Papers by Pavel Kostenetskiy
The modern manycore coprocessors and GPUs demonstrate very high performance on certain problems. ... more The modern manycore coprocessors and GPUs demonstrate very high performance on certain problems. Recent research has shown that these coprocessors can be used to accelerate database operations. But, to the best of our knowledge, there is only a little prior work on using coprocessors in multiprocessor database systems. This paper focuses on evaluation of database multiprocessor architectures with manycore coprocessors and GPUs. We implemented the emulator of parallel DBMS that uses computing cluster with NVIDIA GPUs or Intel Xeon Phi coprocessors for relational query processing. It allows to simulate simple SELECT and JOIN queries. A number of experiments have been done using this emulator. These experiments have shown that coprocessors are less efficient than CPUs for processing modeled SELECT queries, but more efficient than CPUs for processing INNER JOIN queries. I.
Increasing Efficiency of Data Transfer Between Main Memory and Intel Xeon Phi Coprocessor or NVIDIA GPUS with Data Compression
Springer eBooks, 2015
Efficient data transfer between main memory and Intel Xeon Phi coprocessor or GPU plays crucial r... more Efficient data transfer between main memory and Intel Xeon Phi coprocessor or GPU plays crucial role in using this devices for database processing. This paper addresses this problem by using data compression methods such as RLE, Null Suppression, LZSS and combination of RLE and Null Suppression. The chosen compression methods were implemented for Intel Xeon Phi coprocessors and NVIDIA GPUs. It is shown experimentally that these compression methods can be used to increase the efficiency of database processing using Intel Xeon Phi coprocessor and NVIDIA GPUs under certain conditions imposed on the data under treatment. It is also shown that, when a compression method allows one to process data without decompression, such a processing procedure can additionally increase the efficiency of this method.
SUSU Supercomputer Resources for Industry and fundamental Science
2018 Global Smart Industry Conference (GloSIC), Nov 1, 2018
The supercomputers of South Ural State University are at the core of the University's researc... more The supercomputers of South Ural State University are at the core of the University's research performance. These facilities enable researchers to perform a broad range of computationally demanding tasks in the fields of engineering, natural and human sciences, and IT. The powerful capabilities of SUSU supercomputers are currently used for more than 250 research projects, including commercial projects as part of the Industry 4.0. The first-class facilities are ranked among the world's most powerful supercomputers by various ranking agencies.
Automatic Visible Defect Detection and Classification System Prototype Development for Iron-and-Steel Works
This paper is written as a part of “Automatic visible defect detection and classification system ... more This paper is written as a part of “Automatic visible defect detection and classification system (AVDDCS) development for Electrolytic Tinning Unit” project for Iron and Steel Works. As a part of pre-design research, a system prototype was developed. It consists of Preprocessor, Classifier, Server, Database and User interface. The difference between this prototype and analogs is the use of convolutional neural networks to improve the accuracy of the classification of visible defects.
Modeling heterogeneous computational cluster hardware in context of parallel database processing
Mathematical modeling is an important approach for creating a parallel database management system... more Mathematical modeling is an important approach for creating a parallel database management system that could efficiently use capabilities, provided by modern heterogeneous computational clusters, equipped with manycore coprocessors or GPUs. To this day, several models heterogeneous computational systems were proposed, but none of them is suited for modeling database processing. In this paper, we address this problem by proposing the Heterogeneous Database Multiprocessor Model. Proposed model consists of several submodels, which describe different aspects of database processing on heterogeneous computational systems. This paper describes hardware platform submodel, which describes the hardware of modeled computational cluster and execution submodel that describes the rules of cooperation between hardware submodel components.
FME Transactions, 2019
Detection and classification of surface defects of the rolled metal are one of the main tasks for... more Detection and classification of surface defects of the rolled metal are one of the main tasks for correctly assessing product quality. Historically, these tasks were performed by a human. However, due to a multitude of production factors, such as high rolling rate and temperature of the metal, the results of such human work are rather low. Replacing a human controller with an artificial intelligence system has been relevant for a long time; it is merely necessary within the concept of Industry 4.0. This paper is devoted to the development of the prototype system automatic detection and classification of defects for one of the Iron-and-Steel Works of the Chelyabinsk region in the Russian Federation. The prototype consists of the Preprocessor, Classifier, Server, Database, and User interface. The main focus is on achieving high classification accuracy, which is planned to be obtained through the use of convolutional neural networks.
High Performance Computing, Networking and Communication Systems, 2007
In this paper we propose a new approach to data distribution and load balancing in multiprocessor... more In this paper we propose a new approach to data distribution and load balancing in multiprocessor database systems with hierarchical architecture. A model of hierarchical database multiprocessor architecture is described. This model is called DMM. It allows us to simulate and analyze various multiprocessors configurations. An important subclass of multiprocessor hierarchies is considered. We call it symmetric hierarchies. For the symmetric hierarchies, a new strategy of data distribution is proposed. This strategy is based on the partial mirror technique. The analytical estimations for disk space overhead due to data replication are obtained. For the regular symmetric hierarchies, the theorems giving estimations of replica building overhead are proven. An efficient method for load balancing is proposed. This method exploits the partial mirror technique. Presented methods are designed for cluster and Grid systems.
HPC TaskMaster – Task Efficiency Monitoring System for the Supercomputer Center
Communications in computer and information science, 2022
Parallel Computational Technologies
Communications in Computer and Information Science, 2019
Efficiency is a major weakness in modern supercomputers. Low efficiency of user applications is o... more Efficiency is a major weakness in modern supercomputers. Low efficiency of user applications is one of the main reasons for that. There are many software tools for analyzing and improving the performance of parallel applications. However, supercomputer users often do not have sufficient knowledge and skills to apply these tools correctly in their specific case. Moreover, users often do not know that their applications work inefficiently. The main goal of our project is to help any HPC user to detect performance flaws in their applications and find out how to deal with them. To this end, we plan to develop an open-source software solution that performs automatic massive analysis of all jobs running on a supercomputer to identify those with efficiency issues and helps users to conduct a detailed analysis of an individual program (using existing software tools) to identify and eliminate the root causes of the loss of efficiency.
