CUDA Programming Model Research Articles

3D Visualisation of Tumour-Induced Angiogenesis Using the CUDA Programming Model and OpenGL Interoperability

In this paper we solve a complex discrete-continuous model of tumour-induced angiogenesis using an explicit time-stepping FDM and simultaneously simulate the model dynamics in 3D. The interoperability between the CUDA programming model and the graphics hardware through OpenGL allows us to generate dynamic interactive 3D realistic visualisations. We use CUDA for the complex parallel calculations and deploy OpenGL for on-the-fly 3D visualisation of the numerical simulations. Clearly, being able to link the numerical results of complex mathematical models to interactive 3D visualisations that can literally update instantaneously to varying model parameters, should provide an invaluable tool for clinical physicians and research scientists. We also give an overview of current medical imaging techniques for studying microcirculatory and blood flow dynamics at the cellular level and indicate how the results presented here could offer potential for future developments in this area.

A Hybrid Continuous-Discrete Model of Tumour-Induced Angiogenesis is Solved Numerically in Parallel and Performance Improvements Analysed

The main aim of this paper is to investigate the potential performance improvements gained from a serial versus parallel implementation of the numerical solution to a system of coupled nonlinear PDEs describing tumour-induced angiogenesis. After applying a suitable finite difference scheme, the resulting hybrid continuous discreet model is solved based on a set of cellular rules defining endothelial cell movement towards a tumour. In addition, the model explicitly incorporates the processes of branching, anastomosis and cell proliferation. Parallel implementations are based on the CUDA programming model with a detailed look at efficient thread deployment and memory management. Results show substantial speedups for the CUDA C language against that of conventional high performance languages, such as C++. Such increased performance highlights the potential for simulating more complex mathematical models of tumour dynamics, such as vascularisation networks, tumour invasion and metastasis, leading to the potential for more rapid experimental results for a range of complex cancer models.

Towards real-time detection of seizures in awake rats with GPU-accelerated diffuse optical tomography.

Advancement in clinically relevant studies like seizure interruption using functional neuro imaging tools has shown that specific changes in hemodynamics precede and accompany seizure onset and propagation. However, preclinical seizure experiments need to be conducted in awake animals with images reconstructed and displayed in real-time. This article describes an approach that can be utilized to tackle these challenges. A subject specific head interface and restraining method was designed to allow for DOT to imaging of hemodynamic changes in unanesthetized rats during evoked acute seizures. Using CUDA programming model, the finite-element based nonlinear iterative algorithm for image reconstruction was parallelized. Early hemodynamic changes were monitored in real time and observed tens of seconds prior to seizure onset. Utilizing the massive parallelization offered by graphic processing units (GPU), DOT was extended to online image reconstruction within 1s. Pre-seizure state related hemodynamic changes were detected in awake rats. 3D monitoring of hemodynamic changes was performed in real time with our parallelized image reconstruction procedure. Diffuse optical tomography (DOT) is a promising neuroimaging tool for the investigation of seizures in awake animals.

Research on SIFT Matching Algorithm Based on GPU

Aiming at the shortage of SIFT algorithm in time comsumption and getting less matching points. On the one hand, the paper improves the original SIFT algorithm, it proposes regional growth algorithm based on SIFT, so you can get many matching points which are good for generating the disparity map; on the other hand, the paper uses the CPU and GPU heterogeneous platforms and analyses the CUDA programming model and memory model. This paper analyses the algorithm in detail, so the algorithm can be carried out on CUDA. Experimental results show that, compared with the original algorithm, the algorithm is about 10 times faster, and generate good disparity map.

Massive Parallelism With Gpus for Centrality Ranking in Complex Networks

Many problems in Computer Science can be modelled using graphs. Evaluating node centrality in complex networks, which can be considered equivalent to undirected graphs, provides an useful metric of the relative importance of each node inside the evaluated network. The knowledge on which the most central nodes are, has various applications, such as improving information spreading in diffusion networks. In this case, most central nodes can be considered to have higher influence rates over other nodes in the network. The main purpose in this work is developing a GPU based and massively parallel application so as to evaluate the node centrality in complex networks using the Nvidia CUDA programming model. The main contribution of this work is the strategies for the development of an algorithm to evaluate the node centrality in complex networks using Nvidia CUDA parallel programming model. We show that the

