ORGANIZATION OF THE COMPUTATIONAL PROCESS IN SUPERCOMPUTERS WITH DYNAMICALLY CONTROLLED PARTITIONS
Abstract
Worldwide interest in high-performance computing is constantly growing. It is prompted by the needs of various branches of science, education, medicine, engineering and other practical activities of mankind. This article is devoted to the organization of the computational process in the performance of information-related tasks in computers with dynamically controlled cases. Even with the high performance of supercomputers, the tasks assigned to them can take considerable time and can be performed, as a rule, repeatedly. The programs implementing these tasks have a complex multi-module structure with information and control links between them. Therefore, they are implemented in batch mode and often at night. The packages can include various tasks that require high computing power and large main and external memory. In this regard, even within the framework of the most powerful supercomputers, optimization of the computing process is required for the purpose of rational consumption of computing resources while requiring the performance of a certain set of tasks (package) in a given time limit. The organization of the computing process should take into account the features of the computer architecture and its operating system. The article offers a solution to the problem of optimizing the computing process for IBM servers and mainframes on power9 and z14 processors in relation to the operating systems that are used on these computers. The solution is proposed by a sequence of related steps. Methods of network planning and management, theory of parallel computing processes and schedules are used for the solution. In the beginning, the structure of the application is presented and the problem statement is given. The solution begins with determining the critical path and time reserves for individual computational work. The possibilities of minimizing the length of the critical path are further considered. After that, the task of distributing the number of shares of physical processors allocated to dynamic partitions of the computer is set and solved, provided that the application is run without increasing the length of the critical path. In conclusion, we develop a time diagram of the computational process with fixing the final result of the employment of physical processors of the computer and the number of necessary logical partitions.
References
[2] Dongarra J., Sterling T., Simon H., Strohmaier E. High-performance computing: clusters, constellations, MPPs, and future directions. Computing in Science & Engineering. 2005; 7(2):51-59. (In Eng.) DOI: 10.1109/MCSE.2005.34
[3] McLay R., Schulz K.W., Barth W.L., Minyard T. Best practices for the deployment and management of production HPC clusters. State of the Practice Reports (SC'11). ACM, New York, NY, USA, 2011. Article 9, 11 pp. (In Eng.) DOI: 10.1145/2063348.2063360
[4] Bogdanov A.V., Korkhov V.V., Mareev V.V., Stankova E.N. Architecture and topology of multiprocessor computing systems. M.: Internet University of Information Technology. 2012. 176 pp. (In Russ.)
[5] Voevodin V.V., Voevodin Vl.B. Parallel Computing. SPb.: BHV-Petersburg, 2002. 608 pp. (In Russ.)
[6] Levin V.K. Supercomputer Trends. Computational nanotechnology. 2014; 1:35-38. Available at: https://elibrary.ru/item.asp?id=23786741 (accessed 17.01.2019). (In Russ.)
[7] Sinharoy B., Kalla R.N., Tendler J.M., Eickemeyer R.J., Joyner J.B. POWER5 system microarchitecture. IBM Journal of Research and Development. 2005; 49(4.5):505-521. (In Eng.) DOI: 10.1147/rd.494.0505
[8] Kuzminskiy M. Addition to the Power Family. Open Systems.DBMS. 2014; 7:16-18. Available at: https://www.osp.ru/os/2014/07/13042909/ (accessed 17.01.2019). (In Russ.)
[9] Sadasivam S.K., Thompto B.W., Kalla R., Starke W.J. IBM Power9 Processor Architecture. IEEE Micro. 2017; 37(2):40-51. (In Eng.) DOI: 10.1109/MM.2017.40
[10] Kyne F. et al. IBM z/OS Parallel Sysplex Operational Scenarios. IBM, International Technical Support Organization, USA, 2009. 532 pp. Available at: http://www.redbooks.ibm.com/redbooks/pdfs/sg242079.pdf (accessed 17.01.2019). (In Eng.)
[11] Geographically dispersed parallel sysplex support for peer-to-peer VTS. z/OS DFSMS Object Access Method Planning, Installation, and Storage Administration Guide for Tape Libraries. V2R1.0. IBM Corporation, USA, 2013, pp. 22-23. Available at: https://www.ibm.com/support/knowledgecenter/SSLTBW_2.1.0/com.ibm.zos.v2r1.idao300/o3035.htm (accessed 17.01.2019). (In Eng.)
