Please use this identifier to cite or link to this item:
Full metadata record
|dc.description.abstract||In recent years, the embedded and mobile nature of modern computing systems has led to an increased need for high performance and energy efficiency. As a result, energy dissipation has become a first class concern in the design of integrated circuits. Towards this direction, approximate (or inexact) computing appears as an emerging and promising solution for energy-efficient systems design, exploiting the inherent error/noise resilience of various applications involving media processing, machine learning, data mining and statistics. In this diploma thesis an exploration of new techniques of approximate computing is performed. Therefore, 5 techniques were developed with the purpose of energy dissipation and high accuracy. After the implementation of the techniques in Verilog code, several simulations were ran in order to compute the energy consumption and error. The tools used to perform these simulations were Synopsys Design Compiler, Mentor Graphics ModelSim and Matlab. Lastly, the comparison between the techniques was possible with the help of Pareto diagrams. Two types were presented. The first type is an Energy-Error diagram and the second an Area-Error diagram.||en_US|
|dc.title||Arithmetic-Aware Approximation Techniques for Energy-Efficient Inexact Circuits||en_US|
|dc.department||Τομέας Τεχνολογίας Πληροφορικής και Υπολογιστών||en_US|
|Appears in Collections:||Διπλωματικές Εργασίες - Theses|
Items in Artemis are protected by copyright, with all rights reserved, unless otherwise indicated.