Sorting with GPUs: A Survey

Abstract

Sorting is a fundamental operation in computer science and is a bottleneck in many important fields. Sorting is critical to database applications, online search and indexing, biomedical computing, and many other applications. The explosive growth in computational power and availability of GPU coprocessors has allowed sort operations on GPUs to be done much faster than any equivalently priced CPU. Current trends in GPU computing shows that this explosive growth in GPU capabilities is likely to continue for some time. As such, there is a need to develop algorithms to effectively harness the power of GPUs for crucial applications such as sorting.

An efficient deterministic perturbation theory for selected configuration interaction methods

“…The efficiency of this approach stems from the fact that sorting algorithms have been extensively optimized on modern computers, due to their ubiquitous appearance in all aspects of computation [35]. The minimization of cache misses on modern computers are likely among the main reasons that sorting approaches appear to be significantly more efficient than other algorithms [9,36,37], although many aspects of sorting algorithms have been designed for optimal usage on modern computers.…”

“…We generally find that a single GPU is about 10 times faster than a single CPU for sorting [9]. There is still much ongoing development of GPU sorting [35]. Much of the current runtime is used to move data onto and off the GPU, so a fully GPU implementation could have even further speed improvements.…”