A Vector-Like Reconfigurable Floating-Point Unit for the Logarithm

Table 3

Min, max, average, and mean squared error of logarithm implementations with respect to GNU functions.

Program/unit	Min	Max	Avg	MSE
SP-ICSILog	4.228𝑒−7	1.210𝑒−4	4.438𝑒−5	2.689𝑒−9
DP-ICSILog	3.140𝑒−9	1.205𝑒−4	4.437𝑒−5	2.688𝑒−9
LAU/VLAU	4.228𝑒−7	1.210𝑒−4	4.437𝑒−5	2.690𝑒−9
SP-MKL	0.0𝑒−0	3.815𝑒−6	5.003𝑒−7	1.65𝑒−14
DP-MKL	0.0𝑒−0	4.44𝑒−16	4.52𝑒−22	4.93𝑒−38