Institute for Computer Science
                                                                   H ∞ S
      Sören Metje
      GPUComputingwithPython
  June 16, 2022                   NewestTrendsinHigh-Performance Data Analytics
      )
  Introduction   GPUComputing    Python Frameworks  Summary       References    Appendix
  Table of contents
      1 Introduction
      2 GPUComputing
      3 PythonFrameworks
      4 Summary
      5 Appendix
  Sören Metje                    NewestTrendsinHigh-Performance Data Analytics     2/37
  Introduction    GPUComputing     Python Frameworks   Summary      References     Appendix
  Motivation
    WhyPython?
      ■ High-level programming
      ■ Compatible with many platforms and systems
      ■ Manyhighqualityframeworksandlibraries
      ■ Widely distributed in many different domains
      ■ Bigcommunity
  Sören Metje                     NewestTrendsinHigh-Performance Data Analytics       3/37
   Introduction            GPUComputing             Python Frameworks           Summary             References            Appendix
   Motivation
      WhyGPUComputing?
        ■ Reducewall-clock time
        ■ Achieve higher cost-efficiency
                                                                                     Source: NVIDIA - GPU-Accelerated Google Cloud [NVId]
   Sören Metje                                    NewestTrendsinHigh-Performance Data Analytics                               4/37