Changelogs

This page lists out the changelogs for the updates to the exacb software.

2025-06-22 : 3.1.0 (v3)

This release introduces support for the JUPITER machine. In particular, it introduces support for JUPITER machine in the CI/CD components jureap/jube and jureap/scaling. It also introduces a new component jureap/jupiter to plot the JUPITER performance data link.

Apart from this, we also introduce three new variants to the jureap/jube catalog to support the JUPITER evaluation runs:

  • jupiter.evaluation.jupiter

  • jupiter.evaluation.jureca

  • jupiter.evaluation.juwels_booster

All the other machine-independent variants (non-jedi variants) should work on JUPITER as well.

You can find the examples of how to use the new features in the comprehensive example repository. For example, the new CI/CD components are used in the Gitlab CI/CD pipeline here.

The example results of the plots on JUPITER are available for the comparison case here and for the scaling case here.

Please do not hesitate to reach out to use if you have any questions or you encounter any errors.

2024-10-04 : 3.0.0 (v3)

Released new branch of exacb tools (v3) with the following features:

  • Updated checklist to align with the JUREAP certificate instructions presented here.

  • Added instructions in tutorial how to use the the reservations feature in the jureap/jube catalog. See the reservations reference and reservation tutorial for more details.

  • Added new feature to the jureap/jube catalog to support recording of the benchmark data to the repository as shown in here. A tutorial on how to use this feature is available here.

  • Added new feature to the jureap/jube catalog to support adding a label to the benchmark data as shown in here. A tutorial on how to use this feature is available here.

  • Added how-to guides for the JEDI Performance step of the JUREAP certification process.

  • Added new variants to ease the process of executing jureca evaluation benchmarks.

    • jedi.evaluation.jedi

    • jedi.evaluation.jureca

    • jedi.evaluation.juwels_booster

  • Added new strong scaling variant strong.standalone to the allow a single large strong scaling benchmark to be run.

  • Added new CI/CD component jureap/jedi <reference.jureap.jedi> to run evaluation runs on JEDI.

  • Bugfix: Added the weak scaling variants to the jureap variants <jureap.variants>.

2024-07-25

  • We have added support for running the jureap/jube catalog on JEDI

  • Adapted the quickstart example to conduct weak scaling examples.

    The code changes required to run the weak scaling example are available here.

    Some notes: In the weak scaling example, the problem size is a function of the number of nodes. Therefore, we need to make the variable problem_size depend on the variable nodes. However, this can cause conflicts with how problem_size was defined before. Therefore, the original definition of problem_size is restricted to strong.tiny tags as shown here

2024-07-04

  • The jureap/jube CI/CD component now supports extra tags to be added to select between the multiple usecases specified in the same JUBE file. See the usecase reference for more details