By Marcel Leutenegger at École Polytechnique Fédérale de Lausanne (EPFL)
|27 Jul 2010||Helen Chen||
I regularily deal with a huge number of elementary functions as rounding, transcendentals, exponential and logarithm and was unsatisfied with the average performance. So, I started to provide my own assembler subroutines executing with full floating-point performance on any Intel Pentium II+ or compatible system. On recent Pentium 4 computers, I was able to improve the average performance by two to five times.
This archive is dedicated to any number cruncher using MATLAB 6.0 or newer.