New Generation of WIPL-D in-Core Multi-GPU Solver

Acceleration of in-core LU decomposition by using multiple GPUs in parallel and comparison of performance of two different GPU types, is presented in this paper. The presented algorithm for LU decomposition enables highly efficient utilization of an arbitrary number of GPUs. Comparison of performance of up to 8 GTX 680 GPUs and up to 4 GTX 1080 Ti GPUs is shown. The results presented show that a dense MoM matrix with 100 000 complex unknowns in single precision can be LU decomposed in 3.7 minutes, on a personal computer equipped with 4 GTX 1080 Ti GPUs.