Search

We introduce a new HPC system high-speed network fabric production monitoring tool, the ibnet sampler plugin for LDMS version 4. Large-scale testing of this tool is our work in progress. When deployed appropriately, the ibnet sampler plugin can provide extensive counter data, at frequencies up to 1 Hz. This allows the LDMS monitoring system to be useful for tracking the impact of new network features on production systems. We present preliminary results concerning reliability, performance impact, and usability of the sampler.

Productivity from day one on supercomputers that leverage new technologies requires significant preparation. An institution that procures a novel system architecture often lacks sufficient institutional knowledge and skills to prepare for it. Thus, the "Center of Excellence" (CoE) concept has emerged to prepare for systems such as Summit and Sierra, currently the top two systems in the Top 500. This paper documents CoE experiences that prepared a workload of diverse applications and math libraries for a heterogeneous system. We describe our approach to this preparation, including our management and execution strategies, and detail our experiences with and reasons for using different programming approaches. Our early science and performance results show that the project enabled significant early seismic science with up to a l4X throughput increase over Cori. In addition to our successes, we discuss our challenges and failures so others may benefit from our experience.

Foam materials are starting to find application in laser-heated Hohlraums used to drive inertial confinement fusion implosions. Foams made using additive manufacturing (AM) techniques are now available and may have advantages over traditional chemical (aerogel) foams. Here, we present new experimental data on laser-heated AM foams. Samples of four different types of printed AM foams were heated using a single 527 nm laser beam at the Jupiter Laser Facility. The laser pulse was ∼180 J square pulse with an FWHM of 1.6 ns and a peak intensity of 3–4 × 1014 W/cm2. The foam densities ranged from 12 to 93 mg/cc (all supercritical for 527 nm light). We measured the backscattered light (power and spectrum), the transmitted light, side-on x-ray images, and the Ti K-shell emission that was used to infer the time-integrated temperature. The fraction of backscattered light was 6%–15% of the input laser energy. The pure carbon foam sample had less backscatter than a C8H9O3 foam of similar density, which was consistent with multi-fluid calculations that predicted less ion heating for the C8H9O3 foam. The level of backscatter and the thermal front speeds for the AM foams were similar to values measured for stochastic (aerogel) foams under similar conditions.

We present a subgrid model for laser propagation in a subcritical foam. Our model describes the expansion of laser-irradiated foam elements that are below the resolution of the simulation grid and predicts the plasma conditions that result from burning down the foam. Our model can be included as a module within a larger multiphysics code, and we have implemented it within the code pF3D, which is used for simulating a laser-plasma interaction. The model predicts a reduced propagation velocity for a laser through a subcritical foam compared to simulating that foam as a homogeneous gas. This is attributed to the laser energy that goes into burning down the foam microstructure. We compare our model against experimental data by simulating a 2 mg/cc SiO2 foam shot performed at the Janus laser facility at the Lawrence Livermore National Laboratory. pF3D simulations with the foam model predict hot ion temperatures. This leads to a reduction in the level of stimulated Brillouin scattering (SBS), bringing the simulated level of SBS into agreement with the data. Intensity fluctuations at the foam front due to laser speckles and refraction result in ion temperature fluctuations when the foam burns down. These drive long-lived electron density fluctuations on scales that are large compared to the pore size.

We have begun to use 350–500 kJ of 1/3-micron laser light from the National Ignition Facility (NIF) laser to create millimeter-scale, bright multi-keV x-ray sources. In the first set of shots we achieved 15%–18% x-ray conversion efficiency into Xe M-shell (∼1.5–2.5 keV), Ar K-shell (∼3 keV) and Xe L-shell (∼4–5.5 keV) emission (Fournier et al., Phys. Plasmas 17, 082701, 2010), in good agreement with the emission modeled using a 2D radiation-hydrodynamics code incorporating a modern Detailed Configuration Accounting atomic model in non-LTE (Colvin et al., Phys. Plasmas, 17, 073111, 2010). In this paper we first briefly review details of the computational model and comparisons of the simulations with the Ar/Xe NIF data. We then discuss a computational study showing sensitivity of the x-ray emission to various beam illumination details (beam configuration, pointing, peak power, pulse shape, etc.) and target parameters (size, initial density, etc.), and finally make some predictions of how the x-ray conversion efficiency expected from NIF shots scales with atomic number of the emitting plasma.

