TAU Performance System Workshop on LUMI, March 7-9, CSC

This workshop will focus on performance data collection, analysis, and performance optimization using the various tools of the TAU Performance System (TAU core and APEX).

After describing and demonstrating how performance data can be collected in a straightforward manner, the bulk of the workshop will cover how to instrument application codes and to analyze the performance data collected to explain where the time is spent on the CPU as well as kernels executing on the AMD GPUs.
The workshop will include some sample codes that illustrate the different instrumentation and measurement choices available to the users.

Note: The last day of the workshop will be dedicated to working on participants’ own codes.

Topics will cover

• generating performance profiles and traces with memory utilization and headroom
• I/O
• interfaces to ROCm including ROCProfiler and ROCTracer with support for collecting hardware
  performance data.

Additionally, the workshop will cover instrumentation of OpenMP programs using OpenMP Tools Interface (OMPT) including support for target offload and measurement of a program’s memory footprint.

We will demonstrate

• scalable tracing using OTF2 and visualization using the Vampir trace analysis tool.
• Performance data analysis using ParaProf and PerfExplorer using the performance data management framework (TAUdb) that includes TAU’s performance database.

The workshop will also feature cross experiment analysis including comparing the effects of multi-core architectures on code performance. The demonstrations will include using TAU with HIPCC with compiler-based instrumentation with TAU’s LLVM plugin. The workshop will also cover using TAU in the ECP Extreme-Scale Scientific Software Stack (E4S) [https://e4s.io] using container technology.

We will attempt to collect and analyze performance data for additional user codes during the hands-on
portion of the workshop.

The workshop will run daily from 9am to 4 pm. LUMI accounts will be created for the participants.

Prerequisites

• Familiarity with Linux command line use
• Some experience in scientific code development
• At least basic understanding of factors affecting code performance
• Preferably actively working on a code targeting LUMI (for the hands-on part on the 3rd day)

Registration ends on February 27, 2023 at 23:59.

Lecturers: Sameer Shende and Kevin Huck