Image Processing on GPU: Application of Integral Image

In this paper we present an integral image algorithm that can run in real-time on a Graphics Processing Unit (GPU). Our system exploits the parallelisms in computation via the NVIDA CUDA programming model, which is a software platform for solving non-graphics problems in a massively parallel high performance fashion. We compare the performance of the parallel approach running on the GPU with the sequential CPU implementation across a range of image sizes

Solving the Caputo Fractional Reaction-Diffusion Equation on GPU

We present a parallel GPU solution of the Caputo fractional reaction-diffusion equation in one spatial dimension with explicit finite difference approximation. The parallel solution, which is implemented with CUDA programming model, consists of three procedures: preprocessing, parallel solver, and postprocessing. The parallel solver involves the parallel tridiagonal matrix vector multiplication, vector-vector addition, and constant vector multiplication. The most time consuming loop of vector-vector addition and constant vector multiplication is optimized and impressive performance improvement is got. The experimental results show that the GPU solution compares well with the exact solution. The optimized GPU solution on NVIDIA Quadro FX 5800 is 2.26 times faster than the optimized parallel CPU solution on multicore Intel Xeon E5540 CPU.

Применение технологий CUDA и MPI к решению задач молекулярной динамики

Применение технологий CUDA и MPI к решению задач молекулярной динамики

Padding Free Bank Conflict Resolution for CUDA-Based Matrix Transpose Algorithm

The advances of Graphic Processing Units (GPU) technology and the introduction of CUDA programming model facilitates developing new solutions for sparse and dense linear algebra solvers. Matrix Transpose is an important linear algebra procedure that has deep impact in various computational science and engineering applications. Several factors hinder the expected performance of large matrix transpose on GPU devices. The degradation in performance involves the memory access pattern such as coalesced access in the global memory and bank conflict in the shared memory of streaming multiprocessors within the GPU. In this paper, two matrix transpose algorithms are proposed to alleviate the aforementioned issues of ensuring coalesced access and conflict free bank access. The proposed algorithms have comparable execution times with the NVIDIA SDK bank conflict free matrix transpose implementation. The main advantage of proposed algorithms is that they eliminate bank conflicts while allocating shared memory exactly equal to the tile size (T x T) of the problem space. However, to the best of our knowledge an extra space of Tx(T+1) needs to be allocated in the published research. We have also applied the proposed transpose algorithm to recursive gaussian implementation of NVIDIA SDK and achieved about 6% improvement in performance.

Development of a convex polyhedral discrete element simulation framework for NVIDIA Kepler based GPUs

Understanding the dynamical behavior of Granular Media (GM) is extremely important to many industrial processes. Thus simulating the dynamics of GM is critical in the design and optimization of such processes. However, the dynamics of GM is complex in nature and cannot be described by a closed form solution for more than a few particles. A popular and successful approach in simulating the underlying dynamics of GM is by using the Discrete Element Method (DEM). Computational viable simulations are typically restricted to a few particles with realistic complex interactions or a larger number of particles with simplified interactions. This paper introduces a novel DEM based particle simulation code (BLAZE-DEM) that is capable of simulating millions of particles on a desktop computer utilizing a NVIDIA Kepler Graphical Processor Unit (GPU) via the CUDA programming model. The GPU framework of BLAZE-DEM is limited to applications that require large numbers of particles with simplified interactions such as hopper flow which exhibits task level parallelism that can be exploited on the GPU. BLAZE-DEM also performs real-time visualization with interactive capabilities. In this paper we discuss our GPU framework and validate our code by comparison between experimental and numerical hopper flow.

All-pairs Shortest Path Algorithm based on MPI+CUDA Distributed Parallel Programming Model

In view of the problem that computing shortest paths in a graph is a complex and time-consuming process, and the traditional algorithm that rely on the CPU as computing unit solely can't meet the demand of real-time processing, in this paper, we present an all-pairs shortest paths algorithm using MPI+CUDA hybrid programming model, which can take use of the overwhelming computing power of the GPU cluster to speed up the processing. This proposed algorithm can combine the advantages of MPI and CUDA programming model, and can realize two-level parallel computing. In the cluster-level, we take use of the MPI programming model to achieve a coarse-grained parallel computing between the computational nodes of the GPU cluster. In the node-level, we take use of the CUDA programming model to achieve a GPU-accelerated fine grit parallel computing in each computational node internal. The experimental results show that the MPI+CUDA-based parallel algorithm can take full advantage of the powerful computing capability of the GPU cluster, and can achieve about hundreds of time speedup; The whole algorithm has good computing performance, reliability and scalability, and it is able to meet the demand of real-time processing of massive spatial shortest path analysis

