Waguter Wang

ABSTRACT: Magnetic fusion is a long-term solution for producing electrical power for the world, and the large thermonuclear international device being constructed will produce net energy and a path to fusion energy provided the computer modeling is accurate. To effectively address the requirements of the high-end fusion simulation community, application developers, algorithm designers, and hardware architects must have reliable simulation data gathered at scale for scientifically valid configurations. This paper presents detailed benchmarking results for a set of magnetic fusion applications with a wide variety of underlying mathematical models including both particle-in-cell and Eulerian codes using both implicit and explicit numerical solvers. Our evaluation on a petascale Cray XE6 platform focuses on profiling these simulations at scale identifying critical performance characteristics, including scalability, memory/network bandwidth limitations, and communication overhead. Overall results are a key in improving fusion code design, and are a critical first step towards exascale hardware-software co-design — a process that tightly couples applications, algorithms, imple- mentation, and computer architecture.

© 2015 The Physiological Society. The present study evaluated learning-related synaptic operations across the serial stages of the olfactory cortex-hippocampus network. Theta frequency stimulation produced very different time-varying responses in the Schaffer-commissural projections than in the lateral perforant path, an effect associated with distinctions in transmitter mobilization. Long-term potentiation had a higher threshold in LPP field potential studies but not in voltage clamped neurons; coupled with input/output relationships, these results suggest that LTP threshold differences reflect the degree of input divergence. Theta pulse stimulation erased LTP in CA1 but not in the dentate gyrus, although adenosine eliminated potentiation in both areas, suggesting that theta increases extracellular adenosine to a greater degree in CA1. Moreover, adenosine A1 receptor antagonism had larger effects on theta responses in CA1 than in the DG, and concentrations of ecto-5'-nucleotidase were much higher in CA1. Input/output curves for two connections in the piriform cortex were similar to those for the LPP, whereas adenosine modulation again correlated with levels of CD73. In sum, multiple relays in a network extending from the piriform cortex through the hippocampus can be differentiated along three dimensions that potently influence throughput and plasticity. A model that incorporates the regional differences, supplemented with data for three additional links, suggests that network output goes through three transitions during the processing of theta input. It is proposed that individuated relays allow the circuit to deal with different types of behavioural problems.