• Journal of Digital Contents Society
• /
• v.18 no.6
• /
• pp.1151-1156
• /
• 2017

It is common to achieve lower performance in traversing tree data structures in GPU than one expects. In this paper, we analyze the reason of lower-than-expected performance in GPU tree traversal and present that the warp divergences is caused by the branch instructions ("if${\ldots}$ else") which appear commonly in tree traversal CUDA codes. Also, we compare the parallel shifted sort algorithm which can reduce the number of warp divergences with a kd-tree CUDA implementation to show that the shifted sort algorithm can work faster than the kd-tree CUDA implementation thanks to less warp divergences. As the analysis result, the shifted sort algorithm worked about 16-fold faster than the kd-tree CUDA implementation for $2^{23}$ query points and $2^{23}$ data points in $R^3$ space. The performance gaps tend to increase in proportion to the number of query points and data points.

• Journal of Digital Contents Society
• /
• v.15 no.3
• /
• pp.347-355
• /
• 2014

The general octree structure is common for various applications including computer graphics, geometry information analysis and query. Unfortunately, the general octree approach causes duplicated sample data and discrepancy between sampling and representation positions when applied to sample continuous spatial information, for example, signed distance fields. To address these issues, some researchers introduced the dual octree. In this paper, the weakness of the dual octree approach will be illustrated by focusing on the fact that the dual octree cannot access some specific continuous zones asymptotically. This paper shows that the primal tree presented by Lefebvre and Hoppe can solve all the problems above. Also, this paper presents a three-dimensional primal tree traversal algorithm based the Morton codes which will help to parallelize the primal tree method.

• Journal of Korea Spatial Information System Society
• /
• v.7 no.2 s.14
• /
• pp.81-89
• /
• 2005

Declustering and parallel index structures are important research areas to improve a performance of databases. Previous researches proposed several distribution schemes for parallel R-trees, however there is no search schemes to be suitable for the index. In this paper, we propose schemes to improve the performance of range queries for distribute parallel indexes. The proposed schemes use the features that a parallel disk can read multiple nodes from various disks. The proposed schemes are verified using various implementations and performance evaluations. We propose new schemes which can read multiple nodes from multiple disks in contrast that to the previous schemes which can read a node from disk. The experimental evaluation shows that the proposed schemes give us the performance improvement by 40% from the previous researches.

• Journal of KIISE:Computer Systems and Theory
• /
• v.33 no.1_2
• /
• pp.1-14
• /
• 2006

Sort-last parallel rendering using a cluster of GPUs has been widely used as an efficient method for visualizing large- scale volume datasets. The performance of this method is constrained by load balancing when data parallelism is included. In previous works static partitioning could lead to self-balance when only task level parallelism is included. In this paper, we present a load balancing scheme that adapts to the characteristic of volume dataset when data parallelism is also employed. We effectively combine the hierarchical data structures (octree and BSP tree) in order to skip empty regions and distribute workload to corresponding rendering nodes. Moreover, we also exploit a 3D clustering method to determine visibility order and save the AGP bandwidths on each rendering node. Experimental results show that our scheme can achieve significant performance gains compared with traditional static load distribution schemes.

• The KIPS Transactions:PartD
• /
• v.19D no.2
• /
• pp.151-160
• /
• 2012

Most of the distributed high-dimensional indexing structures provide a reasonable search performance especially when the dataset is uniformly distributed. However, in case when the dataset is clustered or skewed, the search performances gradually degrade as compared with the uniformly distributed dataset. We propose a method of improving the k-nearest neighbor search performance for the distributed vector approximation-tree based on the strongly clustered or skewed dataset. The basic idea is to compute volumes of the leaf nodes on the top-tree of a distributed vector approximation-tree and to assign different number of bits to them in order to assure an identification performance of vector approximation. In other words, it can be done by assigning more bits to the high-density clusters. We conducted experiments to compare the search performance with the distributed hybrid spill-tree and distributed vector approximation-tree by using the synthetic and real data sets. The experimental results show that our proposed scheme provides consistent results with significant performance improvements of the distributed vector approximation-tree for strongly clustered or skewed datasets.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.6
• /
• pp.707-711
• /
• 2010

Most relational database systems provide $B^+$-trees as their main index structures, and use bulk-loading techniques for creating new $B^+$-trees on existing data from scratch. Although bulk loadings are more effective than inserting keys one by one, they are still time-consuming because they have to sort all the keys from large data. To improve the performance of bulk loadings, this paper proposes an efficient parallel bulk loading method for $B^+$-trees based on CUDA, which is a parallel computing architecture developed by NVIDIA to utilize computing powers of graphic processor units for general purpose computing. Experimental results show that the proposed method enhance the performance more than 70 percents compared to existing bulk loading methods.

