sibsutisВестник СибГУТИThe Herald of the Siberian State University of Telecommunications and Information Science1998-6920СибГУТИ10.55648/1998-6920-2024-18-1-29-39sibsutis-821Research ArticleСтатьиСистема анализа оптимизации доступа к общей памяти для реализации стандарта PMIxShared Memory Access Optimization Analysis System for PMIx Standard Implementationhttps://orcid.org/0000-0002-3844-5481КрамаренкоК. Е.KramarenkoK. E.

Крамаренко Константин Евгеньевич, инженер-программист ИФП СО РАН; ст. преподаватель СибГУТИ

630090, Новосибирск, пр-кт. Ак. Лаврентьева, 13,

630102, Новосибирск, ул. Кирова, 86

Konstantin E. Kramarenko, Software engineer of ISP SB RAS; Senior lecturer, SibSUTIS 

630090, Novosibirsk,

630102, Novosibirsk, Kirov St. 86

kostya.kram@gmail.comhttps://orcid.org/0000-0001-6759-1448ПоляковА. Ю.PolyakovA. Y.

Поляков Артём Юрьевич, старший архитектор программного обеспечения 

Santa Clara, California

Artem Yu. Polyakov, Senior software architect

Santa Clara, California

 

artemp@nvidia.comhttps://orcid.org/0000-0002-8272-9924ЕфимовА. В.EfimovA. V.

Ефимов Александр Владимирович, ведущий инженер-программист ИФП СО РАН; к.т.н., доцент СибГУТИ 

630090, Новосибирск, пр-кт. Ак. Лаврентьева, 13,

630102, Новосибирск, ул. Кирова, 86

 

Alexander V. Efimov, Senior software engineer ISP SB RAS; Ph.D., docent, SibSUTIS

630090, Novosibirsk,

630102, Novosibirsk, Kirov St. 86

alexandr.v.efimov@gmail.comИнститут физики полупроводников им. А. В. Ржанова СО РАН (ИФП СО РАН); Сибирский гос. унив. телекоммуникаций и информатики (СибГУТИ)РоссияThe Institute of Semiconductor Physics, Siberian Branch of the Russian Academy of Sciences (ISP SB RAS); Siberian State University of Telecommunications and Information Science (SibSUTIS)Russian FederationNVIDIAСоединённые Штаты АмерикиNVIDIAUnited States2024171220231812939Copyright © Крамаренко К.Е., Поляков А.Ю., Ефимов А.В., 20242024Крамаренко К.Е., Поляков А.Ю., Ефимов А.В.Kramarenko K.E., Polyakov A.Y., Efimov A.V.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://vestnik.sibsutis.ru/jour/article/view/821

Рассмотрены алгоритмы синхронизации доступа к общей памяти с преобладанием операций чтения, которые характерны для реализаций PMIx. Проведен анализ схемы синхронизации N(mutex+signal), результаты показали увеличение пропускной способности примерно на 20 % в режиме «только чтение» и дальнейшее направление проведения оптимизации.

The results in the field of system software stacks optimization for distributed computing systems based on the process management Interface (PMI) are obtained. Algorithms for synchronizing access to shared memory with a predominance of reading operations which are typical for PMIx implementation,are considered. The locking schemes N(mutex+signal) are proposed.

