Support community workshops, tutorials, and summer schools

NCAR plays an important role in helping scientific communities focus on the grand challenge problems of the Earth System. Targeted workshops with flexible formats that adapt to the audience and topics are particularly useful in highlighting areas where more research is needed. These events expose data scientists and computational scientists to the unique problems of climate and Earth System science. They also make NCAR scientists aware of new advances in computer hardware, computational science, and data science that can advance NCAR’s scientific mission and improve its facilities.

NCAR Multi-core 6 Workshop

A group of 42 computational experts familiar with creating and optimizing weather and climate applications, along with representatives from industry, gathered at the NCAR Mesa Lab on 13-14 September 2016 to attend the sixth Heterogeneous Multi-core Workshop. Their objective was to share and discuss the latest findings in programming for emerging disruptive computing technologies. This workshop is part of CISL’s ongoing processor technology tracking activity. It is focused on an emerging class of processors, now increasingly in the mainstream, with the common design feature that they trade off lower clock speed for a greatly increased CPU count, and typically offer a large amount of peak floating-point computing power in the form of many SIMD/vector processing elements. The result is a system design that is potentially much faster and more efficient in terms of energy per floating point calculation, but also one that can be much more difficult to program. By persisting year to year, the workshop helps to build a sense of community, create continuity, and foster a clearer understanding of the progress being made in both applications and architectures.

Multi-core 6 participants
The Multi-Core 6 workshop is held annually at NCAR and focuses on tracking the community’s progress in exploiting emerging microprocessors with very high levels of parallelism. The 42 participants came from South Korea, Germany, the United Kingdom, and the U.S. Key vendors in this tech sector and researchers from multiple U.S. government laboratories and agencies were also represented.

 

The NCAR workshop provided a forum for experts to share experiences and have open discussions that lead to an improved collective understanding of how these new technologies can be most useful to atmospheric science. The workshop specifically focused on the algorithms, programming models, design strategies, and tools that will be needed to create a new generation of applications to exploit these architectures. The workshop also enables this community of developers to participate in a forum where they can collectively provide technical feedback to vendors, develop and share necessary software tools, techniques, and standards, and further develop a new generation of applications for the benefit of Earth System science.

Three themes emerged from this year’s conference. First, multi-core system designs are beginning to appear with limited amounts of high-bandwidth memory (HBM). HBM provides O(1) terabyte per second to a new generation of many-core processors, such as Intel’s Knights Landing (KNL) and NVIDIA’s P100 “Pascal” system. As the latter are just becoming available, many presenters focused on KNL results and the intricacies of using HBM effectively. Second, the Multi-core 6 Workshop had multiple presenters reporting on the status of software tools, programming paradigms, and frameworks designed to ease the difficulty of programming and optimizing atmospheric applications on these systems. Finally, presentation of performance results for specific models and the intercomparison of processors and vendors continued to be a strong focus of the workshop.

Third Annual Graduate workshop on Environmental Data Analytics

Held 25-29 July 2016, this workshop is part of series designed to prepare the next generation of researchers and practitioners to work within, and contribute to, the data-rich era. Each workshop will bring together graduate students and senior scientists in environmental statistics and related fields to explore contemporary topics in applied environmental data modeling. This event hosted 27 graduate students in either statistics or ecology and guided them through hands-on computing and modeling tutorials.

This workshop also offered them an opportunity to share research findings and explore open questions within and at the interface of environmental, ecological, climatic, and statistical sciences. Finally, the workshop also included a visit to the National Ecological Observatory Network (NEON) headquarters to learn about NSF plans for the systematic monitoring, instrumentation, and data collection at the scale of the North American continent. The popularity and impact of this event is evidenced by an applicant pool of more than 50 students. Moreover, the lectures were given by prominent researchers in Bayesian methods and showcased state-of-the-art R packages for computation.

Funding

The Multi-core 6 Workshop is supported by NCAR Core and discretionary funds. The Third Annual Graduate workshop on Environmental Data Analytics was supported by NCAR Core funding plus an NSF CAREER Award 1253225 (Andrew Finley).