Lately, time-multiplexed I/Os for multi-device implementations (e.g., multi-FPGA systems), have come into practical use. They realize multiple I/O signal transmissions between two devices in one system clock cycle using one I/O wire between the devices and multiple I/O clock cycles. Though they ease the limitation of the number of I/O-pins of each device, the system clock period becomes much longer approximately in proprotion to the maximum number of multiplexed I/Os on a signal path. There is no conventional partitioning algorithm considering the effect of time-multiplexed I/Os directly. We introduce a new cost function for evaluating the suitability of a bipartition for multi-device implementations with time-multiplexed I/Os. We propose a performance-driven bipartitioning method VIOP which minimizes the value of the cost function. Our method VIOP combines three algorithms, such that i) min-cut partitioning, ii) coarse performance-driven partitioning, iii) fine performance-driven partitioning. For min-cut partitioning and coarse performance-driven partitioning, we employ a well-known conventional bipartitioning algorithms CLIP-FM and DUBA, respectively. For fine performance-driven partitioning for the final improvement of a partition, we propose a partitioning algorithm CAVP. By our method VIOP, the average cost was improved by 10.4% compared with the well-known algorithms.

This paper presents a novel approach to improving the IDDQ-based diagnosability of a CMOS circuit by dividing the circuit into independent partitions and using a separate power supply for each partition. This technique makes it possible to implement multiple IDDQ measurement points, resulting in improved IDDQ-based diagnosability. The paper formalizes the problem of partitioning a circuit for this purpose and proposes optimal and heuristic based solutions. The effectiveness of the proposed approach is demonstrated through experimental results.

This paper presents a timing-driven iterative improvement circuit partitioning algorithm under path delay constraints for the general delay model. The proposed algorithm is an extension of the Fiduccia & Mattheyses (FM) method so as to handle path delay constraints and consists of the clustering and iterative improvement phases. In the first phase, we reduce the size of a given circuit, with a new clustering algorithm to obtain a partition in a short computation time. Next, the iterative improvement phase based on the FM method is applied, and then a new path-based timing violation removal algorithm is also performed so as to remove all the timing violations. From experimental results for ISCAS89 benchmarks, we have demonstrated that the proposed algorithm can produce the partitions which mostly satisfy the timing constraints.

In this paper, we extend the circuit partitioning algorithm which we have proposed for multi-FPGA systems and present a new algorithm in which the delay of each critical signal path is within a specified upper bound imposed on it. The core of the presented algorithm is recursive bipartitioning of a circuit. The bipartitioning procedure consists of three stages: 0) detection of critical paths; 1) bipartitioning of a set of primary inputs and outputs; and 2) bipartitioning of a set of logic-blocks. In 0), the algorithm computes the lower bounds of delays for paths with path delay constraints and detects the critical paths based on the difference between the lower and upper bound dynamically in every bipartitioning procedure. The delays of the critical paths are reduced with higher priority. In 1), the algorithm attempts to assign the primary inputs and outputs on each critical path to one chip so that the critical path does not cross between chips. Finally in 2), the algorithm not only decreases the number of crossings between chips but also assigns the logic-blocks on each critical path to one chip by exploiting a network flow technique. The algorithm has been implemented and applied to MCNC PARTITIONING 93 benchmark circuits. The experimental results demonstrate that it resolves almost all path delay constraints with maintaining the maximum number of required I/O blocks per chip small compared with conventional alogorithms.

