BazTech - Yadda

1

Measure of adequacy for the supercomputer job management system model

Baranov Anton, Lyakhovets Dmitriy, Savin Gennady, Shabanov Boris, Telegin Pavel

Annals of Computer Science and Information Systems

|

2019

|

Vol. 18

423--426

EN

In this paper we investigate the problem of modelling modern supercomputer job management systems (JMS). When modelling the JMS, one of the main issues is the adequacy of the model used in experimental studies. The paper attempts to determine the measure of the JMS model adequacy by comparing the characteristics of two job streams, one of which was acquired from a real supercomputer and the other is obtained from the JMS model. We show that the normalized Euclidean distance between vectors of jobs residence times obtained from the job streams of the real system and the JMS model can serve as a measure of the adequacy of the JMS model. The paper also defines the reference value of the measure of adequacy corresponding to the JMS model with virtual nodes.

2

Architektura węzłowa : superkomputer klasy Beowulf

Lenarczyk P., Piotrowski Z.

Elektronika : konstrukcje, technologie, zastosowania

|

2018

|

Vol. 59, nr 2

12--14

PL

Zapotrzebowanie na możliwości obliczeniowe nieustannie wzrasta w wielu dziedzinach wiedzy. Dotyczy to również działów, które wcześniej uznawane były za niewymagające obliczeniowo. Szczególną odpowiedzią jest technologia superkomputerów, których liczba dynamicznie zwiększa się. Możliwość rozwoju poszczególnych dziedzin może zostać w znacznym stopniu ułatwiona dzięki rozwojowi technologii Obliczeń Ogólnego Przeznaczenia z użyciem typowych graficznych procesorów masowo równoległych. W artykule zawarto kompletny opis budowy superkomputera w architekturze węzłowej, wraz z opisem problemów związanych z praktyczną implementacją.

EN

Nowadays computational demands are rapidly growing in many scientific areas. This also applies to engineering branches that were previously considered as not very computationally demanding. The answer is supercomputer technology, the number of which is dynamically increasing. Attention should be given to supercomputers, with General Purpose Graphical Processing Unit technology. Such coprocessor could easily enhance computational power in many scientific areas of interest. The paper describes node architecture of supercomputer with description of practical implementation problems.

3

Instability issues in the server room cooling system

Kwiatkowski T., Prusiński P., Zgorzelski D., Kozioł S

Journal of Power Technologies

|

2017

|

Vol. 97, nr 3

252--258

EN

Since 2009, the National Center for Nuclear Research in Poland has been constructing its own High Performance Computing (HPC) Centre under the name Świerk Computing Centre (CIŚ). Now, it is ready reaching its target - 1 PFLOPS - in December 2015. However at the early operation stage, one of its major problems was the unstable work of the HPC cluster cooling system, resulting in increased maintenance costs. The main aim of this work is to thoroughly investigate the origin of the problem and to find the best solution for it based on results from a Computational Fluid Dynamics (CFD) analysis. The constructors suspected that the oscillations in the flow domain are caused by thermal flow stratification, but something else was proved. In this paper, a wide range of cases will be analyzed, covering different work regimes of the installation as well as various geometry modifications. Finally, certain improvement to the current design will be suggested by the CFD Analysis Group.

4

Torus-connected cycles: A simple and scalable topology for interconnection networks

Bossard A., Kaneko K.

International Journal of Applied Mathematics and Computer Science

|

2015

|

Vol. 25, no. 4

723--735

EN

Supercomputers are today made up of hundreds of thousands of nodes. The interconnection network is responsible for connecting all these nodes to each other. Different interconnection networks have been proposed; high performance topologies have been introduced as a replacement for the conventional topologies of recent decades. A high order, a low degree and a small diameter are the usual properties aimed for by such topologies. However, this is not sufficient to lead to actual hardware implementations. Network scalability and topology simplicity are two critical parameters, and they are two of the reasons why modern supercomputers are often based on torus interconnection networks (e.g., Fujitsu K, IBM Sequoia). In this paper we first describe a new topology, torus-connected cycles (TCCs), realizing a combination of a torus and a ring, thus retaining interesting properties of torus networks in addition to those of hierarchical interconnection networks (HINs). Then, we formally establish the diameter of a TCC, and deduce a point-to-point routing algorithm. Next, we propose routing algorithms solving the Hamiltonian cycle problem, and, in a two dimensional TCC, the Hamiltonian path one. Correctness and complexities are formally proved. The proposed algorithms are time-optimal.