Projects

Paving the Way for Next-Gen Scientific Software

We are creating tools to simplify the transition to next-generation scientific software infrastructure.


Digital transformation concept

Supporting seamless access to cutting-edge scientific software components

Scientific software libraries provide an essential resource for high-quality, reusable software components that scientists and engineers can adapt for cutting-edge applications. However, much of this software infrastructure is quickly becoming obsolete, creating a gap that could slow scientific progress in the coming years. This project, titled “Collaborative Research: Frameworks: Basic ALgebra LIbraries for Sustainable Technology with Interdisciplinary Collaboration (BALLISTIC),” is creating tools to simplify the transition to next-generation scientific software infrastructure.

Several recent developments, including new system design constraints and high levels of heterogeneity, have made it clear that the essential software infrastructure of computational science and engineering needs to evolve. Math libraries have historically been in the vanguard of software that must be adapted first to such changes, both because these low-level workhorses are so critical to the accuracy and performance of so many different types of applications, and because they have proved to be outstanding vehicles for finding and implementing solutions to the problems that novel architectures pose.

With the BALLISTIC project, the principal designers of two key libraries, LAPACK and ScaLAPACK (abbreviated Sca/LAPACK), are working enhance and update these libraries for the ongoing revolution in processor architecture, system design, and application requirements. To accomplish this, we are incorporating them into a layered package of software components—the BALLISTIC ecosystem—that provides users seamless access to state-of-the-art solver implementations through familiar and improved Sca/LAPACK interfaces. The resulting new software components will be capable of running at every level of the hardware hierarchy by delivering seamless access to the most up-to-date algorithms, numerics, and performance.

Outcomes

Publications


Resources


About

Sponsors


  • National Science Foundation

Focus Areas


  • Machine Learning (Supervised, Unsupervised, Reinforcement Learning)

Get in touch

Want to discuss opportunities to work with ICSI? We’d love to hear from you.

2150 Shattuck Ave., #250
Berkeley, CA 94704

+1 (510) 666-2900

contact @ icsi.berkeley.edu