Journal of Physics: Conference Series, 2021
The National Research University Higher School of Economics launched its HPC cluster and created ... more The National Research University Higher School of Economics launched its HPC cluster and created a new division named the Supercomputer Simulation Unit. Now the university HPC cluster occupies seventh place in rating the most powerful computers of the CIS TOP50. The HPC cluster uses to solve machine learning problems, population genomics, hydrodynamics, atomistic and continuous modeling in physics, generative probabilistic models, financial row forecasting algorithms, and other actual problems. Paper describes the HSE HPC resources and experience of their use for scientific and educational tasks.
Administration, Monitoring and Analysis of Supercomputers in Russia: a Survey of 10 HPC Centers
Supercomputing Frontiers and Innovations, 2021
Programming and Computer Software, 2013
The paper is dedicated to issues concerning simulation and analysis of hierarchical multiprocesso... more The paper is dedicated to issues concerning simulation and analysis of hierarchical multiprocessor systems oriented to database applications. Requirements for a parallel database system model are given. A survey and comparative analysis of known parallel database system models are presented. A new multipro cessor database system model is introduced. This model allows us to simulate and evaluate arbitrary hierar chical multiprocessor configurations in the context of the OLTP class database applications. Examples of using the database multiprocessor model for simulation study of multiprocessor database systems are pre sented.
HPC TaskMaster – Task Efficiency Monitoring System for the Supercomputer Center
Communications in computer and information science, 2022
Automatic Visible Defect Detection and Classification System Prototype Development for Iron-and-Steel Works
2018 Global Smart Industry Conference (GloSIC), 2018
This paper is written as a part of “Automatic visible defect detection and classification system ... more This paper is written as a part of “Automatic visible defect detection and classification system (AVDDCS) development for Electrolytic Tinning Unit” project for Iron and Steel Works. As a part of pre-design research, a system prototype was developed. It consists of Preprocessor, Classifier, Server, Database and User interface. The difference between this prototype and analogs is the use of convolutional neural networks to improve the accuracy of the classification of visible defects.
The paper is devoted to the issues concerning with modeling and simulating of hierarchical databa... more The paper is devoted to the issues concerning with modeling and simulating of hierarchical database multiprocessor systems. The requirements for a model of parallel database systems are defined. A new computational model of the hierarchical database multiprocessor architecture is described. This model is called DMM (Database Multiprocessor Model) and designed for OLTP workload. It allows us to simulate and analyze an arbitrary multiprocessors configuration for database application. The practical experience of exploiting the DMM model is discussed.
Modeling heterogeneous computational cluster hardware in context of parallel database processing
2017 40th International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO), 2017
Mathematical modeling is an important approach for creating a parallel database management system... more Mathematical modeling is an important approach for creating a parallel database management system that could efficiently use capabilities, provided by modern heterogeneous computational clusters, equipped with manycore coprocessors or GPUs. To this day, several models heterogeneous computational systems were proposed, but none of them is suited for modeling database processing. In this paper, we address this problem by proposing the Heterogeneous Database Multiprocessor Model. Proposed model consists of several submodels, which describe different aspects of database processing on heterogeneous computational systems. This paper describes hardware platform submodel, which describes the hardware of modeled computational cluster and execution submodel that describes the rules of cooperation between hardware submodel components.
2018 41st International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO), 2018
The development of parallel database management systems is an urgent problem due to the rapid inf... more The development of parallel database management systems is an urgent problem due to the rapid information volume growth. Nowadays the basic principles of DBMS performance improvement include the use of multiprocessor systems [8]. At the same time, acceleration could be achieved by using new hardware architectures, such as hybrid clusters with manycore coprocessors. The implementation of such architectures is limited by the high cost of hardware and its configuration. Therefore, the development of models that allow determining several characteristics and comparing different database queries runtime without both using real hardware and taking into account the exact execution details is a highly topical problem. This paper describes the development of a mathematical model that explores the effectiveness of a new manycore accelerator with Intel Xeon Phi Knights Landing hardware architecture in terms of parallel database processing.
Simulation of the parallel database column coprocessor
The paper proposes a mathematic model allowing exploration of effectiveness of different hardware... more The paper proposes a mathematic model allowing exploration of effectiveness of different hardware cluster computing configurations based on multi-core coprocessors while processing databases using approach of distributed columnar indices.
In this paper we propose a new approach to data distribution and load balancing in multiprocessor... more In this paper we propose a new approach to data distribution and load balancing in multiprocessor database systems with hierarchical architecture. A model of hierarchical database multiprocessor architecture is described. This model is called DMM. It allows us to simulate and analyze various multiprocessors configurations. An important subclass of multiprocessor hierarchies is considered. We call it symmetric hierarchies. For the symmetric hierarchies, a new strategy of data distribution is proposed. This strategy is based on the partial mirror technique. The analytical estimations for disk space overhead due to data replication are obtained. For the regular symmetric hierarchies, the theorems giving estimations of replica building overhead are proven. An efficient method for load balancing is proposed. This method exploits the partial mirror technique. Presented methods are designed for cluster and Grid systems.
SUSU Supercomputer Resources for Industry and fundamental Science
2018 Global Smart Industry Conference (GloSIC)
Uploads
Papers by Pavel Kostenetskiy