Call for Position Papers
Workshop on Modeling & Simulation of Systems and Applications
August 10-12, 2016 ♦ University of Washington, Seattle
Workshop URL: http://hpc.pnl.gov/modsim/2016/
Submission URL: https://easychair.org/conferences/?conf=modsim20160
EasyChair Submission Deadline: Friday, June 10, 2016 Now closed.
Submission Guidelines Download
All accepted paper and poster authors have been notified. Thank you for your contributions to ModSim 2016.
ModSim 2016 Topics
Advanced Architectures (including beyond CMOS)
Powerful components, such as GPU and RISC processors, 3D memory devices, and ASIC devices linked by increasingly sophisticated interconnection technologies, are frequently combined in novel and unproven ways. This practice, coupled with ever-present demands for greater execution and power efficiencies, lower acquisition costs, and higher reliability, creates even more complexity. ModSim can enable exploration of the most novel architectures, cultivating an understanding of what does and does not work (and why) and guiding complex engineering processes to successful conclusions.
Researchers are invited to submit abstracts describing their research, tools, and techniques for modeling and simulating computer architectures and components, particularly topics related to modeling architectures beyond CMOS, such as neuromorphic computing.
Integration of Measurement and ModSim
New architectures and systems offer a wealth of information regarding performance and power. However, understanding how to capture this information is often complex. Current state-of-the-art monitoring systems can point to various definitions of counter events, while different depth and breadth across vendors and significant idiosyncrasies only add to the complexity. Benchmarking is another measurement technique that offers a snapshot of system and application behavior at the application or microarchitecture-probe level.
In ModSim, a range of methodologies and tools employ measurements. Among them, model validation is vital. ModSim 2016 will take an in-depth look at the “virtuous circle” of measurement and modeling. Researchers are invited to submit abstracts related to: standardization of counter interfaces and definition of events, design of monitoring systems, monitoring systems architectures, trackable events, measurement tools and interfaces, ingesting measurement data into simulators, and measurement-based simulation verification and validation.
ModSim Methods and Tools
While the types and numbers of modeling and simulation tools and techniques continue to grow, their value and use still depend on deciding what users want ModSim to achieve. Whether solving a complex, real-world problem or enhancing general research and applications, interoperable, best-practice-based, validated models can contribute valuable understanding.
Researchers are invited to submit their abstracts related to ModSim methods and/or tools, including: interfaces, best practices, how different methodologies can cooperate, verification and validation supporting interoperability, and integration of ModSim for various layers of the hardware-software stack, to name only a few.
Modeling Integrated Workflows
Numerous science communities have adopted scientific workflow concepts and tools to manage large-scale science computations. These workflows automate data movement and processing tasks associated with collecting and analyzing data. As domain science communities increasingly rely on workflow systems to run complex tasks, it has become essential that these workflow systems operate correctly. ModSim techniques and methods must be used to provide this verification function. Researchers are invited to submit their abstracts related to the following areas: whole-model simulations; resource (computers, instruments, storage, or networks) selection and optimization; dynamic and interactive workflows; and other aspects of end-to-end data generation and movement (streaming data from detectors to storage/computers, bulk data transfers to/from storage systems, visualization or analysis data returned to scientists).