High Parallel Skyline Computation over Low-Cardinality Domains

Abstract

Abstract.A Skyline query retrieves all objects in a dataset that are not dominated by other objects according to some given criteria. Although there are a few parallel Skyline algorithms on multicore processors, it is still a challenging task to fully exploit the advantages of such modern hardware architectures for efficient Skyline computation. In this paper we present high-performance parallel Skyline algorithms based on the lattice structure generated by a Skyline query. We compare our methods with the state-of-the-art algorithms for multicore Skyline processing. Experimental results on synthetic and real datasets show that our new algorithms outperform state-of-the-art multicore Skyline techniques for low-cardinality domains. Our algorithms have linear runtime complexity and fully play on modern hardware architectures.

Template Skycube Algorithms for Heterogeneous Parallelism on Multicore and GPU Architectures

“…There are no single-node parallel skycube algorithms, but skylines can be computed on FPGAs [38], GPUs [1,2,10], and multicore CPUs [3,8,9,12,13,17,20,26,29]. For the CPU, PSkyline [17,29] is a naive divide-and-conquer algorithm that distributes the data to each core, independently computes a local skyline, and then merges the results.…”

“…VSkyline [9] orthogonally proposes using SIMD registers to accelerate DTs, which are available on modern multicore machines. Scalagon [12,13] constructs a partial order of data values and traverses the resultant partial order lattice to produce the skyline, which is effective when the number of distinct values for each attribute is low. Hybrid [8] applies tiling and point-based partitioning.…”

“…Meanwhile, there is minimal complexity introduced by cross-device task parallelism, because each device processes an independent atomic task (i.e., cuboid). It is relatively easy to incorporate new parallel skyline algorithms, algorithms that target particular settings (e.g., low cardinality domains [12,13]), or specialisations on other architectures for which parallel skyline algorithms exist (e.g., FPGAs [38]).…”

Parallel Skyline Computation Exploiting the Lattice Structure

“…Since a large number of dominance tests can often be performed independently, Skyline computation has a good potential to exploit multicore architectures as described in (Selke, Lofi, and Balke, 2010;Im, Park, and Park, 2011;Liknes, Vlachou, Doulkeridis, and Nørvåg, 2014). This paper is an extended version of (Endres, and Kießling, 2014) and we present enhanced algorithms for high-performance parallel Skyline computation which do not depend on tuple comparisons, but on the lattice structure constructed by a Skyline query over low-cardinality domains. Following (Morse, Patel, and Jagadish, 2007;Chomicki, Ciaccia, and Meneghetti, 2013) many Skyline applications involve domains with small cardinalities -these cardinalities are either inherently small (such as star ratings for hotels), or can naturally be mapped to low-cardinality domains (such as price ranges on hotels).…”