GPU accelerated blood flow computation using the Lattice Boltzmann Method

Nita, Cosmin; Itu, Lucian Mihai; Suciu, Constantin

doi:10.1109/hpec.2013.6670324

Cited by 20 publications

(23 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An additional benefit of such an approach would be the simplified mesh generation process, which would be orders of magnitude faster and could potentially be completely automated, thereby removing the need for an engineer skilled in this field. With a Lattice Boltzmann method, again, the mesh generation procedure is simplified to the point of being trivial and the implicit parallelizability of the method means that 2D or 3D flow fields can be computed significantly faster on generic CPU-based supercomputing platforms or using graphics processing units [80].…”

Section: Future Of Virtual Fractional Flow Reservementioning

confidence: 99%

Advances in three-dimensional coronary imaging and computational fluid dynamics

Poon

Hayat²,

Thondapu³

et al. 2015

Coronary Artery Disease

View full text Add to dashboard Cite

Percutaneous coronary intervention (PCI) has shown a high success rate in the treatment of coronary artery disease. The decision to perform PCI often relies on the cardiologist's visual interpretation of coronary lesions during angiography. This has inherent limitations, particularly due to the low resolution and two-dimensional nature of angiography. State-of-the-art modalities such as three-dimensional quantitative coronary angiography, optical coherence tomography and invasive fractional flow reserve (FFR) may improve clinicians' understanding of both the anatomical and physiological importance of coronary lesions. While invasive FFR is the gold standard technique for assessment of the haemodynamic significance of coronary lesions, recent studies have explored a surrogate for FFR derived solely from three-dimensional reconstruction of the invasive angiogram, and therefore eliminating need for a pressure wire. Utilizing advanced computational fluid dynamics research, this virtual fractional flow reserve (vFFR) has demonstrated reasonable correlation with invasive measurements and remains an intense area of ongoing study. However, at present, several limitations and computational fluid dynamic assumptions may preclude vFFR from widespread clinical use. This review demonstrates the tight integration of advanced three-dimensional imaging techniques and vFFR in assessing coronary artery disease, reviews the advantages and disadvantages of such techniques and attempts to provide a glimpse of how such advances may benefit future clinical decision-making during PCI.

show abstract

Section: Future Of Virtual Fractional Flow Reservementioning

confidence: 99%

Advances in three-dimensional coronary imaging and computational fluid dynamics

Poon

Hayat²,

Thondapu³

et al. 2015

Coronary Artery Disease

View full text Add to dashboard Cite

show abstract

“…where +1 denotes the second copy of f i data for each nonsolid node used in [19] to avoid race conditions. For geometries with a low number of solid nodes, the left term from Eqn.…”

Section: Fluid Index Arraymentioning

confidence: 99%

“…Fast LBM implementations fuse collision and streaming steps into a single kernel. Unfortunately, the original implementation from [19] uses two separate kernels. The first kernel, launched only for fluid nodes, reads and writes f i values and realizes the collision step.…”

Section: Fluid Index Arraymentioning

confidence: 99%

“…After this optimization, the real number of memory transactions for each completely filled tile is very close to the minimum value defined by Eqn. (19), due to increased cache utilization. The address calculations are performed by using constexpr methods.…”

Section: Implementation Detailsmentioning

confidence: 99%

“…The simulations were run for dense (fluid flow in a square chamber with a moving lid) and sparse 2D and 3D geometries. The sparse geometries for the D3Q19 lattice arrangement were prepared on the basis of similar cases, as presented in [18] (three arrays of randomly arranged spheres with a diameter of 40 lattice units and blood flow in a cerebral aneurysm) and [19] (aorta with coarctation). We also performed some tests on arrays with randomly arranged spheres with porosities lower than those used in [18].…”

Section: Test Geometriesmentioning

confidence: 99%

See 2 more Smart Citations

Sparse Geometries Handling in Lattice Boltzmann Method Implementation for Graphic Processors

Tomczak

Szafran

2018

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

We describe a high-performance implementation of the lattice Boltzmann method (LBM) for sparse geometries on graphic processors. In our implementation we cover the whole geometry with a uniform mesh of small tiles and carry out calculations for each tile independently with proper data synchronization at the tile edges. For this method, we provide both a theoretical analysis of complexity and the results for real implementations involving two-dimensional (2D) and three-dimensional (3D) geometries. Based on the theoretical model, we show that tiles offer significantly smaller bandwidth overheads than solutions based on indirect addressing. For 2D lattice arrangements, a reduction in memory usage is also possible, although at the cost of diminished performance. We achieved a performance of 682 MLUPS on GTX Titan (72% of peak theoretical memory bandwidth) for the D3Q19 lattice arrangement and double-precision data.In the LBM, as in conventional CFD, the geometry, initial and boundary conditions must be specified to solve the initial value problem. The computational domain is uniformly partitioned, with computational nodes placed in vertices of adjacent cells, which become the lattice. One of the lattice structures used in this study, D2Q9, is presented in Fig. 1. The first number in the notation (D2) is the space dimension, while the second (Q9) is the lattice links number.Let f i represent the probability distribution function (PDF) along the i lattice direction: in which case, δ x and δ t are the lattice spacing and the lattice time step, respectively; This is the author's version of an article that has been published in this journal. Changes were made to this version by the publisher prior to publication.The final version of record is available at http://dx.

show abstract