• International Journal of Contents
• /
• v.17 no.3
• /
• pp.67-73
• /
• 2021

Dark matter is barely known because it cannot be explained using the Standard Model. In addition, dark matter has not been detected yet. It is currently being explored through various ways. In this paper, we studied dark matter in an electron-positron collider using MadGraph5. The signal channel is e⁺e^- → 𝜇⁺𝜇^-A' where A' decays to dimuon. We studied the cross-section by increasing the center-of-mass energy. Central processing unit (CPU) time of simulation was compared with that using a local Linux machine and a KISTI-5 supercomputer (Knight Landing and Skylake). Furthermore, one or more cores were used for comparing CPU time among machines. Results of this study will enable the exploration of dark matter in electron-positron experiments. This study also serves as a reference for optimizing high-energy physics simulation toolkits.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.1
• /
• pp.138-144
• /
• 2010

Tachyon primary system which introduces recently is a high performance supercomputer that composed with AMD Barcelona nodes. In this paper, we will verify the performance and parallel scalability of TachyonIn by using multi-zone NAS Parallel Benchmark(NPB) which is one of a program with hybrid parallel method. To test performance of hybrid parallel execution, B and C classes of BT-MZ in NPB version 3.3 were used. And the parallel scalability test has finished with Tachyon's 1024 processes. It is the first time in Korea to get a result of hybrid parallel computing calculation using more than 1024 processes. Hybrid parallel method in high performance computing system with multi-core technology like Tachyon describes that it can be very efficient and useful parallel performance benchmarks.

• Journal of Internet Computing and Services
• /
• v.17 no.4
• /
• pp.115-125
• /
• 2016

Recently, many attentions have been paid to conducting convergence research across diverse disciplines. Along with this convergence era, an IT-based multi-disciplinary convergence project, called EDISON (EDucation-research Integrated Simulation On the Net), has been launched to support the studies of researchers engaged in several computational science and engineering (CSE) fields and to boost learning motivations of CSE students. Since 2011, we have been successfully carrying out the EDISON project. EDISON as a cyber-learning platform enables CSE researchers to share their own high-performance computing (HPC) simulation softwares developed to solve their research problems accompanying large-scale computation and I/O and users to run the softwares with little constraints on the web. Also, the EDISON platform has been utilized as lecture material by many universities in Korea. This article introduces the construction and service statistics of this EDISON platform. Specifically, we explicate several distinctions between EDISON and existing other HPC service platforms and discuss a three-layered technical architecture of the EDISON platform. We then present the up-to-date service statistics of EDISON over the past four years. Finally, we conclude this article and describe future plans.

• Annual Conference of KIPS
• /
• 2020.11a
• /
• pp.78-79
• /
• 2020

최근 하드웨어의 성능 및 소프트웨어 기술이 비약적으로 발전하면서 컴퓨팅을 위한 인프라 환경이 클라우드 기술 기반으로 활발하게 연구, 개발되고 있다. 이에 본 논문에서는 슈퍼컴퓨터로 대표되는 고성능 컴퓨팅을 분야에서 클라우드 기반 인프라 및 서비스를 제공하기 위한 KI Cloud 플랫폼을 소개한다. KI Cloud 플랫폼은 VM 기반으로 IaaS 서비스를 제공하고, 컨테이너 기술을 기반으로 HPC 사용자를 위한 PaaS 서비스를 제공하는 통합 플랫폼으로 설계 및 개발되었다.

• International journal of advanced smart convergence
• /
• v.12 no.4
• /
• pp.164-170
• /
• 2023

With the recent advances in AI (artificial intelligence) and HPC (high-performance computing) technologies, deep learning is proliferated in various domains of the 4th industrial revolution. As the workload volume of deep learning increasingly grows, analyzing the memory reference characteristics becomes important. In this article, we analyze the memory reference traces of deep learning workloads in comparison with traditional workloads specially focusing on read and write operations. Based on our analysis, we observe some unique characteristics of deep learning memory references that are quite different from traditional workloads. First, when comparing instruction and data references, instruction reference accounts for a little portion in deep learning workloads. Second, when comparing read and write, write reference accounts for a majority of memory references, which is also different from traditional workloads. Third, although write references are dominant, it exhibits low reference skewness compared to traditional workloads. Specifically, the skew factor of write references is small compared to traditional workloads. We expect that the analysis performed in this article will be helpful in efficiently designing memory management systems for deep learning workloads.

