Skip to main content

Advertisement

Table 4 Benchmarks used for our experiments.

From: A Precise High-Level Power Consumption Model for Embedded Systems Software

Benchmark Description
DotP128 Dot product of a vector of 128 16-bit elements
m100 Matrix multiplication for 2 100 × 100 square matrices
FIR Computes a real FIR filter, Input data and filter taps are 16-bit
Sobel 3 × 3 Apply Sobel filter of 3 × 3 window to an image of 8192 pixels
Thresholding Performs a thresholding operation on an input image of 8192 pixels
Histogram Takes histogram of an image of 8192, 8-bit pixels
IIR Performs an auto-regressive moving-average (ARMA) filter with auto-regressive filter coefficients and moving-average filter coefficients
FFT16 × 16 Performs a mixed radix forwards FFT using a special sequence of coefficients
Correlation 3 × 3 Performs a point by point multiplication of the 3 × 3 mask with an input image