GPU-based Approach Research Articles

Graphics-processing-unit-accelerated ice flow solver for unstructured meshes using the Shallow-Shelf Approximation (FastIceFlo v1.0.1)

Abstract. Ice-sheet flow models capable of accurately projecting their future mass balance constitute tools to improve flood risk assessment and assist sea-level rise mitigation associated with enhanced ice discharge. Some processes that need to be captured, such as grounding-line migration, require high spatial resolution (under the kilometer scale). Conventional ice flow models mainly execute on central processing units (CPUs), which feature limited parallel processing capabilities and peak memory bandwidth. This may hinder model scalability and result in long run times, requiring significant computational resources. As an alternative, graphics processing units (GPUs) are ideally suited for high spatial resolution, as the calculations can be performed concurrently by thousands of threads, processing most of the computational domain simultaneously. In this study, we combine a GPU-based approach with the pseudo-transient (PT) method, an accelerated iterative and matrix-free solution strategy, and investigate its performance for finite elements and unstructured meshes with application to two-dimensional (2-D) models of real glaciers at a regional scale. For both the Jakobshavn and Pine Island glacier models, the number of nonlinear PT iterations required to converge a given number of vertices (N) scales in the order of 𝒪(N1.2) or better. We further compare the performance of the PT CUDA C implementation with a standard finite-element CPU-based implementation using the price-to-performance metric. The price of a single Tesla V100 GPU is 1.5 times that of two Intel Xeon Gold 6140 CPUs. We expect a minimum speedup of at least 1.5 times to justify the Tesla V100 GPU price to performance. Our developments result in a GPU-based implementation that achieves this goal with a speedup beyond 1.5 times. This study represents a first step toward leveraging GPU processing power, enabling more accurate polar ice discharge predictions. The insights gained will benefit efforts to diminish spatial resolution constraints at higher computing performance. The higher computing performance will allow for ensembles of ice-sheet flow simulations to be run at the continental scale and higher resolution, a previously challenging task. The advances will further enable the quantification of model sensitivity to changes in upcoming climate forcings. These findings will significantly benefit process-oriented sea-level-projection studies over the coming decades.

GPU-accelerated polyenergetic forward projection for 9 MeV industrial CT system

Polyenergetic forward projection has great significance in inspecting hazardous materials, establishing optimal radiographic variables and investigating beam hardening effects. However, it is computationally intensive to perform polyenergetic forward-projection calculations for high-resolution phantoms. To address this issue, a rapid polyenergetic forward-projection algorithm is proposed for a 9 MeV industrial computed tomography (CT) system. The FLUktuierende KAskade (FLUKA) software package is used to generate the 9 MeV X-ray spectrum data. Two voxelised phantoms are used to model scanned objects, one being a multi-material cylinder and the other a single-material turbine blade. An incremental version of Siddon's algorithm is adopted to calculate the intersection lengths between the X-rays and the auxiliary phantoms. Three strategies are utilised to accelerate the calculation, in which: the intersection lengths do not vary with the energy bins and can be used repeatedly until all the energy bins are counted; a graphics processing unit (GPU) is used to accelerate the ray tracing algorithm by utilising a parallel computing technique; and faster memory access is achieved by binding the auxiliary phantoms to texture objects. The simulation results in this paper show that the GPU-based approach not only maintains the image precision but also gains significant speed-ups over the conventional central processing unit (CPU)-based Siddon method. Furthermore, beam hardening artefacts can clearly be seen from the profile curves of the reconstructed slices, indicating that this method is effective.