GHull

A novel algorithm is presented to compute the convex hull of a point set in ℝ 3 using the graphics processing unit (GPU). By exploiting the relationship between the Voronoi diagram and the convex hull, the algorithm derives the approximation of the convex hull from the former. The other extreme vertices of the convex hull are then found by using a two-round checking in the digital and the continuous space successively. The algorithm does not need explicit locking or any other concurrency control mechanism, thus it can maximize the parallelism available on the modern GPU. The implementation using the CUDA programming model on NVIDIA GPUs is exact and efficient. The experiments show that it is up to an order of magnitude faster than other sequential convex hull implementations running on the CPU for inputs of millions of points. The works demonstrate that the GPU can be used to solve nontrivial computational geometry problems with significant performance benefit.

Memory performance estimation of CUDA programs

CUDA has successfully popularized GPU computing, and GPGPU applications are now used in various embedded systems. The CUDA programming model provides a simple interface to program on GPUs, but tuning GPGPU applications for high performance is still quite challenging. Programmers need to consider numerous architectural details, and small changes in source code, especially on the memory access pattern, can affect performance significantly. This makes it very difficult to optimize CUDA programs. This article presents CuMAPz, which is a tool to analyze and compare the memory performance of CUDA programs. CuMAPz can help programmers explore different ways of using shared and global memories, and optimize their program for efficient memory behavior. CuMAPz models several memory-performance-related factors: data reuse, global memory access coalescing, global memory latency hiding, shared memory bank conflict, channel skew, and branch divergence. Experimental results show that CuMAPz can accurately estimate performance with correlation coefficient of 0.96. By using CuMAPz to explore the memory access design space, we could improve the performance of our benchmarks by 30% more than the previous approach [Hong and Kim 2010].

Efficient compilation of CUDA kernels for high-performance computing on FPGAs

The rise of multicore architectures across all computing domains has opened the door to heterogeneous multiprocessors, where processors of different compute characteristics can be combined to effectively boost the performance per watt of different application kernels. GPUs, in particular, are becoming very popular for speeding up compute-intensive kernels of scientific, imaging, and simulation applications. New programming models that facilitate parallel processing on heterogeneous systems containing GPUs are spreading rapidly in the computing community. By leveraging these investments, the developers of other accelerators have an opportunity to significantly reduce the programming effort by supporting those accelerator models already gaining popularity. In this work, we adapt one such language, the CUDA programming model, into a new FPGA design flow called FCUDA, which efficiently maps the coarse- and fine-grained parallelism exposed in CUDA onto the reconfigurable fabric. Our CUDA-to-FPGA flow employs AutoPilot, an advanced high-level synthesis tool (available from Xilinx) which enables high-abstraction FPGA programming. FCUDA is based on a source-to-source compilation that transforms the SIMT (Single Instruction, Multiple Thread) CUDA code into task-level parallel C code for AutoPilot. We describe the details of our CUDA-to-FPGA flow and demonstrate the highly competitive performance of the resulting customized FPGA multicore accelerators. To the best of our knowledge, this is the first CUDA-to-FPGA flow to demonstrate the applicability and potential advantage of using the CUDA programming model for high-performance computing in FPGAs.

Parallel evaluation of Pittsburgh rule-based classifiers on GPUs

Individuals from Pittsburgh rule-based classifiers represent a complete solution to the classification problem and each individual is a variable-length set of rules. Therefore, these systems usually demand a high level of computational resources and run-time, which increases as the complexity and the size of the data sets. It is known that this computational cost is mainly due to the recurring evaluation process of the rules and the individuals as rule sets. In this paper we propose a parallel evaluation model of rules and rule sets on GPUs based on the NVIDIA CUDA programming model which significantly allows reducing the run-time and speeding up the algorithm. The results obtained from the experimental study support the great efficiency and high performance of the GPU model, which is scalable to multiple GPU devices. The GPU model achieves a rule interpreter performance of up to 64 billion operations per second and the evaluation of the individuals is speeded up of up to 3.461×when compared to the CPU model. This provides a significant advantage of the GPU model, especially addressing large and complex problems within reasonable time, where the CPU run-time is not acceptable.

