Search NASA⌕ Search

SEARCH · Search NASA

Results for “computational capability”

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.

At least 343 records · Page 19
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Machine Learning-enabled Scalable Performance Prediction of Scientific Codes

Hardware architectures become increasingly complex as the compute capabilities grow to exascale. Here, we present the Analytical Memory Model with Pipelines (AMMP) of the Performance Prediction Toolkit (PPT). PPT-AMMP takes high-level source code and hardware architecture parameters as input and predicts runtime of that code on the target hardware platform, which is defined in the input parameters. PPT-AMMP transforms the code to an (architecture-independent) intermediate representation, then (i) analyzes the basic block structure of the code, (ii) processes architecture-independent virtual memory access patterns that it uses to build memory reuse distance distribution models for each basic block, and (iii) runs detailed basic-block level simulations to determine hardware pipeline usage. PPT-AMMP uses machine learning and regression techniques to build the prediction models based on small instances of the input code, then integrates into a higher-order discrete-event simulation model of PPT running on Simian PDES engine. We validate PPT-AMMP on four standard computational physics benchmarks and present a use case of hardware parameter sensitivity analysis to identify bottleneck hardware resources on different code inputs. We further extend PPT-AMMP to predict the performance of a scientific application code, namely, the radiation transport mini-app SNAP. To this end, we analyze multi-variate regression models that accurately predict the reuse profiles and the basic block counts. We validate predicted SNAP runtimes against actual measured times.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2011

Base Flow Investigation of the Apollo AS-202 Command Module

In recent years, both Europe and the US are developing hypersonic research and operational vehicles. These include (re)entry capsules (both ballistic and lifting) and lifting bodies such as ExoMars, EXPERT, ARV, CEV and IXV. The research programs are meant to enable technology and engineering capabilities to support during the next decade the development of affordable (possibly reusable) space transportation systems as well as hypersonic weapons systems for time critical targets. These programs have a broad range of goals, ranging from the qualification of thermal protection systems, the assessment of RCS performances, the development of GNC algorithms, to the full demonstration of the performance and operability of the integrated vehicles. Since the aerothermodynamic characteristics influence nearly all elements of the vehicle design, the accurate prediction of the aerothermal environment is a prerequisite for the design of efficient hypersonic systems. Significant uncertainties in the prediction of the hypersonic aerodynamic and the aerothermal loads can lead to conservative margins in the design of the vehicle including its Outer Mould Line (OML), thermal protection system, structure, and required control system robustness. The current level of aerothermal prediction uncertainties results therefore in reduced vehicle performances (e.g., sub-optimal payload to mass ratio, increased operational constraints). On the other hand, present computational capabilities enable the simulation of three dimensional flow fields with complex thermo-chemical models over complete trajectories and ease the validation of these tools by, e.g., reconstruction of detailed wind tunnel tests performed under identified and controlled conditions (flow properties and vehicle attitude in particular). These controlled conditions are typically difficult to achieve when performing in flight measurements which in turn results in large associated measurement uncertainties. Similar problems arise when attempting to rebuild measurements performed in "hot" ground facilities, where the difficulty level is increased by the addition of the free-flow characterization itself. The implementation of ever more sophisticated thermochemical models is no obvious cure to the aforementioned problems since their effect is often overwhelmed by the large measurement uncertainties incurred in both flight and ground high enthalpy facilities. Concurrent to the previous considerations, a major contributor to the overall vehicle mass of re-entry vehicles is the afterbody thermal protection system. This is due to the large acreage (equal or bigger than that of the forebody) to be protected. The present predictive capabilities for base flows are comparatively lower than those for windward flowfields and offer therefore a substantial potential for improving the design of future re-entry vehicles. To that end, it is essential to address the accuracy of high fidelity CFD tools exercised in the US and EU, which motivates a thorough investigation of the present status of hypersonic flight afterbody heating. This paper addresses the predictive capabilities of after body flow fields of re-entry vehicles investigated in the frame of the NATO/RTO - RTG-043 Task Group and is structured as follows: First, the verification of base flow topologies on the basis of available wind-tunnel results performed under controlled supersonic conditions (i.e., cold flows devoid of reactive effects) is performed. Such tests address the detailed characterization of the base flow with particular emphasis on separation/reattachment and their relation to Mach number effects. The tests have been performed on an Apollo-like re-entry capsule configuration. Second, the tools validated in the frame of the previous effort are exercised and appraised against flight-test data collected during the Apollo AS-202 re-entry.