распределенные вычислительные системыинтерфейс управления процессамиPMIPMIxобщая памятьсинхронизация доступаdistributed computing systemsresource managementprocess management interfacePMIPMIxshared memoryРабота выполнена в рамках государственного задания ИФП СО РАН (ГЗ FWGW-2022-0011)ReferencesХорошевский В. Г. Архитектура вычислительных систем. М.: МГТУ им. Н. Э. Баумана, 2008. 520 с.Khoroshevskii V. G. Arkhitektura vychislitel'nykh sistem [Architectures of computer systems]. Moscow, MGTU im. Baumana, 2008. 520 p.Yu W., Wu J.,and Panda D. K. Fast and Scalable Startup of MPI Programs in InfiniBand Clusters // Proc. 11th International Conference on High Performance Computing (HiPC), Bangalore, India, December 19–22, 2004. P. 440–449.Yu W., Wu J., Panda D. K. Fast and Scalable Startup of MPI Programs in InfiniBand Clusters. Proc. 11th International Conference on High Performance Computing (HiPC), Bangalore, India, 19-22 December, 2004, pp. 440-449.Balaji P., Buntinas D., Goodell D., at. al. PMI: A Scalable Parallel Processmanagement Interface for Extreme-scale Systems // Proc. 17th European MPI Users’ Group Meeting Conference on Recent Advances in the Message Passing Interface (EuroMPI’10). Springer-Verlag, Berlin, Heidelberg. P. 31–41.Balaji P., Buntinas D., Goodell D., at. al. PMI: A Scalable Parallel Processmanagement Interface for Extreme-scale Systems. Proc. 17th European MPI Users’ Group Meeting Conference on Recent Advances in the Message Passing Interface (EuroMPI’10), Springer-Verlag, Berlin, Heidelberg, pp. 31-41.Castain R. H., Solt D., Hursey J., and Bouteiller A. PMIx: Process Management for Exascale Environments // Proc. 24th European MPI Users’ Group Meeting (EuroMPI ’17). ACM, New York, NY, USA, Article № 14, 10 p.Castain R. H., Solt D., Hursey J., Bouteiller A. PMIx: Process Management for Exascale Environments. Proc. 24th European MPI Users’ Group Meeting (EuroMPI ’17). ACM, New York, NY, USA, Article no. 14, 10 p.Polyakov A., Karasev B., Hursey J. et. al. A Performance Analysis and Optimization of PMIxbased HPC Software Stacks // Proc. 26th European MPI Users' Group Meeting, September 2019, Article № 9. P. 1–10.Polyakov A., Karasev B., Hursey J. et. al. A Performance Analysis and Optimization of PMIx-based HPC Software Stacks. EuroMPI '19: Proceedings of the 26th European MPI Users' Group Meeting, September, 2019, article no. 9, pp. 1-10.PMIx Consortium. 2017–2020. Process Management Interface for Exascale (PMIx) Standard (version 3.1). [Электронный ресурс]. URL: https://github.com/pmix/pmixstandard/releases/download/v3.1/pmix-standard-3.1.pdf (дата обращения: 10.04.2023).PMIx Consortium. 2017–2020. Process Management Interface for Exascale (PMIx) Standard (version 3.1), available at: https://github.com/pmix/pmixstandard/releases/download/v3.1/pmix-standard-3.1.pdf (accessed: 10.04.2023).Open PMIx master repository [Электронный ресурс]. URL: https://github.com/openpmix/openpmix (дата обращения: 10.04.2023).Open PMIx master repository, available at: https://github.com/openpmix/openpmix (accessed: 10.04.2023).Message Passing Interface Forum. 2015. MPI-3.1: Official document. [Электронный ресурс]. URL: http://www.mpi-forum.org/docs (дата обращения: 10.04.2023).Message Passing Interface Forum. 2015. MPI-3.1: Official document, available at: http://www.mpi-forum.org/docs (accessed: 10.04.2023).Argonne National Laboratory. 2014. Hydra Process Management Framework. [Электронный ресурс]. URL: https://wiki.mpich.org/mpich/index.php/Hydra_Process_Management_Framework (дата обращения: 10.04.2023).Argonne National Laboratory. 