Parallel Range Query Processing on R-Tree with Graphics Processing Unit

Yu, Boseon; Kim, Hyunduk; Choi, Wonik; Kwon, Dongseop

doi:10.1109/dasc.2011.200

Cited by 11 publications

(5 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We know of no experimental evaluation of these algorithms. There are also parallel implementation-based works such as parallel R-trees [40], parallel sweepline algorithms [45], and algorithms focusing on distributed systems [60] and GPUs [59]. No theoretical guarantees are provided in these papers.…”

Section: Related Workmentioning

confidence: 99%

Parallel Range, Segment and Rectangle Queries with Augmented Maps

Sun

Blelloch

2019

2019 Proceedings of the Twenty-First Workshop on Algorithm Engineering and Experiments (ALENEX)

View full text Add to dashboard Cite

The range, segment and rectangle query problems are fundamental problems in computational geometry, and have extensive applications in many domains. Despite the significant theoretical work on these problems, efficient implementations can be complicated. We know of very few practical implementations of the algorithms in parallel, and most implementations do not have tight theoretical bounds. In this paper, we focus on simple and efficient parallel algorithms and implementations for range, segment and rectangle queries, which have tight worst-case bound in theory and good parallel performance in practice. We propose to use a simple framework (the augmented map) to model the problem. Based on the augmented map interface, we develop both multi-level tree structures and sweepline algorithms supporting range, segment and rectangle queries in two dimensions. For the sweepline algorithms, we also propose a parallel paradigm and show corresponding cost bounds. All of our data structures are work-efficient to build in theory (O(n log n) sequential work) and achieve a low parallel depth (polylogarithmic for the multi-level tree structures, and O(n ) for sweepline algorithms). The query time is almost linear to the output size.We have implemented all the data structures described in the paper using a parallel augmented map library. Based on the library each data structure only requires about 100 lines of C++ code. We test their performance on large data sets (up to 10 8 elements) and a machine with 72-cores (144 hyperthreads). The parallel construction achieves 32-68x speedup. Speedup numbers on queries are up to 126-fold. Our sequential implementation outperforms the CGAL library by at least 2x in both construction and queries. Our sequential implementation can be slightly slower than the R-tree in the Boost library in some cases (0.6-2.5x), but has significantly better query performance (1.6-1400x) than Boost.Many data structures are designed for solving range, segment and rectangle queries such as range trees [17], segment trees [19], kd-trees [16], R-trees [14,50,52], priority trees [44], and many others [58,30,42,48], which are then applied to later research in various areas [2,38,4,20,35,25,54,5,8,7,18,49,21]. Many of them are augmented tree structures. The standard range tree has construction time O(n log n) and query time O(k + log 2 n) for input size n and output size k. Using fractional cascading [41,57], the query time can be reduced to O(k + log n). We did not employ such optimizations, but instead show that the our version using parallel augmented maps achieve good parallelism in practice, and is simple and easy for engineering. The terminology "segment tree" refers to different data structures in the literature. Our version is similar to some previous works [27,12,3]. Previous solutions for rectangle queries usually use combinations of range trees, segment trees, interval trees, and priority

show abstract

Section: Related Workmentioning

confidence: 99%

Parallel Range, Segment and Rectangle Queries with Augmented Maps

Sun

Blelloch

2019

2019 Proceedings of the Twenty-First Workshop on Algorithm Engineering and Experiments (ALENEX)

View full text Add to dashboard Cite

show abstract

“…Other works consider the problem of building R-Trees (and possible derivations) from scratch [2], even recurring to hybrid approaches based on the combined use of CPU and GPU for range queries computation [25,26]. While the goals of some of these works are different with respect to those of the present work, it is interesting to notice how solving certain problems is particularly recurrent and challenging when processing massive spatial data by means of massively parallel architectures, that is, (i) find a solution able to distribute the workload in the most possible uniform way (depending also on the data spatial distribution); (ii) arrange spatial data by using proper GPU-friendly, lightweight, regular data structures that allow to use the GPUs features effectively; and (iii) exploit spatial locality as much as possible.…”

