Accelerated impulse response calculation for indoor optical communication channels

In order to compare the accelerated implementation with the original software, we observe performance and accuracy. Both sets of simulations are run on the same workstation - an HP Z400 with an Intel Xeon W3505 processor. The GPU used for the accelerated simulations is an Nvidia GeForce 285 with CUDA version 2.3. Table I offers an overview of the performance improvements provided by the accelerated implementation in multiple scenarios and Figures 1-4 show detailed analysis of the performance across a range of complexity metrics for the Cubicle scenario discussed in [2].