in future technology nodes.

Perturb and simplify: multi-level boolean network optimizer

In this paper, we discuss the problem of optimizing a multi-level logic combinational Boolean network. Our techniques apply a sequence of local perturbations and modifications of the network which are guided by the automatic test pattern generation ATPG based reasoning. In particular, we propose several new ways in which one or more redundant gates or wires can be added to a network. We show how to identify gates which are good candidates for local functionality change. Furthermore, we discuss the problem of adding and removing two wires, none of which alone is redundant, but when jointly added/removed they do not affect functionality of the network. We also address the problem of efficient redundancy computation which allows to eliminate many unnecessary redundancy tests. We have performed experiments on MCNC benchmarks and compared the results to those of misII and RAMBO. Experimental results are very encouraging.

Clock skew optimization for ground bounce control

High speed synchronous digital systems require large switching currents to facilitate rapid signal transitions. These large currents create voltage drops on the power distribution network and necessitate expensive chip packaging with a large number of supply pins. In this paper we propose a novel technique to reduce the dynamic transient current drawn from the supply pins. Our approach is based on sub-dividing the synchronous clocking into multiple sub-clocks with relative skew. This spreads the computation across the entire clock cycle instead of largely occurring at the beginning. Timing constraints must also be obeyed, so that no races or timing errors are introduced. We propose an exact algorithm based on integer linear programming to solve this problem. We have used our method in the design of a 5 GHz ECL encoder chip to achieve a factor of two reduction in ground bounce, as shown by HSPICE simulations. We also obtained order-of-magnitude improvements in ground bounce on benchmarks laid our in submicron CMOS technology. The approach potentially leads to significant reductions in packaging costs.

Perturb and simplify: multilevel Boolean network optimizer

In this paper, we present logic optimization techniques for multilevel combinational networks. Our techniques apply a sequence of perturbations which result in simplification of the circuit. The perturbation and simplification is achieved through wires/gates addition and removal which are guided by the Automatic Test Pattern Generation (ATPG) based reasoning. The main operations of our approaches are incremental transformations of the circuit (such as adding wires/gates and changing gates functionality) to remove some particular wire, At each iteration, a summary information of such wires/gates addition and removal is precomputed first. Then, a transformation is chosen to remove several wires at once. We have performed experiments on MCNC benchmarks and compared the results to those of misII and RAMBO. Experimental results are very encouraging.