• KIPS Transactions on Computer and Communication Systems
• /
• v.8 no.10
• /
• pp.231-238
• /
• 2019

Analyzing next-generation genome sequencing data in a conventional way using single server may take several tens of hours depending on the data size. However, in order to cope with emergency situations where the results need to be known within a few hours, it is required to improve the performance of a single genome analysis. In this paper, we propose a parallelized method for pre-processing genome sequence data which can reduce the analysis time by utilizing the big data technology and the highperformance computing cluster which is connected to the high-speed network and shares the parallel file system. For the reliability of analytical data, we have chosen a strategy to parallelize the existing analytical tools and algorithms to the new environment. Parallelized processing, data distribution, and parallel merging techniques have been developed and performance improvements have been confirmed through experiments.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.2
• /
• pp.610-619
• /
• 2016

Digital manufacturing (DM) technology helps engineers design products promptly and reliably at low production cost by simulating a manufacturing process and the material behavior of a product in use, based on three-dimensional digital modeling. The computing infrastructure for digital manufacturing, however, is usually expensive and, at present, the number of professional design engineers who can take advantage of this technology to a product design accurately is insufficient, particularly in small and medium manufacturing companies. Considering this, the Korea Institute of Science and Technology Information (KISTI) and H University is operating a DM track in the form of Industry-University-Research Institute collaboration to train high-performance-computing-based DM professionals. In this paper, a series of courses to train students to work directly into DM practice in industry after graduation is reported. The operating cases of the DM track for two years since 2013 are presented by focusing on the progress in establishment, lecture and practice contents, evaluation of students, and course quality improvement. Overall, the track management, curriculum management, learning achievement of students have been successful. By expediting more active participation of the students in the track and providing more internship and job offers in the participating companies in addition to collaborative capstone design projects, the track can be expanded by fostering a nationwide training network.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2014.01a
• /
• pp.9-13
• /
• 2014

The performance of large scale software applications has been automatically increasing for last few decades under the influence of Moore's law - the number of transistors on a microprocessor roughly doubled every eighteen months. However, on-chip transistors limitations and heating issues led to the emergence of multicore processors. The energy efficient ARM based System-on-Chip (SoC) processors are being considered for future high performance computing systems. In this paper, we present a case study of two widely used parallel programming models i.e. MPI and MapReduce on distributed memory cluster of ARM SoC development boards. The case study application, Black-Scholes option pricing equation, was parallelized and evaluated in terms of power consumption and throughput. The results show that the Hadoop implementation has low instantaneous power consumption that of MPI, but MPI outperforms Hadoop implementation by a factor of 1.46 in terms of total power consumption to execution time ratio.

• Journal of the Korean Society of Industry Convergence
• /
• v.25 no.2_2
• /
• pp.219-232
• /
• 2022

The aim of this study is to provide the dynamic convergence index that reflected the inherent characteristics of the convergence phenomenon and utilized the nationally-funded R&D projects data, thereby suggesting useful information about the direction of the national convergence R&D strategy. The dynamic convergence index that we suggested was made of two indicators: persistency and diversity. From a time-series perspective, the persistency index, which measures the degree of continuous convergence of multidisciplinary nationally-funded R&D projects, and the diversity index, which measures the degree of binding with heterogeneous research areas. We conducted the empirical experiment with 151,248 convergence R&D projects during the 2015~2021 time period. The results showed that convergence R&D projects in both public health and life sciences appeared the highest degree of persistency. It was presumed that the degree of persistency has increased again due to the COVID-19 pandemic. Meanwhile, the degree of diversity has risen with combining with disciplinary such as materials, chemical engineering, and brain science areas to solve social problems including mental health, depression, and aging. This study not only provides implications for improving the concept and definition of dynamic convergence in terms of persistency and diversity for national convergence R&D strategy but also presented dynamic convergence index and analysis methods that can be practically applied for directing public R&D programs.

• KIISE Transactions on Computing Practices
• /
• v.21 no.1
• /
• pp.13-18
• /
• 2015

Computational science simulations have been used to enable discovery in a broad spectrum of application areas, these simulations show irregular demanding characteristics of computing resources from time to time. The adoption of virtualized high performance cloud, rather than CPU-centric computing platform (such as supercomputers), is gaining interest of interests mainly due to its ease-of-use, multi-tenancy and flexibility. Basically, provisioning a virtual cluster, which consists of a lot of virtual machines, in a real-time has a critical impact on the successful deployment of the virtualized HPC clouds for computational science simulations. However, the cost of concurrently creating many virtual machines in constructing a virtual cluster can be as much as two orders of magnitude worse than expected. One of the main factors in this bottleneck is the time spent to create the virtual images for the virtual machines. In this paper, we propose a novel technique to minimize the creation time of virtual machine images and improve I/O performance of the provisioned virtual clusters. We also confirm that our proposed technique outperforms the conventional ones using various sets of experiments.