Low Overhead yet Accurate Profiling Tools for Multicore

Europe/Istanbul
Zoom

Zoom

Description

In order to contribute to the dissemination of HPC among academia and SMEs, weekly seminars have been held online since December 2020. HPC use cases in different areas ranging from modern computing tools such as parallel programming, GPU programming, big data, AI, and machine learning techniques to applied subjects such as computational fluid dynamics and materials modelling are presented in these talks, which helped increase the awareness in this area.

For information about the project, visit our EuroCC web page.

  • Seminars will be held online via Zoom.
  • Please register to the Zoom meeting with your real name/surname.

NOTE: Please pay attention to the following points during the online seminar event:

  • When asking questions, please tell your name first.
  • Please use the Q&A section for your questions and the chat section for your comments.
Registration
Registrations
    • 2:00 PM 3:30 PM
      Low Overhead yet Accurate Profiling Tools for Multicore 1h 30m

      In this talk, I will present two profiling tools that my group developed to measure data locality aspects of an application running on a multicore system. Our tools make use of performance monitoring units in a unique way to lower the overhead of profiling. One of such tools is ComDetective, which monitors inter-thread communication. The second tool is ReuseTracker, which analyzes reuse distance in private and shared caches. We expect that both ComDetective and ReuseTracker pave the way for a new generation of low-overhead profiling tools and they can be used by the end-users to improve data locality aspects of their software.

      Speaker: Dr Didem Unat (Koc University)