Ameliorating Memory Contention of OLAP operators on GPU Processors