Date of Award
Master of Science in Computer Engineering - (M.S.)
Electrical and Computer Engineering
Alexandros V. Gerbessiotis
Power flow analysis consists of computationally intensive calculations on large matrices, consumes several hours of computational time, and has shown the need for the implementation of application-specific parallel machines. The potential of Single-Instruction stream Multiple-Data stream (SIMD) parallel architectures for efficient operations on large matrices has been demonstrated as seen in the case of many existing supercomputers. The unsuitability of existing parallel machines for low-cost power system applications, their long design cycles, and the difficulty in using them show the need for application-specific SIMI) machines. Advances in VLSI technology and Field-Programmable Gate-Arrays (FPGAs) enable the implementation of Custom Computing Machines (CCMs) which can yield better performance for specific applications. The advent of SoftCore processors made it possible to integrate reconfigurable logic as a slave to a peripheral bus and has demonstrated the ability in the rapid prototyping of complete systems on programmable chips. This thesis aims at designing and implementing an FPGA-based SIMI) machine for power flow analysis. It presents the architecture of an SIMI) machine that consists of an array of processing elements with mesh interconnection and a Soft-Core processor; the latter is used as the host. The FPGAbased SIMI) machine is implemented on the Annapolis Microsystems Wildstar-II board that contains multiple Virtex-II FPGAs. The Soft-Core processor used is the Xilinx Microblaze and the application targeted is matrix multiplication.
Kunta, Tirupathi Rao, "Design of an FPGA-based parallel SIMD machine for power flow analysis" (2003). Theses. 625.