Walpot, Louis M. G.↗
DOE OSTI2024

Aggregate data‐driven dynamic modeling of active distribution networks with DERs for voltage stability studies

Abstract Electric distribution networks increasingly host distributed energy resources based on power electronic converter (PEC) toward active distribution networks (ADN). Despite advances in computational capabilities, electromagnetic transient models are limited in scalability because of their reliance on exact data about the distribution system and each of its components. Similarly, the use of the DER_A model, which is intended to examine the combined dynamic behavior of many DERs, is limited by the difficulty in parameterization. There is a need for improved dynamic models of DERs for use in large power system simulations for stability analysis. This paper proposes an aggregate model‐free, data‐driven approach for deriving a dynamic partitioned model (DPM) of ADNs. Detailed residential distribution feeders were first developed, including PEC‐based DERs and composite load models (CMLDs), from which the aggregated DPM was derived. The performance was evaluated through various case studies and validated against the detailed ADN model and state‐of‐the‐art DER_A model with CMLD. The data‐driven DPM achieved a of over 90%, accurately representing the aggregated dynamic behavior of ADNs. Furthermore, the DPM significantly accelerated the simulation process with a computational speedup of 68 times compared to the detailed ADN and a 3.5 times speedup compared to the DER_A CMLD model.

42 ENGINEERING↗
NASA NTRSDate not supplied

Engineering the Interface: Advanced Surface Technologies for Lunar Dust Management and Equipment Longevity

Through the Artemis program, NASA intends to develop a sustainable human foothold on the Moon, ultimately paving the way for crewed exploration of Mars. The Moon's hostile environment poses numerous obstacles, including exposure to radiation, temperature extremes, micrometeoroid threats, and particularly the persistent problem of lunar dust. Lunar dust impacts nearly every aspect of surface operations through adhesion and abrasion mechanisms, with contamination from anthropogenic activities (landing, rovers) far outweighing natural phenomena. Multiple adhesion pathways contribute to surface contamination in the lunar environment, including van der Waals forces, electrostatic forces, chemical reaction, and magnetic forces from elemental iron deposits. Sharp asperities from micrometeoroid bombardment and atmospheric absence increase interaction potential and enable mechanical interlocking. Low cohesion between dust particles exacerbates these challenges, as minimal interaction potential between dust and nearby surfaces overcomes particle cohesion, causing contamination. Lunar dust adhesion mitigation technologies can be categorized as either active, requiring external energy, or passive, relying on intrinsic material properties. Ultrasonic and electrodynamic technologies have been developed to the highest technology readiness level for active approaches. Passive strategies primarily focus on surface chemistry and topography modifications. At NASA Langley Research Center, approaches include surface migration agents to reduce surface energy, topographical modification using laser ablation patterning, and tailored surface conductivity to reduce intrinsic adhesion force. Performance has been evaluated using custom-built ultrasonic and centrifuge instruments. Plume-surface interactions from lunar landers can propel micrometer-sized particles at velocities up to 1000 m s-1.8 These particles pose risks to landers, habitats and infrastructure, leading to erosion, degradation, and reduced component lifespan. A panel recovered from Surveyor III was determined to have been severely abraded because of lunar dust displaced from the Apollo 12 lunar module that landed 160 m away. The performance of metallic surfaces has been evaluated via high velocity single particle impact using the laser-induced project impact test (LIPIT) facility at the University of Utah. Peridynamics modeling, a form of continuum mechanics that uses a nonlocal approach enabling greater simulation capabilities of crack initiation and fracture, has also been utilized to gain greater insight into material response during impact events. Lunar dust contamination challenges extend to power generation systems and moving equipment. Cables, rotation stages, and other mechanisms may experience limited range of motion and reduced lifetime due to dust infiltration. NASA Langley Research Center has evaluated traditional aerospace alloys, softgoods, wear resistant ceramics, and several polymer and polymer composite materials. Test methods have included traditional techniques like Taber abrasion testing, as well as designed test configurations developed in the DUSTE (dust, ultraviolet radiation, and space thermal environmental) chamber that reproduce mechanism functions in operational environment. Beyond laboratory experiments, several flight experiments have been conducted. Materials were exposed to the low Earth orbit environment on the Materials International Space Station Experiment (MISSE) and to the lunar surface environment through the Aegis Aerospace Regolith Adherence Characterization (RAC) payload and the Honeybee Robotics PlanetVac payload. Determining lunar dust's impact on surface exploration and habitation requires comprehensive experimental and computational capabilities combined with lessons learned from initial lunar activities. Identifying the greatest environmental challenges and developing mitigation technologies provides the clearest path toward successfully, expeditiously, and efficaciously completing NASA's mission. This presentation will discuss ongoing efforts at NASA Langley Research Center and collaborator contributions to these critical objectives.

