Performance analysis of HPC applications with PIKA

NHR Tutorial (Online)

Date will be announced soon, planned for autumn 2024

Speaker: Frank Winkler

PIKA is a hardware performance monitoring stack that was specially developed for the Taurus cluster of the TU Dresden in order to identify potentially inefficient jobs. HPC users have the option of analyzing the efficiency of their jobs using an interactive web interface. This tutorial contains a brief introduction to job-specific monitoring and describes how HPC users can evaluate their own jobs in terms of performance and resource utilization with the help of PIKA.

Agenda

• Introducing the PIKA stack
  • Which performance metrics are measured?
• Live demonstration
  • Performance analysis of recorded user jobs
• Hands-on Session
  • Submit a sample job and analyze it
  • Analysis of your own user job data (questions from the group of participants are welcome)

Handouts

The course material (slides) will be made available to the class participants.

HPC-Certification Forum Links

Prerequisites

• The user should already work with his own HPC jobs on the Taurus cluster and therefore have basic HPC knowledge (see HPC-CF Skill Tree Link: K-B HPC Knowledge).

• It is an advantage if your own jobs have already run successfully for at least 5 minutes, which can then be analyzed with PIKA.

• In order to understand the performance data recorded by PIKA, a " PE-B Performance Engineering" certificate is an advantage. This course is about a basic understanding of resource utilization using hardware counters and the interactive use of the PIKA web interface.

• Optional: PE-B Performance Engineering

Learning Objectives

• Does my HPC job use the requested resources efficiently?
  • Without the use of additional profiling tools, PIKA can be seen as the first stage of the optimization cycle from the "PE-B Performance Engineering" HPC-CF Skill Tree Link (PE5-B Optimization Cycle) to get an initial overview of job performance.
• Basic understanding of resource utilization using the hardware counter and interactive use of the PIKA web interface (see HPC-CF Skill Tree Link: PE2.2.2.5-B PIKA)

Registration

Link to Registration Page for Monday:
https://event.zih.tu-dresden.de/nhr/vihps

You will receive the access data shortly before the event by email to your registered email address.

Further Information

Course language: English
Target group: HPC Basics / HPC User / HPC Dev

If you have any further questions, please contact Anja Gerbes ().