
Accuracy and Physical Characterization of Approximate Arithmetic Circuits
Author(s) -
Daniela Catelan,
Ricardo Santos,
Liana Duenha
Publication year - 2020
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/wscad.2020.14065
Subject(s) - field programmable gate array , computer science , electronic circuit , ranging , set (abstract data type) , power (physics) , circuit design , dissipation , physical design , field (mathematics) , computer hardware , arithmetic , computer engineering , embedded system , electrical engineering , engineering , mathematics , telecommunications , physics , quantum mechanics , pure mathematics , thermodynamics , programming language
With the end of Dennard's scale, designers have been looking for new alternatives and approximate computing (AC) has managed to attract the attention of researchers, by offering techniques ranging from the application level to the circuit level. When applying approximate circuit techniques in hardware design, the program user may speed up the application while a designer may save area and power dissipation at the cost of less accuracy on the operations results. This paper discusses the compromise between accuracy versus physical efciency by presenting a set of experiments and results of tailor-made approximate arithmetic circuits on Field-Programmable Gate Array (FPGA) platforms. Our results reveal that an approximate circuit with accuracy control could not be useful if the goal is to save circuit area or even power dissipation. Even for circuits that seem to have power efciency, we should care about the size and prototyping platform where the hardware will be used.