Mustafa I. Ali

Multiobjective VLSI cell placement using distributed simulated evolution algorithm

Simulated evolution (SimE) is a sound stochastic approximation algorithm based on the principles of adaptation. If properly engineered it is possible for SimE to reach near-optimal solutions in less time than simulated annealing. Nevertheless, depending on the size of the problem, it may have large run-time requirements. One practical approach to speed up the execution of the SimE algorithm is to parallelize it. This is all the more true for multi-objective cell placement, where the need to optimize conflicting objectives (interconnect wirelength, power dissipation, and timing performance) adds another level of difficulty. In this paper a distributed parallel SimE algorithm is presented for multiobjective VLSI standard cell placement. Fuzzy logic is used to integrate the costs of these objectives. The algorithm presented is based on random distribution of rows to individual processors in order to partition the problem and distribute computationally intensive tasks, while also efficiently traversing the complex search space. A series of experiments are performed on ISCAS-85/89 benchmarks to compare speedup with serial implementation and other earlier proposals. Discussion on comparison with parallel implementations of other iterative heuristics is included.

Reconfigurable broadcast scan compression using relaxation-based test vector decomposition

An effective reconfigurable broadcast scan compression scheme that employs partitioning of test sets and relaxation-based decomposition of test vectors is proposed. Given a constraint on the number of tester channels, the technique classifies test sets into acceptable and bottleneck vectors. The bottleneck vectors are then decomposed into a set of vectors that meets the given constraint. The acceptable and decomposed test vectors are partitioned into the smallest number of partitions while satisfying the tester channels constraint to reduce the decompressor area. Thus, the technique by construction satisfies a given tester channel constraint at the expense of an increased test vector count and number of partitions, offering a tradeoff between test compression, the test application time and the area of test decompression circuitry. Experimental results demonstrate that the proposed technique achieves better compression ratios compared with other techniques of test compression.

Evaluating Parallel Simulated Evolution Strategies for VLSI Cell Placement

Simulated Evolution (SimE) is an evolutionary metaheuristic that has produced results comparable to well established stochastic heuristics such as SA, TS and GA, with shorter runtimes. However, for optimization problems with a very large set of elements, such as in VLSI cell placement and routing, runtimes can still be very large and parallelization is an attractive option for reducing runtimes. Compared to other metaheuristics, parallelization of SimE has not been extensively explored. This paper presents a comprehensive set of parallelization approaches for SimE when applied to multiobjective VLSI cell placement problem. Each of these approaches are evaluated with respect to SimE characteristics and the constraints imposed by the problem instance. Conclusions drawn can be extended to parallelization of SimE when applied to other optimization problems.

A Reconfigurable Broadcast Scan Compression Scheme Using Relaxation Based Test Vector Decompos

An effective reconfigurable broadcast scan compression scheme that employs test set partitioning and relaxation-based test vector decomposition is proposed. Given a constraint on the number of tester channels, the technique classifies the test set into acceptable and bottleneck vectors. The bottleneck vectors are then decomposed into a set of vectors that meet the given constraint. The acceptable and decomposed test vectors are partitioned into the smallest number of partitions while satisfying the tester channels constraint to reduce the decompressor area. Thus, the technique by construction satisfies a given tester channels constraint at the expense of increased test vector count and number of partitions, offering a tradeoff between test compression, test application time and test decompression circuitry area. Experimental results demonstrate that the proposed technique achieves better compression ratios compared to other test compression techniques.

Evaluating parallel simulated evolution strategies for VLSI cell placement

Simulated evolution (SimE) is an evolutionary metaheuristic that has produced results comparable to well established stochastic heuristics such as SA, TS and GA, with shorter runtimes. However, for problems with a very large set of elements to optimize, such as in VLSI placement and routing, runtimes can still be very large and parallelization is an attractive option. Compared to other metaheuristics, parallelization of SimE has not been extensively explored. This paper presents a comprehensive set of parallelization approaches for SimE when applied to multiobjective VLSI cell placement problem. Each of these approaches are evaluated with respect to SimE characteristics and the constraints imposed by the problem instance. Conclusions drawn can be extended to parallelization of other SimE based optimization problems

Comparative evaluation of parallelization strategies for evolutionary and stochastic heuristics

In this paper we present an evaluation of selected parallel strategies for Simulated Annealing and Simulated Evolution, identifying the impact of various issues on the effectiveness of parallelization. Issues under consideration are the characteristics of these algorithms, the problem instance, and the implementation environment. Observations are presented regarding the impact of parallel strategies on runtime and achievable solution quality. Effective parallel algorithm design choices are identified, along with pitfalls to avoid. We further attempt to generalize our assessments to other heuristics.

Parallel Strategies for Stochastic Evolution

The paper discusses the parallelization of stochastic evolution metaheuristic, identifying effective parallelization for a distributed parallel environment. Multiobjective VLSI cell placement is used as an optimization problem. A comprehensive set of parallelization approaches are tested and an effective strategy is identified in terms of two underlying factors: workload division and the effect of parallelization on metaheuristics search intelligence. The strategies are compared with parallelization of another similar evolutionary metaheuristic called simulated evolution. The role of the two mentioned underlying factors is discussed in parallelization of stochastic evolution, the parallelized version of which has not been presented before.