In PR #407 the new Amdahl code was brought in to this lesson, as a black-box code that shows desirable performance characteristics and is a useful demonstration of the benefits and limits of parallelized code.
The actual deployment strategy used is that of download a tarball of the amdahl code, specifically broken out as part of the workshop, but the Amdahl code actually does have its own repo, and is installable via 'pip' or from git, or by other methods more consistent with HPC best practice.
The tar-ball scheme was partially an expediency to accommodate an upcoming workshop, but this is worth revisiting with a bit more patience with a view to a better long-term soluiton.
In PR #407 the new Amdahl code was brought in to this lesson, as a black-box code that shows desirable performance characteristics and is a useful demonstration of the benefits and limits of parallelized code.
The actual deployment strategy used is that of download a tarball of the amdahl code, specifically broken out as part of the workshop, but the Amdahl code actually does have its own repo, and is installable via 'pip' or from git, or by other methods more consistent with HPC best practice.
The tar-ball scheme was partially an expediency to accommodate an upcoming workshop, but this is worth revisiting with a bit more patience with a view to a better long-term soluiton.