In circuit partitioning for FPGAs, partitioned signal nets are connected using I/O blocks, through which signals are coming from or going to external pins. However, the number of I/O blocks per chip is relatively small compared with the number of logic-blocks, which realize logic functions, accommodated in the FPGA chip. Because of the I/O block limitation, the size of a circuit implemented on each FPGA chip is usually small, which leads to a serious decrease of logic-block utilization. It is required to utilize unused logic-blocks in terms of reducing the number of I/O blocks and realize circuits on given FPGA chips. In this paper, we propose an algorithm which partitions an initial circuit into multi-FPGA chips. The algorithm is based on recursive bi-partitioning of a circuit. In each bi-partitioning, it searches a partitioning position of a circuit such that each of partitioned subcircuits is accommodated in each FPGA chip with making the number of signal nets between chips as small as possible. Such bi-partitioning is achieved by computing a minimum cut repeatedly applying a network flow technique, and replicating logic-blocks appropriately. Since a set of logic-blocks assigned to each chip is computed separately, logic-blocks to be replicated are naturally determined. This means that the algorithm makes good use of unused logic-blocks from the viewpoint of reducing the number of signal nets between chips, i.e. the number of required I/O blocks. The algorithm has been implemented and applied to MCNC PARTITIONING 93 benchmark circuits. The experimental results demonstrate that it decreases the maximum number of I/O blocks per chip by a maximum of 49% compared with conventional algorithms.

We developed a parallel bordered-block-diagonal (BBD) matrix solution for parallel circuit simulation. In parallel circuit sumulation on a MIMD parallel computer, a circuit is partitioned into as many subcircuits as the processors of a parallel computer. Circuit partition produce a BBD matrix. In parallel BBD matrix solution, diagonal blocks are easily solved separately in each processor. It is difficult, however, to solve the interconnection (IC) submatrix of a BBD matrix effectively in parallel. To make matters worse, the more a circuit is partitioned into subcircuits for highly parallel circuit simulation, the larger the size of an IC submatrix becomes. From an examination, we found that an IC submatrix is more dense (about 30% of all entries are non-zeros) than a normal circuit matrix, and the non-zeros per row in an IC submatrix are almost constant with the number of subcircuits. To attain high-speed circuit simulation, we devised a data structure for BBD matrix processing and an approach to parallel BBD matrix solution. Our approach solves the IC submatrix in a BBD matrix as well as the diagonal blocks in parallel using all processors. In this approach, we allocate an IC submatrix in block-wise order rather than in dot-wise order onto all processors. Thus, we balance the processor perfomance with the communication capacity of a parallel computer system. When we changed the block size of IC submatrix allocation from dot-wise order to 88 block-wise order, the 88 block-wise order allocation almost halved the matrix solution time. The parallel simulation of a sample circuit with 3277 transistors was 16.6 times faster than a single processor when we used 49 processors.

We have studied a circuit partitioning approach in the view of parallel circuit simulation on a MIMD parallel computer. In parallel circuit simulation, a circuit is partitioned into equally sized subcircuits while minimizing the number of interconnection nodes. Besides circuit partitioning time should be short enough compared with the total simulation time. From the details of circuit simulation time, we found that balancing subcircuits is critical for low parallel processing, whereas minimizing the interconnection nodes is critical for highly parallel processing. Our circuit partitioning approach consists of four steps： Grouping transistors, initial partitioning the transistor-groups, minimizing the number of interconnection nodes, and balancing the subcircuits. It is based on an algorithmic approach, and can directly control the tradeoffs between balancing subcircuits and minimizing the interconnection nodes by adjusting the parameters. We partitioned a test circuit with 3277 transistors into 4, 9, ... , 64 subcircuits, and did parallel simulations using PARACS, our parallel circuit simulator, on an AP1000 parallel computer. The circuit partitioning time was short enough-less than 3 percent of the total simulation time. The highest performance of parallel analysis using 49 processors was 16 times that of a single processor, and that for total simulation was 9 times.

This paper presents an efficient heuristic algorithm for min-cut bisection of weighted hypergraphs. The proposed algorithm is based on a heuristic algorithm proposed by Kahng, which was devised for non-weighted hypergraph bisection, adopting a non-weighted graph called intersection graph to represent a given hypergraph. In the proposed algorithm, instead of an intersection graph, a bipartite graph called netgraph is newly introduced to explicitly represent the weights of nodes of a hypergraph. Using the netgraph, it is easy to partition a weighted hypergraph into two hypergraphs with same size. Computation time of the proposed method is O(m2), where m is the number of nodes of a given hypergraph. Experimental results with real circuit data show that the proposed method produces better solutions in shorter computation time compared with existing methods.