Kamil Halbiniak, Norbert Meyer, Krzysztof Rojek · Concurrency and Computation Practice and Experience · 2023
Summary This work explores the performance of single‐ and multi‐GPU computing on state‐of‐the‐art NVIDIA‐ and AMD‐based server‐class hardware using various programming interfaces to accelerate a real‐world scientific application for solidification modeling based on the phase‐field method. The main computations of this memory‐bound application correspond to 20 stencils computed across grid nodes. We investigate the application's scalability for two basic schemes of organizing computation: without and with hiding data transfers behind computation, combined with using either peer‐to‐peer inter‐GPU data transfers through NVIDIA NVLink and AMD Infinity interconnects or communication over the PCIe and main memory. Among the studied programming interfaces is CUDA, HIP, and OpenMP Accelerator Model. While the first two are designed to write the codes for a specific hardware platform, OpenMP enables code portability between NVIDIA and AMD GPUs. The resulting performance is experimentally assessed on computing platforms containing NVIDIA V100 (up to 8 GPUs) and A100 (one GPU), as well as AMD MI210 (one device) and MI250 (up to 8 logical GPUs).
Cited by 7
S.Z. Zhu, Mi Zhao, Jiaxu Shen et al. · SSRN Electronic Journal · 2025
Cited by 0Open access
David Herrero‐Perez, Humberto Martínez Barberá · Applied Sciences · 2025
This work evaluates the computing performance of finite element analysis in structural mechanics using modern multi-GPU systems. We can avoid the usual memory limitations when using one GPU device for many-core computing using multiple GPUs for scientific computing. We use a GPU-awareness MPI approach implementing a suitable smoothed aggregation multigrid for preconditioning an iterative distributed conjugate gradient solver for GPU computing. We evaluate the performance and scalability of different models, problem sizes, and computing resources. We take an efficient multi-core implementation as the reference to assess the computing performance of the numerical results. The numerical results show the advantages and limitations of using distributed many-core architectures to address structural mechanics problems.
Cited by 3Open access
Giordana Florimbi, Himar Fabelo, Emanuele Torti et al. · IEEE Access · 2020
Several causes make brain cancer identification a challenging task for neurosurgeons during the surgical procedure. The surgeons' naked eye sometimes is not enough to accurately delineate the brain tumor location and extension due to its diffuse nature that infiltrates in the surrounding healthy tissue. For this reason, a support system that provides accurate cancer delimitation is essential in order to improve the surgery outcomes and hence the patient's quality of life. The brain cancer detection system developed as part of the “HypErspectraL Imaging Cancer Detection” (HELICoiD) European project meets this requirement exploiting a non-invasive technique suitable for medical diagnosis: the hyperspectral imaging (HSI). A crucial constraint that this system has to satisfy is providing a real-time response in order to not prolong the surgery. The large amount of data that characterizes the hyperspectral images, and the complex elaborations performed by the classification system make the High Performance Computing (HPC) systems essential to provide real-time processing. The most efficient implementation developed in this work, which exploits the Graphic Processing Unit (GPU) technology, is able to classify the biggest image of the database (worst case) in less than three seconds, largely satisfying the real-time constraint set to 1 minute for surgical procedures, becoming a potential solution to implement hyperspectral video processing in the near future.
Cited by 52Open access
H I Alzeini, Sh A Hameed, Mohamed Hadi Habaebi · arXiv (Cornell University) · 2014
The overwhelmingly increasing amount of stored data has spurred researchers seeking different methods in order to optimally take advantage of it which mostly have faced a response time problem as a result of this enormous size of data. Most of solutions have suggested materialization as a favourite solution. However, such a solution cannot attain Real- Time answers anyhow. In this paper we propose a framework illustrating the barriers and suggested solutions in the way of achieving Real-Time OLAP answers that are significantly used in decision support systems and data warehouses.
Cited by 0Open access
Young Jong Mo, Joongheon Kim, Jong‐Kook Kim et al. · 2017
In this paper we measure and verify the performance improvements in deep learning computation under the support of GPU-enabled multi-core parallel computing platforms. To measure the performance practically, we built our own computing platforms using a GPU hardware (1152 cores) and the TensorFlow software library. In order to evaluate the performance with GPU, we conducted the deep learning computation with various numbers of hidden layers in multilayer perceptron. As presented in the comparative performance results, utilizing GPU hardware improved the performance in terms of computation time (about 3 times or even more).
