Effective GPU Strategies for LU Decomposition

GPUs are becoming an attractive computing platform not only for traditional graphics computation but also for general-purpose computation because of the computational power, programmability and comparatively low cost of modern GPUs. This has lead to a variety of complex GPGPU applications with significant performance improvements. The LU decomposition represents a fundamental step in many computationally intensive scientific applications and it is often the costly step in the solution process because of the impact of size of the matrix. In this paper we implement three different variants of the LU decomposition algorithm on a Tesla C1060 and the most significant LU decomposition that fits the highly parallel architecture of modern GPUs is found to be Update through Column with shared memory access implementation. Keywords—LU decomposition, CUDA, GPGPU