Performance analysis tools beyond C, C++ and Fortran - Task 064

Fit to programme

This task has been identified by the working groups as part of the agenda behind WP 1.2.

The task number is 064.

Summary

Many performance analysis tools cater to more ‘traditional’ HPC languages: C, C++, Fortran, and to a lesser extent Python. Many of these tools have long histories and are very sophisticated. However, there are other languages that are seeing a rise in usage on HPC clusters, e.g., Julia and Rust. We think it is relevant to the DRI community to scope out what the performance tool suppport is for these languages, as tool support is a really important aspect of effective software engineering. This task proposes that a submitter identify one or more programming languages that are less ‘traditional’ in HPC software engineering, and explore what the performance tool support is. A report covering tool availability and features (including links to relevant material will be sufficient).

Methodology

This task is very much exploratory and will likely require the submitter to liaise with members of language communities. Having some familiarity of performance tools in the more traditional languages is also beneficial to be able to compare and contrast. Using these comparisons, it is highly beneficial to see a list of identified gaps in the current tools space.

Outcomes

A report covering tool availability and features (including links to relevant material will be sufficient), to be hosted on the SHAREing webpages.