SIMD-vectorized implementation of high order IRK integrators
Mikel, Joseba Makazaga, Ander Murua
We present a preliminary version of a SIMD-vectorized implementation of the sixteenth order 8-stage implicit Runge-Kutta integrator IRKGL16 implemented in the Julia package IRKGaussLegendre.jl. For numerical integrations of typical non-stiff problems performed in double precision, we show that a vectorized implementation of IRKGL16 that exploits the SIMD-based parallelism can clearly outperform high order explicit Runge-Kutta schemes available in the standard package DifferentialEquations.jl.