A High-Performance Deeply Pipelined Architecture for Elementary Transcendental Function Evaluation
Jing Chen, Xue Liu · 2017
Many scientific applications rely on the evaluation of elementary transcendental functions (e.g. sin, cos, exp, log). Software math libraries are a popular approach for realizing such functions, and frequently use series expansion or lookup-table-based techniques. However, software approaches necessarily suffer from the traditional overheads of fetching/decoding instructions, limited cache sizes, and so on. In this paper, we present a hardware accelerator for such elementary transcendental functions that delivers high computational throughput. The accelerator design is generic in the sense that it is not tied to a specific function. However, we demonstrate its utility by accelerating logarithm. The proposed accelerator is applicable to a wide range of high-throughput scientific computing applications.