Meeting your Performance Portability goals with RAJA : A case study using CoMD – a proxy application for molecular dynamics. C/C++ with RAJA
Explore the source record for details and available documents.
Adedoyin, Adetokunbo Adelana↗
SEARCH · Search NASA
Results for “performance proxy”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
A flexible and differentiable coil proxy for stellarator equilibrium optimization
Balancing plasma performance and coil cost is a significant challenge when designing a stellarator power plant. Most current stellarator designs are produced through two-stage optimization: stage-1 for the equilibrium and stage-2 for a coil design that reproduces its magnetic configuration. Because few proxies connect both stages, two-stage optimization can produce plasmas that have high-quality physical properties but overly complex coils. In recent years, single-stage optimization has increasingly been used to optimize the plasma and coils simultaneously in order to improve the plasma–coil balance. However, all existing single-stage tools are specialized for filament coils, cannot model coil systems containing permanent magnets or dipole arrays, and continue to be challenged by numerical problems. The quasi-single-stage optimization finds a middle ground by integrating a coil optimization subproblem into stage-1 optimization. We present a flexible, differentiable coil complexity proxy based on the newly developed QUADCOIL coil optimization code. QUADCOIL is fast and can target realistic coil metrics and constraints that are unavailable to codes with comparable speed. We demonstrate the effectiveness and flexibility of the QUADCOIL proxy by presenting two QSS optimization studies. The first study produces a permanent magnet solution for the MUSE stellarator with 34% fewer magnets than previous solutions. The second study produces a coil solution for the ARIES-CS stellarator with 30% reductions in both peak and root-mean-square force.
Fu, Lanke [Princeton Plasma Physics Laboratory (PP↗
ASC Tri-lab Co-design Level 2 Milestone Report 2015
In 2015, the three Department of Energy (DOE) National Laboratories that make up the Advanced Sci- enti c Computing (ASC) Program (Sandia, Lawrence Livermore, and Los Alamos) collaboratively explored performance portability programming environments in the context of several ASC co-design proxy applica- tions as part of a tri-lab L2 milestone executed by the co-design teams at each laboratory. The programming environments that were studied included Kokkos (developed at Sandia), RAJA (LLNL), and Legion (Stan- ford University). The proxy apps studied included: miniAero, LULESH, CoMD, Kripke, and SNAP. These programming models and proxy-apps are described herein. Each lab focused on a particular combination of abstractions and proxy apps, with the goal of assessing performance portability using those. Performance portability was determined by: a) the ability to run a single application source code on multiple advanced architectures, b) comparing runtime performance between \native" and \portable" implementations, and c) the degree to which these abstractions can improve programmer productivity by allowing non-portable implementation details to be hidden from the application developer. This report captures the work that was completed for this milestone, and outlines future co-design work to be performed by application developers, programming environment developers, compiler writers, and hardware vendors.
97 MATHEMATICS AND COMPUTING↗
Inter-Procedural Resource Sharing in High Level Synthesis through Function Proxies
Modular design is becoming increasingly important in High Level Synthesis (HLS) flows. Current HLS flows generate hierarchical and modular designs that mimic the structure and call graph of input specification translating functions into modules. Function calls are translated instantiating the callee module in the data-path of its caller, allowing for resource sharing when the same function is called multiple times. However, if two different callers invoke the same function, current HLS flows cannot share the instance of the module between the two callers, even if they invoke the function in a mutually exclusive way. In this paper, we propose a methodology that enables sharing of (sub)modules across modules boundaries. Sharing is obtained through function proxies, which act as forwarders of function calls in the original specification to shared modules without adding performance penalties. Building on the concept of function proxies, we propose a methodology and the related components to perform HLS of function calls through function pointers, without requiring complete static knowledge of the alias set (point-to set). We show that module sharing through function proxies provides valuable area savings without significant impacts on the execution delays, and that our synthesis approach for function pointers enables dynamic polymorphism.
