Details
Presenter(s)
Display Name
NANDAGOPAL RAJASEKARAN
- Affiliation
- Country
Abstract
Power function is one of the elementary arithmetic operation, useful for wide variety of applications including image-processing, signal processing, and recently neural network modules. Floating-point implementation preserves the capability of storing wide range of real numbers, when compared to fixed-point and integer data-format. In this work, power function is realized in floating-point format using logarithmic, and anti-logarithmic compute modules. A novel piece-wise-linear (PWL) method to approximately represent nonlinear logarithmic and anti-logarithmic functions is proposed.