Advance applied computational science research

Meeting the grand challenges in simulating the Earth System requires more than just migrating standard algorithms to larger computational platforms. New hardware, new parallel computational approaches, taking advantage of coprocessors, and more efficient algorithms are all needed to reach the resolution and complexity levels necessary to support scientific breakthroughs in modeling. This attention is also required to address the analysis and manipulation of the large data sets now common in the geosciences.

Advancing applied computational science research enhances the effective use of current and future computational systems by improving mathematical and computational methods for Earth System models and related observations.

The next three sections describe CISL’s efforts to accelerate NCAR software applications on existing as well as future hardware. In the past, application performance improvements came “automatically” – largely from advances in hardware performance. The last decade has seen the gradual end of this regime. Now the emphasis is on acceleration through increased parallelism. CISL research and development in this area has employed four strategies.

First, CISL launched efforts to accelerate the performance of NCAR’s computational models through parallelism. This means developing tools and techniques for achieving efficiency at higher thread counts and vector lengths than previously required. The target here is emerging many-core architectures such as Intel’s Xeon Phi and NVIDIA’s Pascal GPU architectures. This effort resulted in a considerable reduction of the overall computational cost of the production and high-resolution configurations of the Community Atmosphere Model. We have also optimized a version of the MPAS dynamical core that achieves a significantly reduced runtime on both Intel’s Xeon Phi and NVIDIA’s Pascal GPU architecture versus the current-generation Intel Xeon-based node used in Cheyenne.

Second, CISL continues to evaluate the impact that data compression may have on the fidelity of climate simulation data and evaluate different types of lossy compression approaches, both of which are not trivial given the diversity of data produced by CESM. Recent work explores the properties of two distinct lossy approaches (transform and predictive) in the context of CESM, a step toward the development of an automated multi-method approach. In addition, a study was completed that challenged the climate modeling community to test whether compression of model output degrades its scientific content.

Third, CISL focused on accelerating the end-to-end workflow for climate modeling. In the past, optimization efforts were focused on reducing the computational cost of the models. However, as computational science has become more data-centric, attention is now being given to reducing the execution time for data analysis and post processing. Several new parallel Python tools have been developed that reduce the post-processing time of climate simulation data by a factor of 10 to 100.

Finally, CISL’s numerical experts and computer scientists are working with scientists in other NCAR laboratories to pioneer new numerical schemes and parallel algorithms to achieve algorithmic acceleration. Conceptually, algorithmic acceleration means achieving the same numerical accuracy in less time or by using fewer cyber-resources. One example of this work is the development of a transport scheme for the cubed-sphere geometry that has high accuracy but still maintains positive concentrations. Moreover, this method only depends on neighboring elements and so does not degrade the parallelism in the other parts of the numerical procedures.

This work is supported by NSF Core funding, with supplemental funding supplied by other sources as noted in the following reports.