The evolution of supercomputers from very expensive special machines to cheaper high performing computing (HPC) has made very fast and advance systems available to an audience much vaster now than just 10 years ago. However, more computing power per installation and more installations mean more electrical power needed and more complex heat management. Evaluating the Total Cost Ownership (TCO) of a HPC solution over its lifespan is a good approach to compare different options and proceed towards the most efficient one. Density, location, size of the installation, type of cooling chosen, square footage, level of redundancy needed, complexity of the data centre and type of interconnect network are factors that contribute to low maintenance cost and less non IT related capital investments. Also, with rising costs of energy and HPC power-hungry systems, energy expenditure in the buying decision is a very important factor to be considered. Thus, HPC platforms are trying to be more cost efficient, not only for green IT but also for the economic health of the HPC data centers.

In this seminar we will review the concept of the total cost of ownership in HPC as well as the main issues and considerations to optimize acquisition costs of supercomputers during its lifetime. This seminar will be given in English.

Die Themenvergabe und die Terminabsprache erfolgt in einem Vorbesprechungstermin und über die Veranstaltungsmailingliste.

Beachten Sie auch unsere allgemeinen organisatorischen Hinweise zu Seminaren.

Das Seminar eignet sich für Studierende der Informatik in den Diplom- und Bachelorstudiengängen. Studierende anderer Studiengänge müssen die Anrechnung mit dem jeweiligen Prüfungsausschuss klären.

Interessierte Zuhörer sind auch herzlich willkommen.

• Dr. Manuel Dolz (Ansprechpartner)
• Prof. Dr. Thomas Ludwig

1. “HPC data center cost considerations: investment, operation and maintenance. DKRZ example”. Assigned to: Anna-Lena ¹⁾
2. “HPC data center benefit considerations: science and industry”.
3. “HPC data center computer room efficiency. Classical vs. container-based, Efficiency metrics, Water vs. air cooling”. Assigned to: Federico ²⁾
4. “HPC data center TCO model”. Assigned to: Rickmer ³⁾
5. “HPC data center energy-efficient techniques: job scheduling, best programming practices, energy-saving hardware/software mechanisms”. Assigned to: Hakob ⁴⁾
6. “HPC data center store optimization: store vs. recompute, compression vs. information loss, deduplication”.
7. “HPC data center costs and benefits of accelerator usage”.
8. “HPC data center node virtualization”.
9. “HPC data center Nationwide efficiency: Tier levels”.
10. “Current projects on reducing TCO of data centers: Montblanc, GreenFlash, Exa2Green, etc”. Assigned to: Grischa ⁵⁾

• 01.04.2014
  • Introduction ^{6) 7)}

• 15.04.2014
  • „The Costs of HPC-Based Science in the Exascale Era“ ^{8) 9)}

• 26.05.2014 Presentation session 1 - Room 119 - Time: 16:15-18:00
  • Presenter 1: „HPC data center cost considerations“ - Anna-Lena

• 03.06.2014 Presentation session 2 - Room 119 - Time: 14:00-16:00
  • Presenter 1: „HPC data center TCO model“ - Rickmer
  • Presenter 2: „HPC data center energy-efficient techniques“ - Hakob
  • Visit to the computer room - Time: 16:00

• 01.07.2014 Presentation session 3 - Room 023 - Time: 14:00-16:00
  • Presenter 1: „Current projects on reducing TCO of data centers“ - Grischa
  • Presenter 2: „HPC data center computer room efficiency“ - Federico

• 08.07.2014 Discussion session - Time: 15:00-16:00
  • Discussion at Hadley's Café

• The Datacenter as a Computer. An Introduction to the Design of Warehouse-Scale Machines. Luiz André Barroso and Urs Hölzle, Google Inc. Link
• Brainware for green HPC. Christian Bischof, Dieter an Mey, Christian Iwainsky. Computer Science - Research and Development, November 2012, Volume 27, Issue 4, pp 227-233. Link
• Department of Energy Exascale Strategy. U.S. Dept. of Energy. Report of June 2013. Link
• Cost‐Benefit Quantification for HPC: An Inevitable Challenge. Report on the Birds‐of‐a‐Feather Session at SC'13 in Denver. Thomas Ludwig, Sandra Wienke, Albert Reuther, Amy Apon and Earl Joseph. Link
• Vor 50 Jahren: IBM stellt teuerstes Computersystem aller Zeiten vor. Link