sorry for the late announcement.
This time we will have a talk from Mario Mulansky about optimization of numerical algorithms. AND we will have a limited amound of free pizza and beers :).
Efficient computational methods are crucial in many parts of science. Here, we present a strategy to speed up Runge-Kutta-based ODE simulations of large systems with nearest-neighbor coupling. By introducing granularity we are able to transform the algorithm from bandwidth bound to CPU bound. By additionally employing SIMD instructions we are able to boost the efficiency even further. In total, a performance increase of up to a factor three is reached when using cache optimization and SIMD instructions compared to a standard implementation.