HLS, function proxy, FPGA, modular HLS,↗
Sensitivity analysis of numerical modeling input parameters on floating offshore wind turbine loads
Abstract. Floating wind turbines must withstand a unique and challenging set of loads from the wind and ocean environment. To de-risk development, accurate predictions of these loads are necessary. Uncertainty in modeling predictions leads to larger required safety factors, increasing project costs and the levelized cost of energy. Complex aero-hydro-elastic modeling tools use many input parameters to represent the wind, waves, current, aerodynamic loads, hydrodynamic loads, and structural properties. It is helpful to understand which of these parameters ultimately drives a design. In this work, an ultimate and fatigue-proxy load sensitivity analysis was performed with 35 different input parameters, using an elementary effects approach to identify the most influential parameters for a case study involving the National Renewable Energy Laboratory (NREL) 5 MW baseline wind turbine atop the OC4-DeepCwind semisubmersible during normal operation. The importance of each parameter was evaluated using 14 response quantities of interest across three operational wind speed conditions. The study concludes that turbulent wind velocity standard deviation is the parameter with the strongest sensitivity; this value is important not just for turbine loads, but also for the global system response. The system center of mass in the wind direction is found to have the highest impact on the system rotation and tower loads. The current velocity is found to be the most dominating parameter for the system global motion and consequently the mooring loads. All tested wind turbulence parameters in addition to the standard deviation are also found to be influential. Wave characteristics are influential for some fatigue-proxy loading but do not significantly impact the extreme ultimate loads in these operational load cases. The required number of random seeds for stochastic environmental conditions is considered to ensure that the sensitivities are due to the input parameters and not due to the seed. The required number of analysis points in the parameter space is identified so that the conclusions represent a global sensitivity. The results are specific to the platform, turbine, and choice of parameter ranges, but the demonstrated approach can be applied widely to guide focus in parameter uncertainty.
17 WIND ENERGY↗
The effects of dose, dose rate, and irradiation type and their equivalence on radiation-induced segregation in binary alloy systems via phase-field simulations
Radiation-induced segregation is a phenomenon commonly observed in many alloys which consists of the redistribution of elements (solute or interstitial impurities) under irradiation. The onset and development of radiation-induced segregation can only occur when a sufficient flux of defects is sustained and defect sinks are present. Irradiation dose, dose rate, and particle types all affect defect flux. In this work, we employ a phase-field model to examine the effects of dose, dose rate, and type of incident particles on radiation-induced segregation behavior in a model binary alloy. The phase-field model takes into account the formation and evolution of point defects as well as defect clusters, the diffusion and clustering of alloy species, the presence of additional extrinsic defect sinks in the form of dislocations, and two different methods of radiation-damage insertion, which are intended to simulate either light-ion/electron irradiation via Frenkel pairs or heavy-ion irradiation in the form of cascades. Our results show a dose-rate and particle-type dependence on the amount of solute segregation. We show that the material systems exposed to higher dose rates are less subjected to solute segregation at equivalent doses. We also show that such dose-rate-dependence behavior is due to a delay of the incubation dose at which radiation-induced segregation effectively starts. Particle type and the presence of dislocations can accentuate this behavior. Our model predictions correlate with many experimental observations made over the years on radiation-induced segregation providing credence to the simulation results. The methodology presented in this study allows for a first-order prediction of the dose rate at which proxy irradiation experiments could be performed to approximate radiation-induced segregation behaviors seen in targeted irradiation conditions.
36 MATERIALS SCIENCE↗
Performance Assessment of OpenMP Compilers Targeting NVIDIA V100 GPUs
Heterogeneous systems are becoming increasingly prevalent. In order to exploit the rich compute resources of such systems, robust programming models are needed for application developers to seamlessly migrate legacy code from today’s systems to tomorrow’s. Over the past decade and more, directives have been established as one of the promising paths to tackle programmatic challenges on emerging systems. This work focuses on applying and demonstrating OpenMP offloading directives on five proxy applications. We observe that the performance varies widely from one compiler to the other; a crucial aspect of our work is reporting best practices to application developers who use OpenMP offloading compilers. While some issues can be worked around by the developer, there are other issues that must be reported to the compiler vendors. By restructuring OpenMP offloading directives, we gain an 18x speedup for the su3 proxy application on NERSC’s Cori system when using the Clang compiler, and a 15.7x speedup by switching max reductions to add reductions in the laplace mini-app when using the Cray-llvm compiler on Cori.
