Kepler GPU accelerated recursive sorting using dynamic parallelism

Abstract

Summary This paper focuses on the performance gain obtained on the Kepler graphics processing units (GPUs) for multi‐key quicksort. Because multi‐key quicksort is a recursive‐based algorithm, many of the researchers have found it tedious to parallelize the algorithm on the multi and many core architectures. A survey of the state‐of‐the‐art string sorting algorithms and a robust insight of the Kepler GPU architecture gave rise to an intriguing research idea of matching the template of multi‐key quicksort with the dynamic parallelism feature offered by the Kepler‐based GPU's. The CPU parallel implementation has an improvement of 33 to 50% and 62 to 75 improvement when compared with 8‐bit and 16‐bit parallel most significant digit radix sort, respectively. The GPU implementation of multi‐key quicksort gives 6× to 18× speed up compared with the CPU parallel implementation of parallel multi‐key quicksort. The GPU implementation of multi‐key quicksort achieves 1.5× to 3× speed up when compared with the GPU implementation of string sorting algorithm using singleton elements in the literature. Copyright © 2016 John Wiley & Sons, Ltd.

A radix sorting parallel algorithm suitable for graphic processing unit computing

“…T parallel−CUDA and T parallel−NOpenCL are defined by T p = T kernel + T ovehead + T other (5) where T kernel is the total of the execution times of the kernels on the GPU, T ovehead is the total of the data transfer overhead on the CPU and the GPU, and T other is the total of the execution times of the data structure initialization, and so on. 50 The speedup ratio reflects the overall efficiency improvement of the parallel algorithm in the corresponding architecture compared to the CPU sequential algorithm and can be used for objective evaluation of the actual system speed.…”

“…4 The Compute Unified Device Architecture (CUDA) uses the parallel computing engine of the NVIDIA Graphic Processing Unit (GPU) to achieve a more efficient computing solution than the CPU for solving many complex computing tasks. 5 However, there are certain problems. For example, in terms of software porting, NVIDIA GPUs and AMD GPUs are not compatible with each other, and parallel algorithms are not portable.…”

“…The Compute Unified Device Architecture (CUDA) uses the parallel computing engine of the NVIDIA Graphic Processing Unit (GPU) to achieve a more efficient computing solution than the CPU for solving many complex computing tasks 5 . However, there are certain problems.…”

“…and(3)are both prefix calculations with time complexity O(n/p + log p). In steps (4) and(5), an auxiliary array Q is used to update the array number, which takes O(n/p) times. Therefore, the total time complexity of the radix sorting parallel algorithm is O(m(n/p + log p) ).…”

How Many CPU Cores is an FPGA Worth? Lessons Learned from Accelerating String Sorting on a CPU-FPGA System

“…This is a building block of suffix sorting, used in string matching, and database index construction [13]. Parallel string sorting algorithms have been proposed on CPUs [14] and GPUs [15], however, to the best of our knowledge, no hardware accelerator for this problem has been made available yet. Indeed, handling variable-length keys in hardware is not only challenging per se but also involves key comparisons that can become expensive as keys are arbitrarily long.…”