Parallel Optoelectronic Computing System 2.parallel Optoelectronic Computing System with Reconngurable Optical Interconnection
نویسنده
چکیده
It is expected that high performance parallel computing systems can be realized by using ne grain electronic PE arrays and two-dimensional optical devices such as surface emitting laser diode arrays and photo detector arrays. Although processing architectures for utilizing such devices are intrinsically based on massively parallel processing with parallel optical input and output, the interconnects can not be implemented by using conventional macro-scale wiring technology. In such case, optical interconnection palys an important role to realize high preformance computing for some application elds such as high speed visual image processing, robot vision, automated visual inspection, and pattern recognition. In this paper, the design concept of the system, connguration of an experimental system, and algorithms for the parallel optoelectronic computing system are shown. Keys of the system design are how generality of processing is kept and how the optical interconnection is implemented. Ishikawa et al. have proposed a fully parallel hierarchical processing architecture with a reconngurable optical interconnection and a compact PE architecture for general purpose processing l]. In order to directly connect between arbitrary PEs, Kirk et al. have designed a reconngurable optical interconnection and showed basic experimental results 2] using a phase modulation type spatial light modulator, PAL-SLM (Parallel Aligned Spatial Light Modulator) developed by Hamamatsu K.K..3]. The interconnection subsystem is based on computer generated hologram(CGH) optimized by using simulated annealing algorithm. It realizes shift invariant interconnection for SIMD type parallel processing. Reconngurability of the interconnection is the most important for realizing generality of processing. It is implemented by rewriting the CGH on the PAL-SLM by using liquid crystal display controlled by a personal computer. Consequently, practical functions of interconnection are obtained and demonstrated. Recently, a vertical cavity surface emitting laser diode (VCSEL) array for integration of the optical interconnection is examined the system and the computing capabilities of the whole system for parallel processing including usage of the reconngurable interconnection are demonstrated. The block diagram of the system is shown in Fig. 1. The VCSEL array used in the system is developed by NEC Corpp4] and has 7 x 8 LDs (wavelength: 975nm, output power: 1mW, pitch: 125 m) in a chip. In order to match with the VCSEL array, the PAL-SLM used in the system is developed for the wavelength to modulate the light from the VCSEL array.The PD array is 16 x 16 PIN photo diode array which has parallel ampliied output. The sensitivity of …
منابع مشابه
Architecture Description and Prototype Demonstration of Optoelectronic Parallel-Matching Architecture
We propose an optoelectronic parallel-matching architecture (PMA) that provides powerful processing capability for distributed algorithms comparing with traditional parallel computing architectures. The PMA is composed of a parallel-matching (PM) module and multiple processing elements (PE's). The PM module is implemented by a large-fan-out free-space optical interconnection and parallel-matchi...
متن کاملDesign trade-offs in optoelectronic parallel processing systems using smart-SLMs
Optoelectronic devices with free-space optical interconnections offer new possibilities in massively parallel processing. The trade-offs involved in system design and device selection for optoelectronic implementations are examined. System design trade-offs are approached from algorithmic and technological standpoints. From the algorithmic standpoint, new architectures based on expander graphs,...
متن کاملAn MIMD Algorithm for Finding Maximum and Minimum on OMTSE Architecture
Optical Multi-Trees with Shuffle Exchange (OMTSE), recently proposed, is an efficient model of optoelectronic parallel computer. The OMTSE interconnection system consists of n factor networks called TSE networks, which are organized in the form of an n× n matrix. Each factor network has n number of leaf nodes. The network has a total of 3n/2 nodes. The diameter and bisection width of the networ...
متن کاملHardware Description Language for Optical Processing (hadlop): a Simulation Environment for Parallel Optoelectronic Architectures.
We present a simulation environment for parallel optoelectronic data-processing systems, and we especially consider the fusion of optoelectronic integrated circuits and optical interconnection modules. hadlop, which stands for hardware description language for optical processing, is a simulator that works at the digital design level. So far, hadlop has allowed algorithm and architecture studies...
متن کاملOptoelectronic parallel-matching architecture: architecture description, performance estimation, and prototype demonstration.
We propose an optoelectronic parallel-matching architecture (PMA) that provides powerful processing capabilities in global processing compared with conventional parallel-computing architectures. The PMA is composed of a global processor called a parallel-matching (PM) module and multiple processing elements (PE's). The PM module is implemented by a large-fan-out free-space optical interconnecti...
متن کامل