Cited by 35
Zhihui Zhu, Xiao Shan Yang, Hao Li et al. · Journal of Central South University · 2023
Cited by 2
Marek Woda, Adam Hajduga · Lecture notes in computer science · 2018
Cited by 0
Ujwala Bhangale, Surya S. Durbha, Roger L. King et al. · Remote Sensing of Environment · 2017
Cited by 31
Ryan A. Rossi, Rong Zhou · 2016
Massively parallel architectures such as the GPU are becoming increasingly important due to the recent proliferation of data. In this paper, we propose a key class of hybrid parallel graphlet algorithms that leverages multiple CPUs and GPUs simultaneously for computing k-vertex induced subgraph statistics (called graphlets). In addition to the hybrid multi-core CPU-GPU framework, we also investigate single GPU methods (using multiple cores) and multi-GPU methods that leverage all available GPUs simultaneously for computing induced subgraph statistics. Both methods leverage GPU devices only, whereas the hybrid multi-core CPU-GPU framework leverages all available multi-core CPUs and multiple GPUs for computing graphlets in large networks. Compared to recent approaches, our methods are orders of magnitude faster, while also more cost effective enjoying superior performance per capita and per watt. In particular, the methods are up to 300 times faster than a recent state-of-the-art method. To the best of our knowledge, this is the first work to leverage multiple CPUs and GPUs simultaneously for computing induced subgraph statistics.
Cited by 15
Yoshihito Sano, Yoshiaki Kadono, Naoki Fukuta · 2014
There are strong demands to utilize multi-core computing resources effectively for large-scale and highly detailed multi-agent simulations. We have proposed a framework to assist parameter tuning process of multi-core programming for simulation developers to utilize many parallel cores in their simulation programs efficiently. However, due to its massive computation costs, it is not easy task to seek the sufficient compilation and execution parameters and analyze their performance characteristics for various execution settings. In this paper, we present a preliminary analysis of parameter optimization based on BLMAB by utilizing our framework. We show how our BLMAB-based approach can effectively be used on the parameter optimization process.
Cited by 0
John Stone · 2008
• General introduction to GPU computing • Explore CUDA algorithms for computing electrostatic fields around molecules – Detailed look at CUDA implementations of a simple direct Coulomb summation algorithm – Multi-GPU direct Coulomb summation – Cutoff (range-limited) summation algorithm • CUDA acceleration of parallel molecular dynamics simulation on GPU clusters with NAMD • Wrap-up
Cited by 0
Keeyoung Son, Seonguk Choi, Keunwoo Kim et al. · 2024
In this paper, signal integrity (SI) design and analysis of high-speed interconnect of silicon interposer based heterogeneous integration for artificial intelligence (AI) computing was carried out. With the increasing popularity of AI services, there has been a significant surge in the demand for AI computing capabilities. As a result, high-performance, high-density AI computing modules based on multi-GPU architectures, rather than single-GPU configurations, have emerged as a prominent solution. For high-density GPU integration, the silicon interposer based heterogeneous integration of multi-GPU architecture is arisen. In this research, we design multi-GPU integration on silicon interposer and high-speed GPU links considering routability. Furthermore, we analyzed the designed high-speed GPU link considering SI, but also fabrication cost. Consequently, multiple GPUs are integrated on a silicon interposer, but it has been determined that adequate SI for GPU links cannot be secured on the interposer. Instead, utilizing a package for these connections ensures the necessary SI and offers significant advantages in fabrication cost with reducing metal layers of silicon interposer.
Cited by 4
Dajun Chang, Li Li, Ying Chang et al. · EURASIP Journal on Advances in Signal Processing · 2022
Cited by 0Open accessRetracted
Q. Y. Xiong, S. Y. Huang, Zhuoning Yuan et al. · AIP Advances · 2025
The rapid development of an emerging computing device, the graphical processing unit (GPU), has significantly enhanced our ability to conduct full kinetic particle-in-cell (PIC) simulations in space physics. In this paper, we propose an approach that leverages multiple GPUs to facilitate large-scale PIC simulations. This method can effectively reduce data transmission frequency and latency during the computing process. The data communication between GPU devices is optimized through a combination of Message Passing Interface (MPI)-NVIDIA Collective Communications Library (NCCL) running pattern. Our implementation surpasses the expected linear acceleration, achieving superior computing performance and operational efficiency. The instances of large-scale PIC simulations are presented based on physical models of magnetic reconnection, plasma turbulence, and quasi-perpendicular shock. The importance of large-scale simulations is demonstrated in terms of grid resolution, macroparticles used per cell, and the mass ratio between ions and electrons. The multi-GPU enabled fully kinetic PIC simulation demonstrates its capability to efficiently handle large-scale PIC simulations as a crucial requirement for the study of space plasma physics.