GPU-based similarity metrics computation and machine learning approaches for string similarity evaluation in large datasets

AbstractThe digital era brings up on one hand massive amounts of available data and on the other hand the need of parallel computing architectures for efficient data processing. String similarity evaluation is a processing task applied on large data volumes, commonly performed by various applications such as search engines, biomedical data analysis and even software tools for defending against viruses, spyware, or spam. String similarities are also used in musical industry for matching playlist records with repertory records composed of song titles, performer artists and producers names, aiming to assure copyright protection of mass-media broadcast materials. The present paper proposes a novel GPU-based approach for parallel implementation of the Jaro–Winkler string similarity metric computation, broadly used for matching strings over large datasets. The proposed implementation is applied in musical industry for matching playlist with over 100k records with a given repertory which includes a collection of over 1 million right owner records. The global GPU RAM memory is used to store multiple string lines representing repertory records, whereas single playlist string comparisons with the raw data are performed using the maximum number of available GPU threads and the stride operations. Further on, the accuracy of the Jaro–Winkler approach for the string matching procedure is increased using both an adaptive neural network approach guided by a novelty detection classifier (aNN) and a multiple-features neural network implementation (MF-NN). Thus, the aNN approach yielded an accuracy of 92% while the MF-NN approach achieved an accuracy of 99% at the cost of increased computational complexity. Timing considerations and the computational complexity are detailed for the proposed approaches compared with both the general-purpose processor (CPU) implementation and the state-of-the-art GPU approaches. A speed-up factor of 21.6 was obtained for the GPU-based Jaro–Winkler implementation compared with the CPU one, whereas a factor of 3.72 was obtained compared with the existing GPU implementation of string matching procedure based on Levenstein distance metrics.

A GPU-Based Genetic Algorithm for the Multiple Allocation P-Hub Median Problem

As the sizes of realistic hub location problems increase as time goes on (reaching thousands of nodes currently) this makes such problems difficult to solve in a reasonable time using conventional computers. This study aims to show that such problems may be solved in a short computing time and with high-quality solutions using the computational power of the GPU (actually available in most personal computers). So, we present a GPU-based approach for the uncapacitated multiple allocations p-hub median problems. Our method identifies the nodes that are likely to be hubs in the optimal solution and improves them via a parallel genetic algorithm. The obtained GPU implementation reached within seconds the optimal or the best solutions for all the known benchmarks we had access to and solved larger instances up to 6000 nodes so far unsolved. Compared to this study, no other article dealing with hub location problems has presented results for instances as large.

Large Scale Data Using K-Means

Because of the exponential growth of high-layered datasets, conventional database querying strategies are inadequate for extracting useful information, and analysts must now devise novel techniques to meet these demands. Such massive articulation data results in a plethora of new computational triggers as a result of both the rise in data protests and the increase of elements/ascribes. Preprocessing the data with a reliable dimensionality reduction method improves the efficacy and precision of mining operations on densely layered data. Therefore, we have compiled the opinions of numerous academics. Cluster analysis is a data analysis tool that has recently acquired prominence in a number of different disciplines. K-means, a common parceling-based clustering algorithm, looks for a fixed number of clusters that can be identified using only their centroids. However, the outcomes depend heavily on the starting points of the clusters' focuses. Again, there is a dramatic rise in the number of distance calculations with increasing data complexity. This is due to the fact that assembling a detailed model typically calls for a substantial and distributed amount of preliminary data. There may be a substantial time commitment involved in preparing a broad collection of ingredients. For huge data sets in particular, there is a cost/benefit analysis to consider when deciding how to create orders: speed vs. accuracy. The k-means method is commonly used to compress and sum vector data, as well as cluster it. For precautious k-means (ASB K-means), we present No Concurrent Specific Clumped K-means, a fast and memory-effective GPU-based method. Our method can be adjusted to use much less GPU RAM than the size of the full dataset, which is a significant improvement over earlier GPU-based k-means methods. Datasets that are too large to fit in RAM may be clustered. The approach uses a clustered architecture and applies the triangle disparity in each k-means focus to remove a data point if its enrollment task or cluster it belongs to remains unaltered, allowing it to efficiently handle big datasets. This reduces the number of data guides that must be transferred between the CPU's Slam and the GPU's global memory.

