Alexander Antonov

A Study of the Dynamic Characteristics of Software Implementation as an Essential Part for a Universal Description of Algorithm Properties

The AlgoWiki open encyclopedia of parallel algorithmic features enables the entire computing community to work together to describe the properties of a multitude of mathematical algorithms and their implementation for various software and hardware platforms. As part of the AlgoWiki project, a structure has been suggested for providing universal descriptions of algorithm properties. Along with the first part of the description, dedicated to machine-independent properties of the algorithms, it is extremely important to study and describe the dynamic characteristics of their software implementation. By studying fundamental algorithm properties such as execution time, performance, data locality, efficiency and scalability, we can give some estimate of the potential implementation quality for a given algorithm on a specific computer and lay the foundation for comparative analysis of various computing platforms with regards to the algorithms presented in AlgoWiki.

Generalized Approach to Scalability Analysis of Parallel Applications

This article describes an approach to scalability analysis of parallel applications, which is a major part of the algorithm description used in AlgoWiki, the Open Encyclopedia of Parallel Algorithmic Features. The proposed approach is based on the suggested definition of generalized scalability of a parallel application. This study uses joined and structured data on an application’s execution and supercomputing co-design technologies. Parallel application properties are studied by analyzing data collected from all available sources of its dynamic characteristics and information about the hardware and software platforms corresponding with the features of an algorithm and its implementation. This allows reasonable conclusion to be drawn regarding potential reasons of changes in the execution quality for any parallel applications and to compare the scalability of various programs.

JobDigest – Detailed System Monitoring-Based Supercomputer Application Behavior Analysis

The efficiency of computing resources utilization by user applications can be analyzed in various ways. The JobDigest approach based on system monitoring was developed in Moscow State University and is currently used in everyday practice of the largest Russian supercomputing center of Moscow State University. The approach features application behavior analysis for every job run on HPC system providing: the set of dynamic application characteristics - time series of values representing utilization of CPU, memory, network, storage, etc. with diagrams and heat maps; the integral characteristics representing average utilization rates; job tagging and categorization with means of informing system administrators and managers on suspicious or abnormal applications. The paper describes the approach principles and workflow, it also demonstrates JobDigest use cases and positioning of the proposed techniques in the set of tools and methods that are used in the MSU HPC Center to ensure its 24/7 efficient and productive functioning.

What Do We Need to Know About Parallel Algorithms and Their Efficient Implementation

The computing world is changing and all devices—from mobile phones and personal computers to high-performance supercomputers—are becoming parallel. At the same time, the efficient usage of all the opportunities offered by modern computing systems represents a global challenge. Using full potential of parallel computing systems and distributed computing resources requires new knowledge, skills and abilities, where one of the main roles belongs to understanding key properties of parallel algorithms. What are these properties? What should be discovered and expressed explicitly in existing algorithms when a new parallel architecture appears? How to ensure efficient implementation of an algorithm on a particular parallel computing platform? All these as well as many other issues are addressed in this chapter. The idea that we use in our educational practice is to split a description of an algorithm into two parts. The first part describes algorithms and their properties. The second part is dedicated to describing particular aspects of their implementation on various computing platforms. This division is made intentionally to highlight the machine-independent properties of algorithms and to describe them separately from a number of issues related to the subsequent stages of programming and executing the resulting programs.

Computational science and HPC education for graduate students: Paving the way to exascale

Abstract The article discusses the experience of teaching supercomputer disciplines to students specializing in Computational Mathematics. Graduates specializing in this field become future developers and users of complex supercomputing applications and systems. The article presents the structure of a training program that has been implemented at the Faculty of Computational Mathematics and Cybernetics of Lomonosov Moscow State University. It focuses on the content of disciplines related to parallel computing with a detailed description of the structure and content of the course “Supercomputing Simulation and Technologies”, which is offered as part of the Master’s degree training program at the Faculty. The content of practical assignments supporting this discipline is discussed in detail, along with the results produced by the students who performed these practical assignments on Lomonosov and IBM Blue Gene/P supercomputers. The main contribution of the paper is twofold: we draw attention to the importance of study of a wide set of parallel algorithms properties and provide a practical methodology to reach this goal.

Hierarchical Domain Representation in the AlgoWiki Encyclopedia: From Problems to Implementations

Algorithm description is the basic unit in the AlgoWiki Open Encyclopedia of Algorithmic Features. However, computational algorithms are not objectives in and of themselves: they are needed to address problems encountered in various fields of science and industry. On the other hand, there are many practical problems that can be tackled using various methods. This warrants the introduction of another basic term that fits between the concepts of a problem and an algorithm. Also, any algorithm can have different implementations, whether related to a single computing platform or to different platforms. The “problem–method–algorithm–implementation” chain is the basis for describing any subject area in AlgoWiki. This paper describes the permitted freedom in describing such chains, which arises when studying the approaches to address various practical problems.

An AlgoView Web-visualization System for the AlgoWiki Project

There are countless ways to define an algorithm structure, which are mostly organized by flow of data, by executed tasks or by data decomposition. The so-called information graph provides a combination of these patterns. A possibility to investigate visually the information graph of a particular algorithm is, therefore, an adequate tool that helps to understand the algorithm itself, determining its resource of parallelism and figuring out how to code it better for parallel computing systems. In this paper, we present our approach to the information graphs visualization system, where online availability and low computational cost are the primary goals.

Application of the V-Ray technology for optimization of the TRFD and FL052 Perfect Club Benchmarks to CRAY Y-MP and CRAY T3D supercomputers

The paper shows an application of the so-called V-Ray Technology for optimizing the TRFD and FLO52 Perfect Club Benchmarks to CRAY Y-MP and CRAY T3D supercomputers. We also discuss briefly the process of the determination of the potential parallelism of programs within V-Ray since this part of the technology played the key role for the successful optimization of the codes.