Improved GPU near neighbours performance for multi-agent simulations

Chisholm, Robert; Maddock, Steve; Richmond, Paul

doi:10.1016/j.jpdc.2019.11.002

Cited by 4 publications

(5 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2 (a), was initiated. Firstly, D3Q27 neighbours [ 39 ] or 3D Moore neighbourhood [ 40 ] of one virus was extracted and added to the cluster matrix. The overlap of unit movement was then determined, as illustrated in Fig.…”

Section: Methodsmentioning

confidence: 99%

Automatic 3D cluster modelling of COVID-19 through voxel-based redistribution

Liu

Wang

2021

Powder Technology

View full text Add to dashboard Cite

Section: Methodsmentioning

confidence: 99%

Automatic 3D cluster modelling of COVID-19 through voxel-based redistribution

Liu

Wang

2021

Powder Technology

View full text Add to dashboard Cite

“…Effectively updating Verlet lists is essential to accomplishing high-performance simulations. One popular strategy is the grid-based technique, also referred to as cell-lists Hoetzlein (2014); Gross et al (2019); Chisholm et al (2020); Bender, Jan (2024). Using this approach, the simulation domain is partitioned into a regular grid of cells, and the locations of the particles are used to assign them to the cells.…”

Section: Existing Workmentioning

confidence: 99%

SOMA-BD: Brownian Dynamics Simulation for Soft Matter on GPU

Crespin,

Porro,

Cerbelaud

et al. 2024

Preprint

View full text Add to dashboard Cite

This paper introduces a high-performance Brownian Dynamics simulation code designed for Nvidia GPUs. The code prioritizes accuracy with double-precision calculations and utilizes a bucket sort neighbor list for efficient particle interaction management. Additionally, a CUDA occupancy-aware autotuning system optimizes thread block size for diverse hardware configurations. Benchmarks demonstrate significant performance gains compared to existing tools, particularly for large-scale simulations with various interaction potentials (Lennard-Jones, Yukawa, DLVO). The software offers user-friendly configuration via JSON files and supports restarting simulations from both binary and XYZ data formats. Future work includes expanding supported interaction potentials, refining autotuning and investigating scalability. This open-source software empowers researchers in material science and related fields to conduct efficient and accurate BD simulations.

show abstract

“…22 But they are not quite suitable in finding all the neighbors within a fixed radius. 23 In general, kd-trees are found to be suitable for static data. Bkd-tree 24 is an improved version of kd-tree, designed to index large multi-dimensional databases.…”

Section: Scalabilitymentioning

confidence: 99%

“…Kd‐trees 21 are also widely used in applications that require finding

k

nearest neighbors, with

k

\ge 1

22 . But they are not quite suitable in finding all the neighbors within a fixed radius 23 . In general, kd‐trees are found to be suitable for static data.…”

Section: Related Workmentioning

confidence: 99%

Single‐linkage clustering of dynamic data

Anju

Nasre

2022

Concurrency and Computation

View full text Add to dashboard Cite

The surge in data sizes in fluid processing applications necessitates partitioning the data into clusters and studying their representatives instead of studying each voxel data point. In addition, the dynamic nature of these data poses further challenges.Under such circumstances, it becomes essential to develop an approach that can handle the delta data with minimal updates to the underlying data structure, without processing the complete data from scratch on every update. However, this poses synchronization challenges in parallelization. In this article, we propose SLCODD (single-linkage clustering of dynamic data), a geometric distance based dynamic clustering and its multi-core parallelization using OpenMP. To improve efficiency, SLCODD exploits geometric properties of the bounding squares. We illustrate trade-offs in various ways of performing point additions to clusters, point deletions, and their batched versions. Using a suite of large inputs, we demonstrate the effectiveness of SLCODD. SLCODD's fully dynamic version achieves a substantial geomean speedup of 8.49× over the static parallel version and of 5× over the dynamic sequential version.

show abstract

Improved GPU near neighbours performance for multi-agent simulations

Cited by 4 publications

References 16 publications

Automatic 3D cluster modelling of COVID-19 through voxel-based redistribution

Automatic 3D cluster modelling of COVID-19 through voxel-based redistribution

SOMA-BD: Brownian Dynamics Simulation for Soft Matter on GPU

Single‐linkage clustering of dynamic data

Contact Info

Product

Resources

About