Fast parallel algorithms for finding elementary circuits of a directed graph: a GPU-based approach

Circuits in a graph are interesting structures and identifying them is of an important relevance for many applications. However, enumerating circuits is known to be a difficult problem, since their number can grow exponentially. In this paper, we propose fast parallel approaches for enumerating elementary circuits of directed graphs based on graphics processing unit (GPU). Our algorithms are based on a massive exploration of the graph in a breadth-first search strategy. Algorithm V-FEC explores the graph starting from different vertices simultaneously. To further reduce the search space, we present T-FEC, another algorithm that uses triplets as an initial set to start exploring. To the best of our knowledge, those are the first parallel GPU-based algorithms for finding all circuits of a given graph. In addition, they find circuits of a given length and circuits with a specific vertex or edge. The evaluation results show that the proposed approaches achieve up to 190x speed-up over Johnson’s algorithm, one of the most efficient sequential algorithms for finding circuits.

GPU-based approach to large scale dynamic vehicle routing problem

Vehicle routing problems (VRPs) are fundamental optimisation problems of transportation systems. In the real-world, VRPs are dynamic in the sense that new customers' requests continuously arrive over time, after a number of vehicles have already started their tours. Dynamic VRPs (DVRPs) require making decisions as fast as possible. This needs resolution methods with high computational efficiency especially for problems with a large number of customers. The aim of this paper is to attempt to achieve this objective. For this, we design a genetic algorithm for the DVRP and we implement it on GPU. The proposed approach inserts new requests into already planned routes then it optimises the resulting solution via genetic operators. To our knowledge, this is the first attempt to solve large DVRP on the GPU using evolutionary algorithm and seems to be efficient according to the experimental results on some published benchmarks and on our large instances (up to 10,000 nodes).

A Novel GPU-Based Approach to Exploit Time-Respectingness in Public Transport Networks for Efficient Computation of Earliest Arrival Time

In static temporal networks, the Earliest Arrival Time (EAT) problem is to calculate the earliest possible time of arrival at a set of vertices from a given source vertex. Applications of the EAT problem include designing efficient evacuation planning in dynamic scenarios, optimal journey planning in transport networks, and optimal flow management in supply chains. There exist several solutions for the EAT problem in the literature, however, there is limited work on GPU-based solutions to leverage the capabilities of the high throughput accelerator for better performance. Further, there is also a need for more efficient methods to process the inherent earliest arrival dependencies in a transport network. In this paper, we propose GPU algorithms for the one-to-all Earliest Arrival Time problem in public transport networks. Its key characteristic is that it is very fast for the best-case networks where all temporal paths are <i>time-respecting</i>. We propose our algorithms in five incremental ways, where the subsequent approaches are the improved version of previous ones. The Selective-check-version is the most improved approach and hence, the key algorithm. It uses shared memory for efficient computations. For the Selective-check version, we observed an average speedup of 6.45 against the Serial Connection-scan algorithm and 2.77 w.r.t. the Edge-version algorithm.

Improving Similarity-Based Retrieval Efficiency by Using Graphic Processing Units in Case-Based Reasoning

The accelerated growth of available data causes case bases of increasing sizes and thus lowers efficiency during the case retrieval phase in Case-Based Reasoning (CBR) systems. Even though, many complex and data-intensive tasks are solved by using Graphic Processing Units (GPUs), its application in CBR research has yet to advance past the early stage phase. In this paper, we present an approach to use CUDA-compatible GPUs for similarity assessment of structural, feature vector based cases. Our approach supports several syntactic and semantic similarity measures and is implemented in the open-source case-based reasoning framework ProCAKE. When comparing to current retrieval techniques that calculate similarities on the CPU, our GPU-based approach outperforms them by a factor of up to 37. In addition, our evaluation indicates that the performance gains increase with higher case complexity.