Section: Related Workmentioning

confidence: 99%

“…In this batch of experiments, whose results are shown in Figure 23, we vary the query area. All the queries are equally sized during a single experiment, while the amount of objects is fixed (700K for uniform, 500K for gaussian and network), as well as the query rate (100%) and the number of hotspots (25) Variable amount of objects and variable query area. In these experiments we consider different amounts of objects, each one issuing a query whose area is decided independently of the other objects and according to a uniform distribution in the [(200u) 2 , (400u) 2 ] range.…”

Section: Performance Analysis For Different Spatial Distributions Amo...mentioning

confidence: 99%

Manycore GPU processing of repeated range queries over streams of moving objects observations

Lettich

Orlando

Silvestri

et al. 2016

Concurrency and Computation

View full text Add to dashboard Cite

The ability to timely process significant amounts of continuously updated spatial data is mandatory for an increasing number of applications. Parallelism enables such applications to face this data-intensive challenge and allows the devised systems to feature low latency and high scalability. In this paper, we focus on a specific data-intensive problem concerning the repeated processing of huge amounts of range queries over massive sets of moving objects, where the spatial extent of queries and objects is continuously modified over time. To tackle this problem and significantly accelerate query processing, we devise a hybrid CPU/GPU pipeline that compresses data output and saves query processing work. The devised system relies on an ad- hoc spatial index leading to a problem decomposition that results in a set of independent data-parallel tasks. The index is based on a point-region quadtree space decomposition and allows to tackle effectively a broad range of spatial object distributions, even those very skewed. Also, to deal with the architectural peculiarities and limitations of the GPUs, we adopt non-trivial GPU data structures that avoid the need of locked memory accesses while favouring coalesced memory accesses, thus enhancing the overall memory throughput. To the best of our knowledge, this is the first work that exploits GPUs to efficiently solve repeated range queries over massive sets of continuously moving objects, possibly characterized by highly skewed spatial distributions. In comparison with state-of-the-art CPU-based implementations, our method highlights significant speedups in the order of 10 20 , depending on the dataset

show abstract

“…In [23], a GPU-based implementation of R-Tree is presented: differently from our approach, parallelism is not exploited to increase the performance of a single query, but to run different queries in parallel. Finally, the work in [30] proposes a technique to speedup processing of large R-Tree structures by storing recently visited nodes on the GPU memory and re-use them for future queries. Different from our approach, the authors focus on structures that do not fit in main memory; only a portion of the computation is performed on the GPU and several interactions between the CPU and the GPU may take place while navigating the tree.…”

Section: Related Workmentioning

confidence: 99%

High-Performance Location-Aware Publish-Subscribe on GPUs

Cugola

Margara

2012

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Adding location-awareness to publish-subscribe middleware infrastructures would open-up new opportunities to use this technology in the hot area of mobile applications. On the other hand, this requires to radically change the way published events are matched against received subscriptions. In this paper we examine this issue in detail and we present CLCB, a new algorithm using CUDA GPUs for massively parallel, high-performance, location-aware publish-subscribe matching and its implementation into a matching component that allows to easily build a full-edged middleware system. A comparison with the state-of-the-art in this area shows the impressive increment in performance that GPUs may enable, even in this domain. At the same time, our performance analysis allows to identify those peculiar aspects of GPU programming that mostly impact the performance of this kind of algorithm

show abstract

Parallel Range Query Processing on R-Tree with Graphics Processing Unit

Cited by 11 publications

References 18 publications

Parallel Range, Segment and Rectangle Queries with Augmented Maps

Parallel Range, Segment and Rectangle Queries with Augmented Maps

Manycore GPU processing of repeated range queries over streams of moving objects observations

High-Performance Location-Aware Publish-Subscribe on GPUs

Contact Info

Product

Resources

About