Davis, Josh↗
A Simple Computational Proxy for Screening Magnetocaloric Compounds
Alternating cycles of isothermal magnetization and adiabatic demagnetization applied to a magnetocaloric material can drive refrigeration in very much the same manner as cycles of gas compression and expansion. The material property of interest in finding candidate magnetocaloric materials is ΔS M , their gravimetric entropy change upon application of a magnetic field under isothermal conditions. There is, however, no general method for screening materials for such an entropy change without actually performing the relevant, time- and effort-intensive magnetic measurements. Here we propose a simple computational proxy based on performing nonmagnetic and magnetic density functional theory calculations on magnetic materials. This proxy, which we term the magnetic deformation Σ M , is a measure of how much the unit cell deforms comparing with the relaxed structures with and without the inclusion of spin polarization. Σ M appears to correlate very well with experimentally measured magnetic entropy change values. Here, the proxy has been tested against 33 ferromagnetic materials with known ΔS M , including nine materials newly measured for this study. It has then been used to screen 134 ferromagnetic materials for which the magnetic entropy has not yet been reported, identifying 30 compounds as being promising for further study. As a demonstration of the effectiveness of our approach, we have prepared one of these compounds and measured its isothermal entropy change. MnCoP, with a T C of 575 K, shows a maximal ΔS M of –6.0 J kg –1 K –1 for an H = 5 T applied field.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Compare linear-system solver and preconditioner stacks with emphasis on GPU performance and propose phase-2 NGP solver development pathway
The goal of the ExaWind project is to enable predictive simulations of wind farms comprised of many megawatt-scale turbines situated in complex terrain. Predictive simulations will require computational fluid dynamics (CFD) simulations for which the mesh resolves the geometry of the turbines and captures the rotation and large deflections of blades. Whereas such simulations for a single turbine are arguably petascale class, multi-turbine wind farm simulations will require exascale-class resources. The primary physics codes in the ExaWind project are Nalu-Wind, which is an unstructured-grid solver for the acoustically incompressible Navier-Stokes equations, and OpenFAST, which is a whole-turbine simulation code. The Nalu-Wind model consists of the mass-continuity Poisson-type equation for pressure and a momentum equation for the velocity. For such modeling approaches, simulation times are dominated by linear-system setup and solution for the continuity and momentum systems. For the ExaWind challenge problem, the moving meshes greatly affect overall solver costs as reinitialization of matrices and recomputation of preconditioners is required at every time step. In this report we evaluated GPU-performance baselines for the linear solvers in the Trilinos and hypre solver stacks using two representative Nalu-Wind simulations: an atmospheric boundary layer precursor simulation on a structured mesh, and a fixed-wing simulation using unstructured overset meshes. Both strong-scaling and weak-scaling experiments were conducted on the OLCF supercomputer Summit and similar proxy clusters. We focused on the performance of multi-threaded Gauss-Seidel and two-stage Gauss-Seidel that are extensions of classical Gauss-Seidel; of one-reduce GMRES, a communication-reducing variant of the Krylov GMRES; and algebraic multigrid methods that incorporate the afore-mentioned methods. The team has established that AMG methods are capable of solving linear systems arising from the fixed-wing overset meshes on CPU, a critical intermediate result for ExaWind FY20 Q3 and Q4 milestones. For the fixed-wing strong-scaling study (model with 3M grid-points), the team identified that Nalu-Wind simulations with the new Trilinos and hypre solvers scale to modest GPU counts, maintaining above 70% efficiency up to 6 GPUs. However, there still remain significant bottlenecks to performance: matrix assembly (hypre), AMG setup (hypre and Trilinos) In the weak-scaling experiments (going from 0.4M to 211M gridpoints), it's shown that the solver apply phases are faster on GPUs, but that Nalu-Wind simulation times grow, primarily due to the multigrid-setup process. Finally, based on the report outcomes, we propose a linear solver path-forward for the remainder of the ExaWind project. Near term, the NREL team will continue their work on GPU-based linear-system assembly. They will also investigate how the use of alternatives to the NVIDIA UVM (unified virtual memory) paradigm affects performance. Longer term, the NREL team will evaluate algorithmic performance on other types of accelerators and merge their improvements back to the main hypre repository branch. Near term, the Trilinos team will address performance bottlenecks identified in this milestone, such as implementing a GPU-based segregated momentum solve and reusing matrix graphs across linear-system assembly phases. Longer term, the Trilinos team will do detailed analysis and optimization of multigrid setup.
17 WIND ENERGY↗
Performance Evaluation of Heterogeneous GPU Programming Frameworks for Hemodynamic Simulations
Preparing for the deployment of large scientific and engineering codes on upcoming exascale systems with GPU-dense nodes is made challenging by the unprecedented diversity of device architectures and heterogeneous programming models. In this work, we evaluate the process of porting a massively parallel, fluid dynamics code written in CUDA to SYCL, HIP, and Kokkos with a range of backends, using a combination of automated tools and manual tuning. We use a proxy application along with a custom performance model to inform the results and identify additional optimization strategies. At scale performance of the programming model implementations are evaluated on pre-production GPU node architectures for Frontier and Aurora, as well as on current NVIDIA device-based systems Summit and Polaris. Real-world workloads representing 3D blood flow calculations in complex vasculature are assessed. Our analysis highlights critical trade-offs between code performance, portability, and development time.