The Design and Development of the Destiny-Class CyberCANOE Hybrid Reality Environment

The Destiny-class CyberCANOE (Destiny) is a Hybrid Reality environment that provides 20/20 visual acuity in a 13-foot-wide, 320-degree cylindrical structure comprised of tiled passive stereo-capable organic light emitting diode (OLED) displays. Hybrid Reality systems combine surround-screen virtual reality environments with ultra-high-resolution digital project-rooms. They are intended as collaborative environments that enable multiple users to work minimally encumbered for long periods of time in rooms surrounded by data in the form of visualizations that benefit from being displayed at resolutions matching visual acuity and/or in stereoscopic 3D. Destiny is unique in that it is the first Hybrid Reality system to use OLED displays and it uses a real-time GPU-based approach for minimizing stereoscopic crosstalk. This paper chronicles the non-trivial engineering research and attention-to-detail that is required to develop a production quality hybrid-reality environment by providing details about Destiny’s design and construction process. This detailed account of how a Hybrid Reality system is designed and constructed from the ground up will help VR researchers and developers understand the engineering complexity of developing such systems. This paper also discusses a GPU-based crosstalk mitigation technique and evaluation, and the use of Microsoft’s augmented reality headset, the HoloLens, as a design and training aid during construction.

A GPU-based Approach for Path Planning Optimization via Travel Length Reduction

Typically, before constructing an object with an additive manufacturing system, the 3D object must be sent through a process called slicing. Slicing converts a 3D object commonly in the form of an STL file into a set of layers by horizontally intersecting a plane with the object at various heights. At each height, called a layer, multiple 2D polygons can be generated. Each polygon represents a boundary for solid geometry and is called an island. Each island is then comprised of multiple path types in an attempt to optimally fill the polygon. To move between each island and each islands’ paths, travels are inserted. Travels are simply motion by the system to move from one area of construction to another. Travels do not contribute to the construction of the object, and so, are considered wasted motion. In large-scale additive manufacturing, objects can be quite large and the distance between islands can be large as well. As a result, these travels can waste a significant amount of time. Ideally, travels would be as short as possible, however, computing global minimal travel paths is computationally expensive. To combat this problem, researchers at Oak Ridge National Lab developed a GPU-based approach to travel insertion based on a unique factoradic representation. This representation was then utilized by the GPU to solve the Traveling Salesman Problem (TSP). This algorithm was able to compute global minimal travel paths quickly resulting in faster object construction. A general investigation was also carried out to determine when a GPU vs CPU implementation would be beneficial.

GPU-Based approach to Large scale Dynamic Vehicle Routing Problem.

Vehicle routing problems (VRPs) are fundamental optimisation problems of transportation systems. In the real-world, VRPs are dynamic in the sense that new customers' requests continuously arrive over time, after a number of vehicles have already started their tours. Dynamic VRPs (DVRPs) require making decisions as fast as possible. This needs resolution methods with high computational efficiency especially for problems with a large number of customers. The aim of this paper is to attempt to achieve this objective. For this, we design a genetic algorithm for the DVRP and we implement it on GPU. The proposed approach inserts new requests into already planned routes then it optimises the resulting solution via genetic operators. To our knowledge, this is the first attempt to solve large DVRP on the GPU using evolutionary algorithm and seems to be efficient according to the experimental results on some published benchmarks and on our large instances (up to 10,000 nodes).

Graph Reachability on Parallel Many-Core Architectures

Many modern applications are modeled using graphs of some kind. Given a graph, reachability, that is, discovering whether there is a path between two given nodes, is a fundamental problem as well as one of the most important steps of many other algorithms. The rapid accumulation of very large graphs (up to tens of millions of vertices and edges) from a diversity of disciplines demand efficient and scalable solutions to the reachability problem. General-purpose computing has been successfully used on Graphics Processing Units (GPUs) to parallelize algorithms that present a high degree of regularity. In this paper, we extend the applicability of GPU processing to graph-based manipulation, by re-designing a simple but efficient state-of-the-art graph-labeling method, namely the GRAIL (Graph Reachability Indexing via RAndomized Interval) algorithm, to many-core CUDA-based GPUs. This algorithm firstly generates a label for each vertex of the graph, then it exploits these labels to answer reachability queries. Unfortunately, the original algorithm executes a sequence of depth-first visits which are intrinsically recursive and cannot be efficiently implemented on parallel systems. For that reason, we design an alternative approach in which a sequence of breadth-first visits substitute the original depth-first traversal to generate the labeling, and in which a high number of concurrent visits is exploited during query evaluation. The paper describes our strategy to re-design these steps, the difficulties we encountered to implement them, and the solutions adopted to overcome the main inefficiencies. To prove the validity of our approach, we compare (in terms of time and memory requirements) our GPU-based approach with the original sequential CPU-based tool. Finally, we report some hints on how to conduct further research in the area.

