How will science at ALCF evolve over the next five years, toward our pre-exascale Aurora system? How might it continue to ten years out, when our exascale system is live? This presentation projects paths forward in the activities defining and surrounding scientific research using the ALCF.

Science: A few examples illustrate some of the exciting scientific questions to be answered, as we state those questions today and will expand or refine them going forward.

Data: Established big data efforts are making steps from the cloud into the HPC world we know in computational science, bringing adaptations of their software stack. Analyses of experimental data and simulation data are reaching leadership scale computing, defining new types of scientific projects for ALCF.

Compute Systems: A highly probable pair of architectural tracks gives a picture of the platforms for computational science through exascale.

Applications: LCF compute system architectures present the next challenges for developing and optimizing scientific applications. Addressing portability and embracing community codes will lead to new ways of providing and maintaining those applications.

Users: Maturing codes and new types of scientific projects may lead to new types of LCF users.

ALCF: The evolution of scientific and data-intensive applications using our systems will affect not just the science team, but also other parts of our organization—storage, networking, scheduling, support, and outreach.