Surface Engineering↗
NASA NTRS1978

CELFE: Coupled Eulerian-Lagrangian Finite Element program for high velocity impact. Part 1: Theory and formulation

A 3-D finite element program capable of simulating the dynamic behavior in the vicinity of the impact point, together with predicting the dynamic response in the remaining part of the structural component subjected to high velocity impact is discussed. The finite algorithm is formulated in a general moving coordinate system. In the vicinity of the impact point contained by a moving failure front, the relative velocity of the coordinate system will approach the material particle velocity. The dynamic behavior inside the region is described by Eulerian formulation based on a hydroelasto-viscoplastic model. The failure front which can be regarded as the boundary of the impact zone is described by a transition layer. The layer changes the representation from the Eulerian mode to the Lagrangian mode outside the failure front by varying the relative velocity of the coordinate system to zero. The dynamic response in the remaining part of the structure described by the Lagrangian formulation is treated using advanced structural analysis. An interfacing algorithm for coupling CELFE with NASTRAN is constructed to provide computational capabilities for large structures.

Lee, C. H.↗
DOE OSTI2022

Universal control of a six-qubit quantum processor in silicon

Future quantum computers capable of solving relevant problems will require a large number of qubits that can be operated reliably. However, the requirements of having a large qubit count and operating with high fidelity are typically conflicting. Spins in semiconductor quantum dots show long-term promise but demonstrations so far use between one and four qubits and typically optimize the fidelity of either single- or two-qubit operations, or initialization and readout. Here, we increase the number of qubits and simultaneously achieve respectable fidelities for universal operation, state preparation and measurement. We design, fabricate and operate a six-qubit processor with a focus on careful Hamiltonian engineering, on a high level of abstraction to program the quantum circuits, and on efficient background calibration, all of which are essential to achieve high fidelities on this extended system. State preparation combines initialization by measurement and real-time feedback with quantum-non-demolition measurements. These advances will enable testing of increasingly meaningful quantum protocols and constitute a major stepping stone towards large-scale quantum computers.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

cclib 2.0: An updated architecture for interoperable computational chemistry

Interoperability in computational chemistry is elusive, impeded by the independent development of software packages and idiosyncratic nature of their output files. The cclib library was introduced in 2006 as an attempt to improve this situation by providing a consistent interface to the results of various quantum chemistry programs. The shared API across programs enabled by cclib has allowed users to focus on results as opposed to output and to combine data from multiple programs or develop generic downstream tools. Initial development, however, did not anticipate the rapid progress of computational capabilities, novel methods, and new programs; nor did it foresee the growing need for customizability. Here, we recount this history and present cclib 2, focused on extensibility and modularity. We also introduce recent design pivots—the formalization of cclib’s intermediate data representation as a tree-based structure, a new combinator-based parser organization, and parsed chemical properties as extensible objects.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Atomistic Simulations for Thermophysical Properties of Uranium-Containing Halide Molten Salts