Blocking Self-Avoiding Walks Stops Cyber-Epidemics: A Scalable GPU-Based Approach

Cyber-epidemics, the widespread of fake news or propaganda through social media, can cause devastating economic and political consequences. A common countermeasure against cyber-epidemics is to disable a small subset of suspected social connections or accounts to effectively contain the epidemics. An example is the recent shutdown of 125,000 ISIS-related Twitter accounts. Despite many proposed methods to identify such a subset, none are scalable enough to provide high-quality solutions in nowadays’ billion-size networks. To this end, we investigate the Spread Interdiction problems that seek the most effective links (or nodes) for removal under the well-known Linear Threshold model. We propose novel CPU-GPU methods that scale to networks with billions of edges , yet possess rigorous theoretical guarantee on the solution quality. At the core of our methods is an $O(1)$ O ( 1 ) -space out-of-core algorithm to generate a new type of random walks, called Hitting Self-avoiding Walks ( $\mathsf{HSAW}$ HSAW s). Such a low memory requirement enables handling of big networks and, more importantly, hiding latency via scheduling of millions of threads on GPUs. Comprehensive experiments on real-world networks show that our algorithms provide much higher quality solutions and are several orders of magnitude faster than the state-of-the art. Comparing to the (single-core) CPU counterpart, our GPU implementations achieve significant speedup factors up to 177x on a single GPU and 338x on a GPU pair.

When the Decomposition Meets the Constraint Satisfaction Problem

This paper explores the joint use of decomposition methods and parallel computing for solving constraint satisfaction problems and introduces a framework called Parallel Decomposition for Constraint Satisfaction Problems (PD-CSP). The main idea is that the set of constraints are first clustered using a decomposition algorithm in which highly correlated constraints are grouped together. Next, parallel search of variables is performed on the produced clusters in a way that is friendly for parallel computing. In particular, for the first step, we propose the adaptation of two well-known clustering algorithms (k-means and DBSCAN). For the second step, we develop a GPU-based approach to efficiently explore the clusters. The results from the extensive experimental evaluation show that the PD-CSP provides competitive results in terms of accuracy and runtime.

Constructions Mechanical and Thermal Conditions Modeling with GPU-based Approach

At present, there are two parallel trends related to spacecraft development tasks and problems. From the one side, due to complexity of constructions and growing use of high-power large-sized antennas, it is necessary to consider the effect of self-shading and research the effects accompanying this phenomenon. From the other side, modern computer tools, especially GPUs (graphics processing unit), allow one to create “mini-supercomputer” based on “common” desktop/laptop and perform CAE (computer-aided engineering)-based processes using 3D models of different LOD (Level-of-detail). In this paper, we present a conception of SA parameters computer modeling based on hierarchical, property-based 3D-model of spacecraft. Thus, one can model different work modes of spacecraft on the project stage, with real-time prediction of effects related to space factors complex combination impact. The result is easy-to-edit, easy-to-calibrate “digital twin” of spacecraft which allows engineer to model and investigate complex space factors combination results for a different orbital work mode..