Cited by 1Open access
Dajun Chang, Li Li, Ying Chang et al. · EURASIP Journal on Advances in Signal Processing · 2021
Abstract Nowadays, with the rapid growth of data volume, massive data has become one of the factors that plague the development of enterprises. How to effectively process data and reduce the concurrency pressure of data access has become the driving force for the continuous development of big data solutions. This article mainly studies the MapReduce parallel computing framework based on multiple data fusion sensors and GPU clusters. This experimental environment uses a Hadoop fully distributed cluster environment, and the entire programming of the single-source shortest path algorithm based on MapReduce is implemented in Java language. 8 ordinary physical machines are used to build a fully distributed cluster, and the configuration environment of each node is basically the same. The MapReduce framework divides the request job into several mapping tasks and assigns them to different computing nodes. After the mapping process, a certain intermediate file that is consistent with the final file format is generated. At this time, the system will generate several reduction tasks and distribute these files to different cluster nodes for execution. This experiment will verify the changes in the running time of the PSON algorithm when the size of the test data set gradually increases while keeping the hardware level and software configuration of the Hadoop platform unchanged. When the number of computing nodes increases from 2 to 4, the running time is significantly reduced. When the number of computing nodes continues to increase, the reduction in running time will become less and less significant. The results show that NESTOR can complete the basic workflow of MapReduce, and simplifies the process of user development of GPU positive tree order, which has a significant speedup for applications with large amounts of calculations.
Cited by 4Open accessRetracted
Naoki Kawada, Kan Okubo, Norio Tagawa et al. · Nippon Onkyo Gakkaishi/Acoustical science and technology/Nihon Onkyo Gakkaishi · 2015
Cited by 1Open access
Jorge Francés, Sergio Bleda, Cristian Neipp et al. · Computer Physics Communications · 2012
Cited by 16
Rongteng Wu · IEEE Access · 2018
As an essential part of current mainstream computing systems, GPUs are not only powerful graphics engines but also highly parallel programmable processors. Collaboration between CPUs and GPUs is required to obtain high computing performance in multi-CPU and multi-GPU heterogeneous systems. It is challenging to develop new parallel algorithms on heterogeneous architectures with multiple CPUs and multiple GPUs for such purposes as communication, load balancing, memory spaces, and synchronization We present a parallel Cholesky block factorization algorithm for heterogeneous multi-CPU and multi-GPU architectures. First, a matrix is partitioned into different-sized blocks based on with the performance of the CPU and GPU. Then, a one-dimensional row block-cyclic distribution strategy is used to allocate row block data to every CPU and GPU to minimize communication. The computing task related to the definite row block will then be executed by the corresponding CPU or GPU. Experiments on a system with two CPUs and eight GPUs show good load balancing, parallelism, communication cost, and scalability.
Cited by 7Open access
Yadi Luo, Jing Li, Jun Lu et al. · IOP Conference Series Materials Science and Engineering · 2019
Abstract Multi-channel multi-core CPU parallel and CPU+GPU heterogeneous parallel are effective means to enhance Network Analysis Application computing. To solve the job scheduling problem of Network Analysis Application with multi-level dispatching, multi-user, multi-task in CPU/GPU heterogeneous cluster environment, a multi-user concurrent job scheduling method for Network Analysis Application in CPU/GPU cluster is proposed. According to the characteristics of Network Analysis Application, it is suggested that state estimation and dispatcher power flow should be accelerated in parallel with CPU, while static security analysis, perturbation power flow calculation and interruption capacity scanning should be accelerated in parallel with CPU+GPU. The job scheduling method proposed in this paper can satisfy the high concurrent requests of multi-level Dispatching & Control Center in the isomorphic and heterogeneous computing environments.
Cited by 0Open access