Programmierung und Architekturen von Cluster-Rechnern (PS-PAC)7.5 ECTS
(englische Bezeichnung: Programming and Architecture of Compute Clusters)
(Prüfungsordnungsmodul: Programmierung und Architekturen von Cluster-Rechnern)

Lehrveranstaltungen:

• • Programmierung und Architekturen von Cluster-Rechnern
    (Vorlesung, 4 SWS, Thorsten Blaß et al., Do, 10:15 - 11:45, K2-119; Fr, 12:15 - 13:45, K2-119; ab 7.5.2020)
  • Übung zu Programmierung und Architekturen von Cluster-Rechnern
    (Übung, 2 SWS, Thorsten Blaß, Do, 14:15 - 15:45, 00.151-113; Mi, 12:15 - 13:45, 00.153-113 CIP; ab 7.5.2020; n.V. (in der ersten Vorlesung))

Inhalt:

There are many problems where a single computer cannot compute a solution in reasonable time. In this case we must use multiple computers working in concert (a cluster). If a researcher wants to use a cluster at another location (because the current cluster is 'full'), then the researcher needs a way to safely run jobs in some other location. This is 'grid computing'. Similarly, a smartphone may not suffice, and we must let the computation run elsewhere (potentially in parallel). If many smartphones need remote services, we need many remote machines, which is again a cluster. This is called 'cloud-computing'. All of the above are topics of this lecture.

In this lecture we will:

• examine use cases

• examine ways to use a cluster, cloud, or grid.

• examine how to optimize applications for clusters (load balancing, message passing optimizations)

• examine message passing libraries, both from how to use them to how they are implemented.

• examine how high-end networking hardware works and how it is programmed

• study various cloud and grid computing systems

• look into the future where clouds and clusters are augmented with compute accelerators (exascale computing).
  

In the exercises you will:

• program a scalable application that runs on a cluster (there are a number of projects to choose from).

• learn how to achieve scalability,

• learn how to use a cluster practically, and

• learn how to interface with a cluster.

Lernziele und Kompetenzen:

The students

• program a scalable application that runs on a cluster

• optimize their implementation in order to achieve scalability

• use a cluster practically and configure it to satisfy their needs

Organisatorisches:

The time and date of the exercise class will be determined later in cooperation. No exercise class the first two weeks, however. The choice of language used in the class (English or German) is by request of the students.

Verwendbarkeit des Moduls / Einpassung in den Musterstudienplan:

1. Computational Engineering (Rechnergestütztes Ingenieurwesen) (Bachelor of Science)
   (Po-Vers. 2010 | TechFak | Computational Engineering (Rechnergestütztes Ingenieurwesen) (Bachelor of Science) | Gesamtkonto | Technische Wahlmodule | Programmierung und Architekturen von Cluster-Rechnern)

Studien-/Prüfungsleistungen:

Programmierung und Architekturen von Cluster-Rechnern (Prüfungsnummer: 399289)

(englischer Titel: Programming and Architecture of Computer Clusters)

Prüfungsleistung, mündliche Prüfung, Dauer (in Minuten): 30, benotet, 7.5 ECTS

Anteil an der Berechnung der Modulnote: 100.0 %

weitere Erläuterungen:
ACHTUNG: Falls erforderlich, findet die Prüfung gemäß §4, Absatz 1, Satz 2 der Corona-Satzung der FAU (vom 17.04.2020) in elektronischer/digitaler Form statt!

Erstablegung: SS 2020, 1. Wdh.: WS 2020/2021, 2. Wdh.: keine Wiederholung

1. Prüfer:

Michael Philippsen

1. Prüfer:

Alexander Ditter