We develop a model for laser propagation and heating in a subcritical foam (homogeneous electron density as a fraction of critical ne,0/nc

An assessment of short wavelength hydrodynamic stability is an essential component in the optimization of NIF ignition target designs. Using highly-resolved massively-parallel 2D Hydra simulations [Marinak, M.M. et al., Physics of Plasmas (1998). 5(4): 1125], we routinely evaluate target designs up to mode numbers of 2000 (λ∼2 μm) [Hammel, B.A. et al., Journal of Physics: Conference Series, 2008. 112(2): p. 02200]. On the outer ablator surface, mode numbers up to ∼300 (λ∼20 μm) can have significant growth in CH capsule designs. At the internal fuel:ablator interface mode numbers up to ∼2000 are important for both CH and Be designs. In addition, “isolated features” on the capsule, such as the “fill-tube” (∼5 μm scale-length) and defects, can seed short wavelength growth at the ablation front and the fuel:ablator interface, leading to the injection of ∼10's ng of ablator material into the central hot-spot. We are developing methods to measure high-mode mix on NIF implosion experiments. X-ray spectroscopic methods are appealing since mix into the hot-spot will result in x-ray emission from the high-Z dopant (Cu or Ge) in the ablator material (Be or CH).

With the next generation of high-power laser facilities for inertial fusion coming online1,2, ensuring laser beam propagation through centimetre-scale plasmas is a key physics issue for reaching ignition. Existing experimental results3,4,5 including the most recent one6 are limited to small laser spots, low-interaction laser beam energies and small plasma volumes of 1–2 mm. Here, we demonstrate the propagation of an intense, high-energy, ignition-size laser beam through fusion-size plasmas on the National Ignition Facility (NIF) and find the experimental measurements to agree with full-scale modelling. Previous attempts to apply computer modelling as a predictive capability have been limited by the inherently multiscale description of the full laser–plasma interaction processes7,8,9,10,11. The findings of this study validate supercomputer modelling as an essential tool for the design of future ignition experiments.

The capsules used in ignition experiments on the National Ignition Facility (NIF) laser will have a layer of frozen DT inside a low-Z shell. Liquid DT will be injected through a narrow fill tube that penetrates the shell and is frozen in place. The fill tube is a perturbation on the surface of the capsule and hydrodynamic instabilities will cause this perturbation to grow during an implosion. Experiments to investigate the growth of perturbations due to fill tubes have been carried out on the Omega laser. The goal of these experiments was to validate simulations at Omega energy scales and thus increase confidence in the use of simulations in planning for NIF experiments. Simulations show that the fill tube leads to a jet of shell material that penetrates into the DT fuel. Simulations will be used to pick experimental conditions in which the jet is small enough that it does not significantly reduce the yield of a NIF implosion. This paper compares experiments in which bumps and stalks were used as fill tube surrogates to 2D simulations of X-ray emission from Omega capsule implosions. Experiments and simulations are in reasonable agreement on the size of a bump or stalk required to produce a jet that is visible above the emission from a (nominally) smooth capsule.

HYDRA simulates a variety of experiments carried out at the National Ignition Facility and other high energy density physics facilities. It has packages to simulate radiation transfer, atomic physics, hydrodynamics, laser propagation, and a number of other physics effects. HYDRA has over one million lines of code, includes MPI and thread-level (OpenMP and pthreads) parallelism, has run on a variety of platforms for two decades, and is undergoing active development.

