The Smoky Mountains Computational Sciences and Engineering Conference (SMC2021) is a premier event for discussing the latest developments in computational sciences and engineering for high-performance computing (HPC) and integrated instruments for science. The conference has been held since 2003. This year, the 19th installment of the conference is virtual and in person at the MeadowView Marriott Resort & Convention Center, Kingsport, Tennessee, USA. The conference theme is “Driving future science and engineering discoveries through the integration of experiment, big data, and modeling and simulation—that focus on accelerated node high-performance computing and integrated instruments for science (edge computing). This year, the program committee will accept vision papers that include authors’ perspectives on the most important directions for research, development, production, and experiences, and needs for investment. We specifically encourage authors to emphasize their positions, grounded in evidence, in the specific areas identified in the sessions below.
Abstract submission and paper registration due date *extended*: March 9, 2021 (submit your abstracts via EasyChair)
Abstracts need to be uploaded via this easy-chair link: https://easychair.org/conferences/?conf=smc20210
Author notification for abstract acceptance: April 2, 2021
Paper submission for review: June 1, 2021
Author notification for paper acceptance: July 16, 2021
Conference ready paper submission: August 3, 2021
Conference presentation: August 24-26th, 2021
Camera-ready paper submission: September 14th, 2021
Session chairs – Olga Ovchinnikova, ORNL and Juan Restrepo, ORNL
This session will address applications that embrace data-driven and first-principle methods, focusing on converging AI methods and approaches with high-performance modeling and simulation applications. Topics will include experiences, algorithms, and numerical methods development and integration with the edge. This session also focuses on mixed-precision, data reduction methods, and scientific libraries and frameworks for converged HPC and AI. Participants will discuss how simulation can be used to train AI models and integrate them to work with simulation applications while quantifying errors.
Session chairs – Teja Kuruganti and Olga Kuchar, ORNL
Participants will discuss multi-domain applications that use federated scientific instruments with data sets and large-scale compute capabilities, including sensors, actuators, instruments for HPC systems, data stores, and other network-connected devices. Some of the AI and HPC workloads are being pushed to the edge (closer to the instruments) while large-scale simulations are scheduled on HPC systems with large capacities. This session will focus on applications that focus on integration across domains and scientific datasets that combine AI and HPC with edge computing.
Session chairs – Arjun Shankar and Neena Imam, ORNL
This session includes programming systems and software technologies for novel computing processors such as neuromorphic, automata, advanced FETs, carbon nanotube processors, and other types of accelerators that meet the SWaP constraints to be deployed at the edge. To connect instruments from the edge to supercomputers, we need to efficiently collect and process data at the edge. Specialized workflows, efficient networks, data transfer toolkits, and communication libraries need to be developed to minimize the latency between edge and supercomputers and close the AI/learning and control loops. This session will present the latest ideas and findings in the programming and software ecosystems for these rapidly changing and emerging fields.
Session chairs – Scott Atchley and David Bernholdt, ORNL
Topics include industry experience and plans for deploying both hardware and software infrastructure needed to support emerging AI and/or classical simulation workloads; for combining on-premises and cloud resources; and for connecting distributed experimental, observational, and data resources and computing facilities using edge technologies. This session will focus on how emerging technologies can be co-designed to support compute and data workflows at scale for next-generation HPC and AI systems.
For more information about the sessions, contact firstname.lastname@example.org
All contributions are planned to be published in SMC2021 proceedings in a CCIS Springer volume (pending approval) and will be peer-reviewed by the program committee. Authors should clearly identify which of the four sessions described above their paper is targeting. Papers that do not fit into a session (either by topic or due to the number of papers accepted for a session) will be considered for short presentations in a poster session. All authors must first submit a 250-word abstract to register their papers. Once the abstract is accepted, we will encourage the authors to submit full papers. We will accept full papers of 12-18 pages. Papers need to be formatted according to Springer’s single-column style. Please use the paper templates available for LaTeX and Word (https://www.springer.com/gp/authors-editors/conference-proceedings/conference-proceedings-guidelines). The copyright will need to be transferred to Springer. A copyright form will be provided to authors which allows users to self-archive.
Abstracts and papers need to be uploaded here: https://easychair.org/conferences/?conf=smc20210
Data Challenge chair – Pravallika Devineni, ORNL
SMC2021 provides an opportunity to tackle scientific data challenges that come from eminent data sets at ORNL. These data sets come from scientific simulations and instruments in physical and chemical sciences, electron microscopy, bioinformatics, neutron sources, urban development, and other areas. These data sets will be used for the SMC Data Challenge (SMCDC2021) competition. For more information please visit: https://smc-datachallenge.ornl.gov