Computing 2D Constrained Delaunay Triangulation Using the GPU

We propose the first graphics processing unit (GPU) solution to compute the 2D constrained Delaunay triangulation (CDT) of a planar straight line graph (PSLG) consisting of points and edges. There are many existing CPU algorithms to solve the CDT problem in computational geometry, yet there has been no prior approach to solve this problem efficiently using the parallel computing power of the GPU. For the special case of the CDT problem where the PSLG consists of just points, which is simply the normal Delaunay triangulation (DT) problem, a hybrid approach using the GPU together with the CPU to partially speed up the computation has already been presented in the literature. Our work, on the other hand, accelerates the entire computation on the GPU. Our implementation using the CUDA programming model on NVIDIA GPUs is numerically robust, and runs up to an order of magnitude faster than the best sequential implementations on the CPU. This result is reflected in our experiment with both randomly generated PSLGs and real-world GIS data having millions of points and edges.

Parallel Catmull-Rom Spline Interpolation Algorithm for Image Zooming Based on CUDA

In order to scale video image real-timely, a GPU-aided parallel interpolation algorithm was proposed. Catmull-Rom Spline algorithm for image zooming was reformed into SIMD (Single instruction, multiple data) mode according to CUDA programming model. Re-sampling of each pixel was completed by a GPU thread. Hence, time-consuming re-sampling procedure of the whole zooming process were handled by parallel threads. The proposed algorithm runs hundreds times faster than traditional algorithm in experiments, and the speed is fast enough for scaling video frames real-timely. In addition, this algorithm can be extended to solve many other image processing related problems, such as image denosing and image segmentation.

Parallel multi-objective Ant Programming for classification using GPUs

Classification using Ant Programming is a challenging data mining task which demands a great deal of computational resources when handling data sets of high dimensionality. This paper presents a new parallelization approach of an existing multi-objective Ant Programming model for classification, using GPUs and the NVIDIA CUDA programming model. The computational costs of the different steps of the algorithm are evaluated and it is discussed how best to parallelize them. The features of both the CPU parallel and GPU versions of the algorithm are presented. An experimental study is carried out to evaluate the performance and efficiency of the interpreter of the rules, and reports the execution times and speedups regarding variable population size, complexity of the rules mined and dimensionality of the data sets. Experiments measure the original single-threaded and the new multi-threaded CPU and GPU times with different number of GPU devices. The results are reported in terms of the number of Giga GP operations per second of the interpreter (up to 10 billion GPops/s) and the speedup achieved (up to 834× vs CPU, 212× vs 4-threaded CPU). The proposed GPU model is demonstrated to scale efficiently to larger datasets and to multiple GPU devices, which allows the expansion of its applicability to significantly more complicated data sets, previously unmanageable by the original algorithm in reasonable time.

A GPU implementation of a structural-similarity-based aerial-image classification

There is an increasing need for fast and efficient algorithms for the automatic analysis of remote-sensing images. In this paper we address the implementation of the semantic classification of aerial images with general-purpose graphics-processing units (GPGPUs). We propose the calculation of a local Gabor-based structural texture descriptor and a structural texture similarity metric combined with a nearest-neighbor classifier and image-to-class similarity on CUDA supported graphics-processing units. We first present the algorithm and then describe the GPU implementation and optimization with the CUDA programming model. We then evaluate the results of the algorithm on a dataset of aerial images and present the execution times for the sequential and parallel implementations of the whole algorithm as well as measurements only for the selected steps of the algorithm. We show that the algorithms for the image classification can be effectively implemented on the GPUs. In our case, the presented algorithm is around 39 times faster on the Tesla C1060 unit than on the Core i5 650 CPU, while keeping the same success rate of classification.

Accelerated Surface Measurement Using Wavelength Scanning Interferometer with Compensation of Environmental Noise

The optical interferometry systems are widely used for surface metrology. However, the environmental noise and the data analysis approach can limit the measurement performance during in- process inspection. This paper introduces a wavelength scanning interferometer (WSI) for micro and nano-scale areal surface measurement. The WSI can measure large discontinuous surface profiles without phase ambiguity problems. The proposed WSI is immune to environmental noise using an active servo control system that serves as a phase compensating mechanism. Furthermore, a parallel CUDA programming model is presented as a solution to accelerate the computing analysis. The presented system can be used for on-line or in-process measurement on the shop floor.