Level 2 Reformulation Linearization Technique–Based Parallel Algorithms for Solving Large Quadratic Assignment Problems on Graphics Processing Unit Clusters

This paper discusses efficient parallel algorithms for obtaining strong lower bounds and exact solutions for large instances of the quadratic assignment problem (QAP). Our parallel architecture is comprised of both multicore processors and compute unified device architecture–enabled NVIDIA graphics processing units (GPUs) on the Blue Waters Supercomputing Facility at the University of Illinois at Urbana–Champaign. We propose novel parallelization of the Lagrangian dual ascent algorithm on the GPUs, which is used for solving a QAP formulation based on the level-2 reformulation linearization technique. The linear assignment subproblems in this procedure are solved using our accelerated Hungarian algorithm [Date K, Rakesh N (2016) GPU-accelerated Hungarian algorithms for the linear assignment problem. Parallel Computing 57:52–72.]. We embed this accelerated dual-ascent algorithm in a parallel branch-and-bound scheme and conduct extensive computational experiments on single and multiple GPUs, using problem instances with up to 42 facilities from the quadratic assignment problem library (QAPLIB). The experiments suggest that our GPU-based approach is scalable, and it can be used to obtain tight lower bounds on large QAP instances. Our accelerated branch-and-bound scheme is able to comfortably solve Nugent and Taillard instances (up to 30 facilities) from the QAPLIB, using a modest number of GPUs.

A GPU Based Approach for Solving the Workflow Scheduling Problem

Cloud computing is considered a new way to use on-demand computing resources. When executing a workflow process in such an environment, task scheduling, a well-known NP-hard problem is a very important step. Many heuristic algorithms have been proposed to solve this problem. In this article, the authors present a GPU-based approach for solving the workflow scheduling problem. The main idea of the approach is to implement a massively parallel version of the simulated annealing algorithm, in an asynchronous way where no information is exchanged among parallel runs. The proposed approach, called PSA algorithm, is against another well-established scheduling HEFT heuristic. Experiments with randomly generated graphs show a much better performance from the proposed approach.

A GPU-Based Automatic Approach for Guide Wire Tracking in Fluoroscopic Sequences

Guide wire tracking in fluoroscopic images has done a significant task in assisting the physicians during radiology-aided interventions. Many groups have tried to detect the guide wire from the fluoroscopic images based on the image properties. The main challenge is that manual intervention is required during the detection. Other groups try to introduce localizers to track guide wires during intervention, which requires additional hardware equipment, and may intervene with the traditional clinical routines. Machine learning methods are also exploited. Although such methods may provide accurate tracking, they often require large amount of data and training time. In this paper, we propose a GPU-based fast and automatic approach to track guide wires in fluoroscopic sequences. We propose a multi-scale filtering and gradient vector field-based real-time tracking method for guide wire tracking from fluoroscopic images. To improve calculation efficiency and meet real-time application requirement, we propose a GPU-based acceleration scheme, and also a Bayesian filter-like motion tracking method to limit the guide wire tracking to a smaller range to improve calculation efficiency. We test our proposed method on two test data sets of fluoroscopic sequences of 102 frames and 72 frames. We achieve an average guide wire detection rate of 96.7%, a false detection rate of 0.0011% and an error distance measure of 0.83 pixels for the first sequence, and 98.8%, 0.000069% and 0.85 pixels, respectively, for the second sequence. With the proposed acceleration method, we finish calculation for the first sequence in nine seconds, thus, efficiency is enhanced by 100 times with the unaccelerated algorithm.

Hybrid Lighting for faster rendering of scenes with many lights

There is growing interest in rendering scenes with many lights, where scenes typically contain hundreds to thousands of lights. Each light illuminates geometry within a finite extent called a light volume. A key aspect of performance is determining which lights apply to what geometry, and then applying those lights efficiently. We present a GPU-based approach using spatial data structures, binning lights by depth analytically while also taking advantage of hardware rasterization. This improves light binning performance by 3–6 $$\times $$ . We also present a GPU memory and cache friendly data structure that takes two passes to build, giving 4–10 $$\times $$ improved performance when applying lighting and an overall improvement of 1.3–4 $$\times $$ for total frametime.