The understanding and prediction of transport due to plasma microturbulence is a key open problem in modern plasma physics, and a Grand Challenge for fusion energy research. Ab initio simulations of such small-scale, low-frequency turbulence are to be based on the gyrokinetic equations, a set of nonlinear integro-differential equations in reduced (five-dimensional) phase space. In the present paper, the extension of the well-established and widely used gyrokinetic code Gene [F. Jenko et al., Phys. Plasmas 7, 1904] from a radially local to a radially global (nonlocal) version is described. The necessary modifications of both the basic equations and the employed numerical methods are detailed, including, e.g., the change from spectral methods to finite difference and interpolation techniques in the radial direction and the implementation of sources and sinks. In addition, code verification studies and benchmarks are presented.

In the context of gyrokinetic flux-tube simulations of microturbulence in magnetized toroidal plasmas, different treatments of the magnetic equilibrium are examined. Considering the Cyclone DIII-D base case parameter set ͓Dimits et al., Phys. Plasmas 7, 969 ͑2000͔͒, significant differences in the linear growth rates, the linear and nonlinear critical temperature gradients, and the nonlinear ion heat diffusivities are observed between results obtained using either an s-␣ or a magnetohydrodynamic ͑MHD͒ equilibrium. Similar disagreements have been reported previously ͓Redd et al., Phys. Plasmas 6, 1162 ͑1999͔͒. In this paper it is shown that these differences result primarily from the approximation made in the standard implementation of the s-␣ model, in which the straight field line angle is identified to the poloidal angle, leading to inconsistencies of order ͑ = a / R is the inverse aspect ratio, a the minor radius and R the major radius͒. An equilibrium model with concentric, circular flux surfaces and a correct treatment of the straight field line angle gives results very close to those using a finite , low ␤ MHD equilibrium. Such detailed investigation of the equilibrium implementation is of particular interest when comparing flux tube and global codes. It is indeed shown here that previously reported agreements between local and global simulations in fact result from the order inconsistencies in the s-␣ model, coincidentally compensating finite ‫ء‬ effects in the global calculations, where ‫ء‬ = s / a with s the ion sound Larmor radius. True convergence between local and global simulations is finally obtained by correct treatment of the geometry in both cases, and considering the appropriate ‫ء‬ → 0 limit in the latter case.

The scaling of turbulence-driven heat transport with system size in magnetically confined plasmas is reexamined using first-principles based numerical simulations. Two very different numerical methods are applied to this problem, in order to resolve a long-standing quantitative disagreement, which may have arisen due to inconsistencies in the geometrical approximation. System size effects are further explored by modifying the width of the strong gradient region at fixed system size. The finite width of the strong gradient region in gyroradius units, rather than the finite overall system size, is found to induce the diffusivity reduction seen in global gyrokinetic simulations.

Abstract. The best hope for reducing long-standing global climate model biases is by increasing resolution to the kilometer scale. Here we present results from an ultrahigh-resolution non-hydrostatic climate model for a near-global setup running on the full Piz Daint supercomputer on 4888 GPUs (graphics processing units). The dynamical core of the model has been completely rewritten using a domain-specific language (DSL) for performance portability across different hardware architectures. Physical parameterizations and diagnostics have been ported using compiler directives. To our knowledge this represents the first complete atmospheric model being run entirely on accelerators on this scale. At a grid spacing of 930 m (1.9 km), we achieve a simulation throughput of 0.043 (0.23) simulated years per day and an energy consumption of 596 MWh per simulated year. Furthermore, we propose a new memory usage efficiency (MUE) metric that considers how efficiently the memory bandwidth – the dominant bottleneck of climate codes – is being used.

Currently major efforts are underway toward refining the horizontal resolution (or grid spacing) of climate models to about 1 km, using both global and regional climate models (GCMs and RCMs). Several groups have succeeded in conducting kilometer-scale multiweek GCM simulations and decadelong continental-scale RCM simulations. There is the well-founded hope that this increase in resolution represents a quantum jump in climate modeling, as it enables replacing the parameterization of moist convection by an explicit treatment. It is expected that this will improve the simulation of the water cycle and extreme events and reduce uncertainties in climate change projections. While kilometer-scale resolution is commonly employed in limitedarea numerical weather prediction, enabling it on global scales for extended climate simulations requires a concerted effort. In this paper, we exploit an RCM that runs entirely on graphics processing units (GPUs) and show examples that highlight the prospects of this approach. A particular challenge addressed in this paper relates to the growth in output volumes. It is argued that the data avalanche of high-resolution simulations will make it impractical or impossible to store the data. Rather, repeating the simulation and conducting online analysis will become more efficient. A prototype of this methodology is presented. It makes use of a bit-reproducible model version that ensures reproducible simulations across hardware architectures, in conjunction with a data virtualization layer as a common interface for output analyses. An assessment of the potential of these novel approaches will be provided.