2014. Hydra Process Management Framework, available at: https://wiki.mpich.org/mpich/index.php/Hydra_Process_Management_Framework (accessed: 10.04.2023).PMIx Reference RunTime Environment (PRRTE). [Электронный ресурс]. URL: https://github.com/openpmix/prrte (дата обращения: 10.04.2023).PMIx Reference RunTime Environment (PRRTE), available at: https://github.com/openpmix/prrte (accessed: 10.04.2023).SchedMD, LLC. 2017. SLURM Workload Manager. [Электронный ресурс]. URL: https://slurm.schedmd.com/ (дата обращения: 10.04.2023).SchedMD, LLC. 2017. SLURM Workload Manager, available at: https://slurm.schedmd.com/ (accessed: 10.04.2023).Chakraborty S., Subramoni H., Perkins J., Panda D. K. SHMEMPMI – Shared Memory Based PMI for Improved Performance and Scalability // Proc. 24th IEEE European MPI Users’ Group Meeting (CCGrid). P. 60–69.Chakraborty S., Subramoni H., Perkins J., Panda D. K. SHMEMPMI – Shared Memory Based PMI for Improved Performance and Scalability. Proc. 24th IEEE European MPI Users’ Group Meeting (CCGrid), pp. 60-69.Calciu I., Dice D., Lev Y., Luchangco V. et. al. NUMA-aware reader-writer locks // Proc. 18th ACM SIGPLAN symposium on Principles and practice of parallel programming. ACM, New York, NY, USA, 2013. P. 157–166.Calciu I., Dice D., Lev Y., Luchangco V. et. al. NUMA-aware reader-writer locks. Proc. 18th ACM SIGPLAN symposium on Principles and practice of parallel programming, ACM, New York, NY, USA, 2013, pp. 157-166.IEEE Standard for Information Technology–Portable Operating System Interface (POSIX(R)) Base Specifications, Issue 7. IEEE Std 1003.1-2017 (Revision of IEEE Std 1003.1-2008) (Jan 2018). P. 1–3951.IEEE Standard for Information Technology–Portable Operating System Interface (POSIX(R)) Base Specifications, iss. 7. IEEE Std 1003.1-2017 (Revision of IEEE Std 1003.1-2008), Jan, 2018, pp. 1-3951.Hsieh W. C., Weihl W. E. Scalable Reader-Writer Locks for Parallel Systems // Proc. Sixth IEEE International Parallel Processing Symposium, 1992. P. 656–659.Hsieh W. C., Weihl W. E. Scalable Reader-Writer Locks for Parallel Systems. Proc. Sixth IEEE International Parallel Processing Symposium, 1992, pp. 656-659.IEEE Standard for Information Technology–Portable Operating System Interface (POSIX(R)) Base Specifications, Issue 7. IEEE Std 1003.1-2017 (Revision of IEEE Std 1003.1-2008) (Jan 2018). P. 1–3951. [Электронный ресурс]. URL: https://doi.org/10.1109/IEEESTD.2018.8277153 (дата обращения: 01.09.2023).IEEE Standard for Information Technology–Portable Operating System Interface (POSIX(R)) Base Specifications, iss. 7. IEEE Std 1003.1-2017 (Revision of IEEE Std 1003.1-2008), Jan, 2018, pp. 1-3951, available at: https://doi.org/10.1109/IEEESTD.2018.8277153 (accessed: 01.09.2023).Mellor-Crummey J., Scott M. Algorithms for scalable synchronization on shared-memory multiprocessors // ACM Trans. Comput. Syst. 1991. № 9. P. 21–65.Mellor-Crummey J., Scott M. Algorithms for scalable synchronization on shared-memory multiprocessors. ACM Trans. Comput. Syst, 1991, no. 9, pp. 21-65.Mellor-Crummey J. Algorithms for Scalable Lock Synchronization on Shared-memory Multiprocessors. Department of Computer Science Rice University. Lecture 18. 17 March 2009. P. 1–49. [Электронный ресурс]. URL: https://cs.anu.edu.au/courses/comp8320/lectures/aux/comp422-Lecture18-HWLocks.pdf (дата обращения: 10.04.2023).Mellor-Crummey J. Algorithms for Scalable Lock Synchronization on Shared-memory Multiprocessors. Department of Computer Science Rice University, Lecture 18. 17 March, 2009, pp. 1-49. available at: https://cs.anu.edu.au/courses/comp8320/lectures/aux/comp422-Lecture18-HWLocks.pdf (accessed: 10.04.2023).Микробенчмарк. [Электронный ресурс]. URL: https://bitly.su/Wtxe0Z (дата обращения: 05.02.2023).Mikrobenchmark [Microbenchmark], available at: https://bitly.su/Wtxe0Z (accessed: 05.02.2023).OpenPMIx. [Электронный ресурс]. URL: https://shorturl.at/kwx14 (дата обращения: 16.04.2023).OpenPMIx, available at: https://shorturl.at/kwx14 (accessed: 16.04.2023).Ефимов А. В., Поляков А. Ю., Крамаренко К. Е., Бочкарев Б. В. Оптимизация доступа к общей памяти для реализации стандарта PMIx // Вестник СибГУТИ. 2020. № 4. С. 28–38.Efimov A. V., Plyakov A. U., Kramarenko K. E., Bochkarev B. V. Optimizatsiya dostupa k obshchei pamyati dlya realizatsii standarta PMIx [Optimization of access to shared memory for the implementation of the PMIx standard]. Vestnik SibGUTI, 2020, no. 4, pp. 28-38.

The authors declare that there are no conflicts of interest present.