Very Large Scale Optical Interconnect Systems For Different Types of Optical Interconnection Networks

Abstract

Abstract-The need for scalable systems in market demands in terms of lower computing costs and protection of customer investment in computing: scaling up the system to quickly meet business growth is obviously a better way of protecting investment: hardware, software, and human resources. A scalable system should be incrementally expanded, delivering linear incremental performance with a near linear cost increase, and with minimal system redesign (size scalability), additionally, it should be able to use successive, faster processors with minimal additional costs and redesign (generation scalability). On the architecture side, the key design element is the interconnection network. The interconnection network must be able to increase in size using few building blocks and with minimum redesign, deliver a bandwidth that grows linearly with the increase in system size, maintain a low or (constant) latency, incur linear cost increase, and readily support the use of new faster processors. The major problem is the ever-increasing speed of the processors themselves and the growing performance gap between processor technology and interconnect technology. Increased central processing unit (CPU) speeds and effectiveness of memory latency-tolerating techniques.

Linear Crossed Cube (LCQ): A New Interconnection Network Topology for Massively Parallel System

“…The different techniques to optimize the network performance are scheduling the tasks or simply the task assignment, load balancing and load sharing. In task scheduling each parallel process is viewed as a collection of related tasks and these tasks are scheduled to suitable nodes evenly [30] [31] [32]. The aim is to assure that no node should remain idle and the distribution of tasks is made in such a way that the load on individual node should remain balance at any point of time.…”