The study presents a comparison of computing systems based on IBM POWER8, IBM POWER9, and Intel Xeon Platinum 8160 processors running parallel applications. Memory subsystem bandwidth was studied, parallel programming technologies were compared, and the operating modes and capabilities of simultaneous multithreading technology were analyzed. Performance analysis for the studied computing systems running parallel applications based on the OpenMP and MPI technologies was carried out by using the NAS Parallel Benchmarks. An assessment of the results obtained during experimental calculations led to the conclusion that IBM POWER8 and Intel Xeon Platinum 8160 systems have almost the same maximum memory bandwidth, but require a different number of threads for efficient utilization. The IBM POWER9 system has the highest maximum bandwidth, which can be attributed to the large number of memory channels per socket. Based on the results of numerical experiments, recommendations are given on how the hardware of a similar grade can be utilized to solve various scientific problems, including recommendations on optimal processor architecture choice for leveraging the operation of high-performance hybrid computing platforms.
The article is devoted to the problem of solving scientific problems in the field of high-performance computing systems. An approach to solving a certain kind of problems in materials science is the use of mathematical modeling technologies implemented by specialized modeling systems. The greatest efficiency of the modeling system is shown when deployed in hybrid high-performance computing systems (HHPC), which have high performance and allow solving problems in an acceptable time with sufficient accuracy. However, there are a number of limitations that affect the work of the research team with modeling systems in the HHPC computing environment: the need to access graphics accelerators at the stage of development and debugging of algorithms in the modeling system, the need to use several modeling systems in order to obtain the most optimal solution, the need to dynamically change settings modeling systems for solving problems. The solution to the problem of the above limitations is assigned to an individual modeling environment functioning in the HHPC computing environment. The optimal solution for creating an individual modeling environment is the technology of virtual containerization. An algorithm for the formation of an individual modeling environment in a hybrid high-performance computing complex based on the «docker» virtual containerization system is proposed. An individual modeling environment is created by installing the necessary software in the base container, setting environment variables, installing custom software and licenses. A feature of the algorithm is the ability to form a library image from a base container with a customized individual modeling environment. In conclusion, the direction for further research work is indicated. The algorithm presented in the article is independent of the implementation of the job management system and can be used for any high-performance computing system.
This article discusses a methodology for assessing the effectiveness of a high-performance research platform. The assessment is carried out for the example of the "Informatika" Center for Collective Use (CCU) established at the Federal Research Center of the Institute of Management of the Russian Academy of Sciences, for solving new materials synthesis problems. The main objective of the "Informatika" Center for Collective Use is to conduct research using the software and hardware of the data center of the FRC IU RAS, including for the benefit of third-party organizations and research teams. The general characteristics of the "Informatika" Center for Collective Use are presented, including the main characteristics of its scientific equipment, work organization and capabilities. The hybrid high-performance computing cluster of the FRC CSC RAS (HHPCC) is part of the data center of the FRC IU RAS and also part of the “Informatika” Center for Collective Use. HHPCC provides computing resources in the form of cloud services as software (SaaS) and platform (PaaS) services. With the aid of special technologies, scientific services are delivered to researchers in the form of subject-oriented applications. Based on the analysis of the structure and operation principles of the Informatika Center, key performance indicators of the Center have been developed taking into account its specific tasks in order to characterize its various activity aspects (development, activities and performance). CCU efficiency evaluation implies calculation, on the basis of the developed indicators, of overall (generalized) indicators that characterize the CCU operation efficiency in various areas. An integral indicator is also calculated showing the overall CCU efficiency. To develop the overall performance indicators and the integral performance indicator, it is suggested to use the methods of weighted average and analysis of hierarchies. The procedure of determining partial performance indicators has been considered. Specific features of the choice of CCU performance indicators for solving new materials synthesis problems have been identified that characterize computing complex capabilities in the creation of a virtualization environment (peak performance of a computing system, real performance of a computing system on specialized tests, equipment loading with applied tasks and program code efficiency).
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.