The first experiments on the National Ignition Facility (NIF) have employed the first four beams to measure propagation and laser backscattering losses in large ignition-size plasmas. Gas-filled targets between 2 and 7 mm length have been heated from one side by overlapping the focal spots of the four beams from one quad operated at 351 nm (3ω) with a total intensity of 2 × 1015 W cm−2. The targets were filled with 1 atm of CO2 producing up to 7 mm long homogeneously heated plasmas with densities of ne = 6 × 1020 cm−3 and temperatures of Te = 2 keV. The high energy in an NIF quad of beams of 16 kJ, illuminating the target from one direction, creates unique conditions for the study of laser–plasma interactions at scale lengths not previously accessible. The propagation through the large-scale plasma was measured with a gated x-ray imager that was filtered for 3.5 keV x-rays. These data indicate that the beams interact with the full length of this ignition-scale plasma during the last ~1 ns of the experiment. During that time, the full aperture measurements of the stimulated Brillouin scattering and stimulated Raman scattering show scattering into the four focusing lenses of 3% for the smallest length (~2 mm), increasing to 10–12% for ~7 mm. These results demonstrate the NIF experimental capabilities and further provide a benchmark for three-dimensional modelling of the laser–plasma interactions at ignition-size scale lengths.

A supervised machine learning algorithm trained on a multi-petabyte dataset of inertial confinement fusion simulations has identified a class of implosions that robustly achieve high yield, even in the presence of drive variations and hydrodynamic perturbations. These implosions are purposefully driven with a time-varying asymmetry, such that coherent flow generation during hotspot stagnation forces the capsule to self-organize into an ovoid, a shape that appears to be more resilient to shell perturbations than spherical designs. This new class of implosions, whose configurations are reminiscent of zonal flows in magnetic fusion devices, may offer a path to robust inertial fusion.

Six of the ten fastest supercomputers in the world in 2014 use a torus interconnection network for message passing between compute nodes. Torus networks provide high bandwidth links to near-neighbors and low latencies over multiple hops on the network. However, large diameters of such networks necessitate a careful placement of parallel tasks on the compute nodes to minimize network congestion. This paper presents a methodological study of optimizing application performance on a five-dimensional torus network via the technique of topology-aware task mapping. Task mapping refers to the placement of processes on compute nodes while carefully considering the network topology between the nodes and the communication behavior of the application. We focus on the IBM Blue Gene/Q machine and two production applications — a laser-plasma interaction code called pF3D and a lattice QCD application called MILC. Optimizations presented in the paper improve the communication performance of pF3D by 90% and that of MILC by up to 47%.

This paper examines whether lossy compression can be used effectively in physics simulations as a possible strategy to combat the expected data-movement bottleneck in future high performance computing architectures. We show that, for the codes and simulations we tested, compression levels of 3–5X can be applied without causing significant changes to important physical quantities. Rather than applying signal processing error metrics, we utilize physics-based metrics appropriate for each code to assess the impact of compression. We evaluate three different simulation codes: a Lagrangian shock-hydrodynamics code, an Eulerian higher-order hydrodynamics turbulence modeling code, and an Eulerian coupled laser-plasma interaction code. We compress relevant quantities after each time-step to approximate the effects of tightly coupled compression and study the compression rates to estimate memory and disk-bandwidth reduction. We find that the error characteristics of compression algorithms must be carefully considered in the context of the underlying physics being modeled.

Predictable performance is important for understanding and alleviating application performance issues; quantifying the effects of source code, compiler, or system software changes; estimating the time required for batch jobs; and determining the allocation requests for proposals. Our experiments show that on a Cray XE system, the execution time of a communication-heavy parallel application ranges from 28% faster to 41% slower than the average observed performance. Blue Gene systems, on the other hand, demonstrate no noticeable run-to-run variability. In this paper, we focus on Cray machines and investigate potential causes for performance variability such as OS jitter, shape of the allocated partition, and interference from other jobs sharing the same network links. Reducing such variability could improve overall throughput at a computer center and save energy costs.