Martin, Aristotle↗
Implementing a neural network interatomic model with performance portability for emerging exascale architectures
The two main thrusts of computational science are increasingly accurate predictions and faster calculations; to this end, the zeitgeist in molecular dynamics (MD) simulations is pursuing machine learned and data driven interatomic models, e.g. neural network potentials, and novel hardware architectures, e.g. GPUs. Current implementations of neural network potentials are orders of magnitude slower than traditional interatomic models and while looming exascale computing offers the ability to run large, accurate simulations with these models, achieving portable performance for MD with new and varied exascale hardware requires rethinking traditional algorithms, using novel data structures, and library solutions. We re-implement a neural network interatomic model in CabanaMD, an MD proxy application, built on libraries developed for performance portability. Our implementation shows significantly improved thread scaling in this complex kernel as compared to a current LAMMPS implementation, across both strong and weak scaling. Our single-source solution enables simulations up to 20 million atoms on a single CPU node and 4 million atoms with improved performance on a single GPU. Furthermore, we also explore parallelism and data layout choices (using flexible data structures called AoSoAs) and their effect on performance, seeing up to ~50% and ~5% improvements in performance on a GPU by choosing the right level of parallelism and data layout respectively.
97 MATHEMATICS AND COMPUTING↗
Proxy Applications for Converged Workloads: DMC LDRD Initiative
Modern scientific applications are complicated and require coordination of several components. Proxy application driven software-hardware co-design plays a vital role in driving innovation among the developments of applications, software infrastructure and hardware architecture. Proxy applications are self-contained and simplified codes that are intended to model the performance-critical computations within applications. Applications executing on modern High Performance Computing (HPC) systems are susceptible to network congestion, insufficient memory bandwidth within and across compute nodes, and inadvertent loss of performance due to bugs and unoptimized programming models. Modern numerical simulations and machine learning models play a critical role in studying physical phenomenon under myriad uncertainties. Such applications often exhibit irregular computation and memory accesses at specific regions of the application code, which can contribute to various performance bottlenecks at scale. To mitigate such issues and prepare the next generation hardware for a variety of computation and data movement contingencies, a well-known practice is to consider "proxy" applications as representative motifs for various classes of scientific applications. While there is disagreement in the HPC community on the mechanisms of construction of the proxy applications, there is a strong consensus on their positive impact in co-design. Proxy Applications for Converged Workloads (PACER) is about facilitating software-hardware co-design through proxy applications with the goal of improving the performance of converged science workflows on heterogeneous systems.
97 MATHEMATICS AND COMPUTING↗
Hygroscopic growth and cloud droplet activation of xanthan gum as a proxy for marine hydrogels: HYGROSCOPICITY OF XG AS MARINE HYDROGEL
Knowledge of the physical characteristics and chemical composition of marine organic aerosols is needed for the quantification of their effects on cloud microphysical processes and solar radiative transfer. Here we use xanthan gum (XG)—a bacterial biopolymer—as a proxy for marine hydrogels. Measurements were performed for pure XG particles and mixtures of XG with sodium chloride, calcium nitrate, and calcium carbonate. The aerosol hygroscopicity parameter (κ) is derived from hygroscopic growth factor measurements (κ gf ) at variable water activity (a w ) and from cloud condensation nuclei activation efficiency (κ ccn ). The Zdanovskii, Stokes, and Robinson (ZSR) hygroscopicity parameter derived for multicomponent systems (κ mix , sol) is used to compare measurements of κ gf and κ ccn . Pure XG shows close agreement of κ gf (at aw = 0.9) and κccn of 0.09 and 0.10, respectively. Adding salts to the system results in deviations of κ gf (at a w = 0.9) from κ ccn . The measured κ gf and ZSR-derived hygroscopicity parameter (κ mix , sol) values for different solutions show close agreement at aw > 0.9, while κ gf is lower in comparison to κ mix , sol at a w < 0.9. The differences between predicted κmix, sol and measured κ gf and κ ccn values are explained by the effects of hydration and presence of salt ions on the structure of the polymer networks. Results from this study imply that at supersaturations of 0.1 and 0.5%, the presence of 30% sea salt by mass can reduce the activation diameter of pure primary marine organic aerosols from 257 to 156 nm and from 87 to 53 nm, respectively.
