Many computational scientists access HPC resources through centralized supercomputing facilities, which serve many users.
Although the details of the systems, tools, and support resources can vary greatly from one facility to another, many high-level aspects are common. Because of inherent lifetimes and technological advances, supercomputer systems typically periodically change every five years. Also, with limited resources available for support and training, facilities typically focus first and foremost on the systems, tools, and environments that they're fielding at any given moment. This situation can pose a challenge for users, who are in many cases using multiple supercomputer facilities simultaneously. The goal of BSSw in supporting supercomputer facilities and users is to provide a place for information that spans facilities and specific computer architectures, complementing local facility resources, and helps users deal wtih multiple environments in a productive and sustainable way.
A vast number of supercomputing facilities exist around the world, and listing them all is impossible. But the following are some of the major facilities used by BSSw contributors.
Performance portablity is often a significant concern for users of multiple facilities. Many facilities offer support to improve the performance of an application on the platform(s) they field, but the staff often find it difficult to speak definitively about an application that they are not deeply familiar with and the application's performance on platforms other than their own. Thus, the onus for performance portability is on the users. In working with facility staff on performance, debugging, or other issues, it is best for everyone when such issues can be quickly isolated and addressed. This goal can be greatly facilitated by emphasizing software quality, documentation, and testing during the development process.