We demonstrate the hohlraum radiation temperature and symmetry required for ignition-scale inertial confinement fusion capsule implosions. Cryogenic gas-filled hohlraums with 2.2 mm-diameter capsules are heated with unprecedented laser energies of 1.2 MJ delivered by 192 ultraviolet laser beams on the National Ignition Facility. Laser backscatter measurements show that these hohlraums absorb 87% to 91% of the incident laser power resulting in peak radiation temperatures of T(RAD)=300 eV and a symmetric implosion to a 100 μm diameter hot core.

Summary form only given. Experiments on the National Ignition Facility (NIF) have employed the first four beams to measure propagation and laser backscattering losses in large ignition-size plasmas. Gas-filled targets between 2 mm and 7 mm length have been heated from one side by overlapping the focal spots of the four beams from one quad operated at 351 nm with a total intensity of 2times1015 W cm-2. The targets were filled with 1 atm of CO2 producing of up to 7 mm long homogeneously heated plasmas with densities of ne = 5times1020 cm-3 and temperatures of Te = 2 keV. The high energy in a NIF quad of beams of 16 kJ, illuminating the target from one direction, creates unique conditions for the study of laser plasma interactions at scale lengths not previously accessible. The propagation through the large-scale plasma was measured with a gated X-ray imager that was filtered for 3.5 keV X-rays. These data indicate that the beams interact with the full length of this ignition-scale plasma during the last ~1 ns of the experiment when applying full laser beam smoothing consisting of phase plates, smoothing by spectral dispersion and polarization smoothing. Measurements that only apply phase plates show, laser beam filamentation and reduced propagation speed. These results demonstrate the NIF experimental capabilities and further provide a benchmark for three-dimensional modeling of the laser-plasma interactions at ignition-size scale lengths

Lawrence Livermore National Laboratory (LLNL) has a long history of working with IBM on Blue Gene® supercomputers. Beginning in November 2001 with the joint announcement of a partnership to expand the Blue Gene research project (including Blue Gene®/L and Blue Gene®/P), the collaboration extends to this day with LLNL planning for the installation of a 96-rack Blue Gene®/Q (called Sequoia) supercomputer. As with previous machines, we envision Blue Gene/Q will be used for a wide array of applications at LLNL, ranging from meeting programmatic requirements for certification to increasing our understanding of basic physical processes. We briefly describe a representative sample of mature codes that span this application space and scale well on Blue Gene hardware. Finally, we describe advances in multi-scale whole-organ modeling of the human heart as an example of breakthrough science that will be enabled with the Blue Gene/Q architecture.

Lasnex calculations and x-ray flux measurements are presented for a series of NIF vacuum hohlraum experiments that were among the first targets shot on NIF as part of the facility commissioning. An important result is that the hohlraum x-ray fluxes are significantly higher than predicted by pre-shot Lasnex calculations employing the baseline "configuration managed" physics packages used in the NIF ignition target calculations. A possible explanation for the high-flux vacuum hohlraum result has been explored via post-shot calculations in which non-baseline emissivity and heat conduction models are used.

The radiation hydrodynamics code Hydra is used to quantify the sensitivity of different NIF ignition point designs to several 3D effects. Each of the 48 NIF quads is included in the calculations and is allowed to have different power. With this model we studied the effect on imploded core symmetry of discrete laser spots (as opposed to idealized azimuthally-averaged rings) and random variations in laser power.

Cyclotron emission from a uniform plasma slab can be approximated by a Rayleigh-Jeans spectrum that cuts off sharply at an effective frequency mω, where ω is the cyclotron frequency. Accurate opacities are used to present significantly improved, numerically derived, formulae for m as a function of plasma temperatures in the range 5 ≤ kT ≤ 50 keV and the dimensionless plasma parameter Λ in the range 10 4 -10 8 . These formulae take account of the angular dependence of the cyclotron opacity and the high-frequency optically thin emission. In addition to being applicable to fusion plamas, these formulae are also valid for the much higher values of Λ characteristic of many astrophysical plasmas.