• Nuclear Engineering and Technology
• /
• v.54 no.8
• /
• pp.2783-2791
• /
• 2022

Molecular dynamics simulations were performed to predict the behavior of graphite atoms under neutron irradiation using large-scale atomic/molecular massively parallel simulator (LAMMPS) package with adaptive intermolecular reactive empirical bond order (AIREBOM) potential. Defect structures of graphite were compared with results from previous studies by means of density functional theory (DFT) calculations. The quantitative relation between primary knock-on atom (PKA) energy and irradiation damage on graphite was calculated. and the effect of PKA direction on the amount of defects is estimated by counting displaced atoms. Defects are classified into four groups: structural defects, energy defects, vacancies, and near-defect structures, where a structural defect is further subdivided into six types by decision tree method which is one of the supervised machine learning techniques.

• Journal of KIISE:Databases
• /
• v.29 no.1
• /
• pp.58-71
• /
• 2002

Generally, multi-dimensional data such as image and spatial data require large amount of storage space. There is a limit to store and manage those large amount of data in single workstation. If we manage the data on parallel computing environment which is being actively researched these days, we can get highly improved performance. In this paper, we propose a parallel high-dimensional index structure that exploits the parallelism of the parallel computing environment. The proposed index structure is nP(processor)-n$\times$mD(disk) architecture which is the hybrid type of nP-nD and lP-nD. Its node structure increases fan-out and reduces the height of a index tree. Also, A range search algorithm that maximizes I/O parallelism is devised, and it is applied to K-nearest neighbor queries. Through various experiments, it is shown that the proposed method outperforms other parallel index structures.

• Applied Microscopy
• /
• v.42 no.3
• /
• pp.124-128
• /
• 2012

Leaf surface structures were investigated in the Chinese juniper Juniperus chinensis by scanning electron microscopy. Adult scale leaves were collected from the tree, air-dried at room temperature, and sputter-coated with gold without further specimen preparation. Approximately fi ve stomata were locally distributed and arranged in clusters on the leaf surface. Stomata were ovoid and ca. 40 ${\mu}m$ long. The epicuticular wax structures of J. chinensis leaves were tubules and platelets. Numerous tubules were evident on the leaf regions where stomata were found. The tubules were cylindrical, straight, and ca. 1 ${\mu}m$ in length. They almost clothed the stomatal guard cells, and occluded the slit-shaped stomatal apertures. Moreover, the wax ridges were flat crystalloids that were connected to the surface by their narrow side. They did not have distinct edges, and their width/height ratio varied. In particular, the wax ridges could be discerned on the leaf regions where stomata were not present nearby. Since the wax ridges did not have distinct edges on their margin, they were identified as platelets. Instances were noted where platelets were oriented either parallel to each other or perpendicular to the cuticle surface. These results can be used in biomimetics to design the hierarchical structures for mimicking the plant innate properties such as hydrophobicity and self-cleaning effects of the leaf surface.

• Korean Journal of Agricultural Science
• /
• v.37 no.3
• /
• pp.465-469
• /
• 2010

In order to develop effective recycling technologies of disused railway wood ties, wood properties and residual creosote oil of wood ties were investigated. Among the total 50 wood ties, 32 wood ties were identified as tropical hardwood tree, such as, Keruing (Dipterocarpus spp.), Kempas (Koompassia malaccensis), Kapur (Dryobalanops spp.) Naytoh (Palaquium rostratum), and so on. Disused wood ties showed mostly sound structure without degradation of cell walls by decay fungi. Disused wood ties showed high strengths of bending and compressive parallel to grain because degradation of wood properties was hardly occurred in use under exterior condition. Disused railway wood ties had relatively poor depth of penetration and residual of creosote oil because of refractory wood structures. These results suggest that disused railway wood ties may be useful as recycling wood wastes.