Fig. 13From: Parallel acceleration of CPU and GPU range queries over large data setsAverage execution times for the iterative CPU method (a) using 16 cores (solid lines), the CPU reduction method (b) using 16 cores, and the ideal hybrid GPU method (shown as dashed lines in a and b) by query size. The Zipf data set skews are appended (e.g., Zipf0 has a skew of 0). Both plots share the same x-axisBack to article page