Energy Efficient Programming

Efficiency in computer science often refers to the runtime or memory usage, that a certain algorithm needs to produce an output. However, efficiency can also describe the amount of energy, that is consumed by the hardware during the runtime of an algorithm.

This course explains the relevant computer architecture components, as well as different coding techniques, that allow computer scientists to design and program energy efficient algorithms. Participants will learn how intrinsic functions work and how they can be applied to an algorithm to reduce its energy consumption. In addition, the course presents a hands-on approach to measuring energy consumption of programs using the Intel VTune Profiler tool. In contrast to the existing clean-IT courses on openHPI (https://open.hpi.de/channels/clean-it-forum), this course provides a deeper dive into specific, energy efficient, architecture coding techniques.

12. April 2023 - 10. Mai 2023
Kurssprache: English
Advanced, Programming

Kursinformationen

In response to the growing energy consumption of software programs and digital technologies, energy efficiency is on the rise as a key metric for software development. However, computer scientists and developers often lack concrete, applicable tools, that help reduce an algorithm's energy consumption.

This course is aimed at students, researchers, or developers that work or study in computer science or a related field and want to learn more about energy efficient coding techniques.

In this course, you will learn:

  • The basic concepts of computer functionality and architecture, such as clock cycles, parallelism, execution pipelines, and superscalar architectures.
  • How to use coding techniques in C and assembly, to significantly minimize the number of instructions executed by the CPU
  • About intrinsics and macros, and how you can apply them to any algorithm to increase energy efficiency

The course consists of educational videos, multiple choice tests, and multiple, detailed coding examples. Depending on your prior knowledge and engagement with the coding material, you will spend 5-10 hours per week.

Prequisite knowledge for this course is fundamental programming experience in C, as well as basic knowledge about computer architecture concepts and functionality. Experience with an IDE (preferably Microsoft Visual Studio) is recommended to evaluate the programming examples.

You must achieve at least 50% of the points for a record of achievement. By accessing at least 50% of the resources, you will already receive a confirmation of participation.

Was Teilnehmende lernen werden

  • Understand the basic concepts of computer functionality and architecture
  • Learn how to use coding techniques in C and assembly, to significantly minimize the number of instructions executed by the CPU
  • Learn about intrinsics and macros, and how to apply them to any algorithm to increase its energy efficiency

Für wen dieser Kurs gedacht ist

  • Students, researchers, or developers that work or study in computer science or a related field and want to learn more about energy efficient coding techniques

Lernmaterial

  • Intro:

    In this section, you'll get to know the course lecturer, Prof. Vesselin Iossifov. We'll also introduce the concepts and contents, that are used and covered in this course.
  • Week 1 - Motivation, Concepts, and Architecture:

    The first week of the course presents all the basic knowledge you need to know, including: The motivation for energy efficient programming, computer functionality and architecture basics, and an introduction into x86-programming.
  • Week 2 - Coding techniques for energy efficient programming:

    The second week dives deep into various coding techniques, that allow energy efficient programming of superscalar CPUs, including: Coding techniques in C and Assembler, as well as Streaming Instruction Extensions (SSE) and vector intrinsics.
  • I like, I wish:

    This survey will be unlocked after the course has finished and provides the opportunity to leave detailed feedback on the course.

Für diesen Kurs einschreiben

Der Kurs ist kostenlos. Legen Sie sich einfach ein Benutzerkonto auf openHPI an und nehmen Sie am Kurs teil!
Jetzt einschreiben

Anforderungen für Leistungsnachweise

  • Den Leistungsnachweis erhält, wer in der Summe aller benoteten Aufgaben mindestens 50% der Höchstpunktzahl erreicht hat.
  • Die Teilnahmebestätigung erhält, wer auf mindestens 50% der Kursunterlagen zugegriffen hat.

Mehr Informationen finden Sie in den Richtlinien für Leistungsnachweise.

Dieser Kurs wird angeboten von

Prof. Dr. Vesselin Iossifov

Prof. Vesselin Iossifov schloss sein Studium 1977 als Dipl.-Ing. für technische Kybernetik an der TH Ilmenau, Deutschland, ab. 1982 erwarb er außerdem einen Dr.-Ing. in Computer Engineering an derselben Universität, mit Schwerpunkt auf Echtzeit-Betriebssystemen.

1990 erhielt er die Facultas Docendi in Parallel Computer Architectures von der TH Ilmenau und schloss seine Habilitation in Parallel Computer Architectures am Institut für Mathematik und Computerwissenschaften in Sofia, Bulgarien, ab.

Von 1994 bis 2018 arbeitete Iossifov als Professor für Mikrocomputertechnik an der Fachhochschule Berlin, Deutschland. Während der Zeit von 1999 bis 2006 war er als Gastprofessor für Netzwerk-zentrierte Berechnung an den Universitäten Liverpool und Reading im Vereinigten Königreich tätig. Zusätzlich ist er seit 2016 Mitglied des VDE Vorstands Berlin-Brandenburg.