[12] Varrette S., Bouvry P., Cartiaux H., Georgatos F. Management of an academic HPC cluster: The UL experience. Proceedings of the 2014 International Conference on High Performance Computing & Simulation (HPCS). Bologna, 2014, pp. 959-967. (In Eng.) DOI: 10.1109/HPCSim.2014.6903792
[13] Creasy R.J. The Origin of the VM/370 Time-Sharing System. IBM Journal of Research and Development. 198; 25(5):483-490. (In Eng.) DOI: 10.1147/rd.255.0483
[14] Alexandrov A. Spirals of hardware virtualization. Open Systems.DBMS. 2007; 3:12-18. Available at: https://www.osp.ru/os/archive/2007/03/ (accessed 17.01.2019). (In Russ.)
[15] Borzenko A. Dynamic partitions in IBM eServer pSeries servers. BYTE Russia. 2004; 7(71). Available at: https://www.bytemag.ru/articles/detail.php?ID=8579 (accessed 17.01.2019). (In Russ.)
[16] zEnterprise System. Processor Resource/Systems Manager Planning Guide. IBM Corporation, USA, 2011. 236 pp. Available at: https://www-01.ibm.com/support/docview.wss?uid=isg202e537c11be0929c8525776100663570&aid=1 (accessed 17.01.2019). (In Eng.)
[17] Ebbers M. et al. IBM HiperSockets Implementation Guide. An IBM Redbooks publication. IBM, International Technical Support Organization, USA, 2014. 182 pp. Available at: http://www.redbooks.ibm.com/redbooks/pdfs/sg246816.pdf (accessed 17.01.2019). (In Eng.)
[18] Polychronopoulos C.D., Kuck D.J. Guided Self-Scheduling: A Practical Scheduling Scheme for Parallel Supercomputers. IEEE Transactions on Computers. 1987; C-36(12):1425-1439. (In Eng.) DOI: 10.1109/TC.1987.5009495
[19] Shirazi B.A., Hurson A.R., Kavi K.M. Scheduling and Load Balancing in Parallel and Distributed Systems. IEEE Computer Society Press, Los Alamitos, CA, USA, 1995. 520 pp. (In Eng.)
[20] Karpov V.E. Introduction to the parallelization of algorithms and programs. Computer Research and Modeling. 2010; 2(3):231-272. Available at: https://elibrary.ru/item.asp?id=15621762 (accessed 17.01.2019). (In Russ.)
[21] Barsky A.B. Parallel Programming. M.: Intuit, 2016. 345 pp. (In Russ.)
[22] Golenko D.I. Statistical Methods of Network Planning and Management. Nauka, Moscow, 1968, 400 pp. (In Russ.)
[23] Taha H.A. Operations Research: An Introduction. Prentice Hall, 2006. 838 pp. (In Eng.)
[24] Wagner H.M. Principles of Operations Research With Applications to Managerial Decisions. Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1969. 937 pp. (In Eng.)
[25] Kovalenko D.A. Methods and means of automatic synthesis of parallel programs on the basis of the theory of structural functional models. Bulletin of the Tomsk Polytechnic University. 2008; 312(5):39-44. Available at: https://elibrary.ru/item.asp?id=11170114 (accessed 17.01.2019). (In Russ.)
This work is licensed under a Creative Commons Attribution 4.0 International License.
Publication policy of the journal is based on traditional ethical principles of the Russian scientific periodicals and is built in terms of ethical norms of editors and publishers work stated in Code of Conduct and Best Practice Guidelines for Journal Editors and Code of Conduct for Journal Publishers, developed by the Committee on Publication Ethics (COPE). In the course of publishing editorial board of the journal is led by international rules for copyright protection, statutory regulations of the Russian Federation as well as international standards of publishing.
Authors publishing articles in this journal agree to the following: They retain copyright and grant the journal right of first publication of the work, which is automatically licensed under the Creative Commons Attribution License (CC BY license). Users can use, reuse and build upon the material published in this journal provided that such uses are fully attributed.
