The FAµST version 3.30.0 follows the previous news which showed the GPU Faust slicing optimization. This time this is the GPU Faust indexing that has been optimized. The figure below demonstrates how the optimization speeds up the related operations since the previous release (3.29.1).
The code to reproduce the figure is available here.
The figure was produced using a NVIDIA GPU GTX 980 configured with CUDA 11.