54 ENVIRONMENTAL SCIENCES↗
Flux balance analysis indicates that methane is the lowest cost feedstock for microbial cell factories
The low cost of natural gas has driven significant interest in using C 1 carbon sources (e.g. methane, methanol, CO, syngas) as feedstocks for producing liquid transportation fuels and commodity chemicals. Given the large contribution of sugar and lignocellulosic feedstocks to biorefinery operating costs, natural gas and other C 1 sources may provide an economic advantage. To assess the relative costs of these feedstocks, we performed flux balance analysis on genome-scale metabolic models to calculate the maximum theoretical yields of chemical products from methane, methanol, acetate, and glucose. Yield calculations were performed for every metabolite (as a proxy for desired products) in the genome-scale metabolic models of three organisms: Escherichia coli (bacterium), Saccharomyces cerevisiae (yeast), and Synechococcus sp. PCC 7002 (cyanobacterium). The calculated theoretical yields and current feedstock prices provided inputs to create comparative feedstock cost surfaces. Our analysis shows that, at current market prices, methane feedstock costs are consistently lower than glucose when used as a carbon and energy source for microbial chemical production. Conversely, methanol is costlier than glucose under almost all price scenarios. Acetate feedstock costs could be less than glucose given efficient acetate production from low-cost syngas using nascent biological gas to liquids (BIO-GTL) technologies. Furthermore, our analysis suggests that research should focus on overcoming the technical challenges of methane assimilation and/or yield of acetate via BIO-GTL to take advantage of low-cost natural gas rather than using methanol as a feedstock.
09 BIOMASS FUELS↗
ExaSAT: An exascale co-design tool for performance modeling
One of the emerging challenges to designing HPC systems is understanding and projecting the requirements of exascale applications. In order to determine the performance consequences of different hardware designs, analytic models are essential because they can provide fast feedback to the co-design centers and chip designers without costly simulations. However, current attempts to analytically model program performance typically rely on the user manually specifying a performance model. Here we introduce the ExaSAT framework that automates the extraction of parameterized performance models directly from source code using compiler analysis. The parameterized analytic model enables quantitative evaluation of a broad range of hardware design trade-offs and software optimizations on a variety of different performance metrics, with a primary focus on data movement as a metric. Finally, we demonstrate the ExaSAT framework’s ability to perform deep code analysis of a proxy application from the Department of Energy Combustion Co-design Center to illustrate its value to the exascale co-design process. ExaSAT analysis provides insights into the hardware and software trade-offs and lays the groundwork for exploring a more targeted set of design points using cycle-accurate architectural simulators.
97 MATHEMATICS AND COMPUTING↗
Design & Implementation of MACSio
We are proposing to develop “MACSio” (pronounced “max-ee-oh”), a Multi-purpose, Application-Centric, Scalable I/O proxy application. It is being proposed to fill a long existing void in co-design proxy applications that allow for I/O performance testing and evaluation of tradeoffs in data model interfaces and parallel I/O paradigms for multi-physics, HPC applications. Two key design features of MACSio will set it apart from existing I/O benchmarking tools. The first is the level of abstraction (LOA) at which MACSio is being designed to operate. The second is the degree of flexibility MACSio is being designed to provide in driving an HPC I/O workload through parameterized, user-defined data objects and a variety of parallel I/O paradigms and I/O interfaces. Combined, these features will allow MACSio to closely mimic I/O workloads for a wide variety of real applications and, in particular, multi-physics applications where data object distribution and composition vary dramatically both within and across parallel tasks. These data objects can then be marshaled using one or more I/O interfaces and parallel I/O paradigms, allowing for direct comparisons of software interfaces, parallel I/O paradigms, and filesystem technologies with the same set of customizable data objects.
97 MATHEMATICS AND COMPUTING↗
Understanding the Machine Learning Needs of ECP Applications
In order to support the codesign needs of ECP applications in current and future hardware in the area of machine learning, the ExaLearn team at Sandia studied the different machine learning use cases in three different ECP applications. This report is a summary of the needs of the three applications. The Sandia ExaLearn team will develop a proxy application representative of ECP application needs, specifically the ExaSky and EXAALT ECP projects. The proxy application will allow us to demonstrate performance portable kernels within machine learning codes. Furthermore, current training scalability of machine learning networks in these applications is negatively affected by large batch sizes. Training throughput of the network will increase as batch size increases, but network accuracy and generalization worsens. The proxy application will contain hybrid model- and data-parallelism to improve training efficiency while maintaining network accuracy. The proxy application will also target optimizing 3D convolutional layers, specific to scientific machine learning, which have not been as thoroughly explored by industry.
97 MATHEMATICS AND COMPUTING↗