| The technological progression of electronic circuit design and manufacturing has led to an era of smaller, faster, and cheaper devices. These benefits are obviously advantageous to both electronics producers and consumers; smaller circuits allow for more complex systems with faster components that are capable of performing more operations in the same amount of time. In light of this, circuits have proliferated in virtually every public and private sector market, with applications no longer limited to personal and high-performance computing, but ranging from telecommunications to biomedicine.; However, such smaller and faster devices have also introduced new challenges regarding their design, particularly in analysis and simulation. As increased circuit speeds (i.e. faster operating frequencies) drive down the corresponding wavelengths to a scale commensurate with components, traditional lumped-element approaches for modeling device operation become insufficient. Instead, a rigorous analysis via a full-wave electromagnetic technique for solving Maxwell's Equations is required. Unfortunately, the computational intensity of these methods renders them impractical for most circuit analysis tools. However, through the incorporation of a dedicated hardware co-processor devoted to accelerating the large amount of necessary arithmetic, such analysis becomes highly practical even on a desktop PC running a commodity microprocessor. To this end, this thesis presents my implementation of a custom hardware-based acceleration system devoted to speeding up the full-wave analysis of electronic circuits. |