Characterizing the thermophysical properties in both fuel and coolant salts are critical in modeling, developing, process optimizing and utilizing molten salt reactors (MSRs), as these properties directly relate to operation metrics and can inform on the selection of candidate salts. The demand for consistent, accurate and publicly available thermophysical property data has become more apparent in recent years as interests have increased from molten salt reactor developers. There are a number of challenges in experimentally measuring properties such as thermal conductivity, viscosity, density and heat capacity , which have led to sparse and often times conflicting data points or molten salts in general. Additionally, there are a number of hazards to consider when synthesizing, storing, using, treating and disposing of molten salts. With the advances in computational capabilities over the last 10 years, the use of atomistic simulations can be implemented to support these efforts. The primary objective of this work is characterize the thermophysical transport properties in a number of molten chloride salts, and in particular NaCl-UCl 3 using ab-initio molecular dynamic (AIMD) simulations. In this binary salt the UCl 3 acts as the primary fissile material and NaCl acts as a carrier salt due with its’ high solubility for actinides A number of studies on the thermophysical properties of NaCl-UCl 3 have been published but there is not a vast amount of viscosity data for this system. In 1975, Desyatnik, et al published a study reporting dynamic viscosities that were calculated from kinematic viscosity measurements, and using the coefficients provided the viscosity in a 70:30 NaCl:UCl 3 mixture is 2.29 cP and 2.88 for a 60:40 mixture. Termini et al. recently reported viscosities in the range of 2.75 – 3 cP for the 63:37 NaCl-UCl 3 mixture in the same temperature range using rolling ball viscosity measurements. Computational viscosity of a similar mixture (64:36) can be obtained from the work Andersson et al. using the reported diffusion coefficients, and the hydrodynamic radius from the pair-radial distribution functions (RDFs). Using Eq (1) (vida infra), the viscosity would be 2.50 cP at 1100K. This is not to say that these values are incorrect due to the varying reported values, but aims to highlight the necessity of this work. The data reported in this ongoing work are computations on a 64:36 mixture of NaCl-UCl 3 at 987K. This work is likely to be expanded into varying concentrations of this mixture along with the inclusion of other salt candidate mixtures.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗
NASA NTRSDate not supplied

Low Latency Flux and Concentration Datasets in Support of Greenhouse Gas Monitoring Based on NASA's GEOS Modeling and Data Assimilation System

We present efforts to develop space-based greenhouse gas monitoring systems that can provide low latency information and traceability to independent observations. Through support from its Carbon Monitoring System program, NASA has developed the capability to assimilate XCO2 retrievals from the Orbiting Carbon Observatory, 2 (OCO-2) into the Goddard Earth Observing System (GEOS) Constituent Data Assimilation System (CoDAS) to create gap-filled, three-dimensional (3D) estimates of CO2 mixing ratio. When OCO-2 data are not available, concentration fields are further informed by a bottom-up flux package based on remotely sensed fire radiative power, nighttime lights, and vegetation reflectance combined with estimates of atmospheric growth rate based on surface in situ data. The 3D nature of this dataset supports evaluation with independent aircraft data, helping to ensure transparency of remotely sensed data products. These quasi-operational data are currently produced 2-3 months behind real time and are distributed via NASA and international dashboard services to a variety of end users. In this presentation, we provide an overview of the system as well as remaining data gaps and modeling challenges. We also highlight the application of this dataset for detecting emissions anomalies associated with COVID-19 and comparing against independent emissions estimates. Finally, we highlight a new NASA initiative called the Earth Information System (EIS), which aims to support open science and applications by leveraging emerging cloud computing capabilities to increase access to NASA’s greenhouse gas datasets, opportunities for co-development, and transparency in methods for analysis and flux attribution.

Lesley Ott↗
DOE OSTI2023

Security and Trust Metrics for Edge Computing

The present state of edge computing is an environment of different computing capabilities connected via a wide variety of communication paths. The energy grid is relying upon distributed energy devices connected at the edge of the internet. Consider the scenario where each edge device is customer-owned distributed energy resource (DER) that is connected via a trustworthy link to a grid service provider. Each DER keeps a local simple trust record of interactions. Information protection is provided by internet https standards, however, trust must be evaluated throughout operation. This paper presents a model for representing and evaluating trust in general and applied to the energy grid as a key example. Actors on the edge may interact with each other as well as with a central datacenter.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2021

Enabling Combustion Science Simulations for Future Exascale Machines

Reacting flow simulations for combustion applications require extensive computing capabilities. Leveraging the AMReX library, the Pele suite of combustion simulation tools targets the largest supercomputers available and future exascale machines. We introduce PeleC, the compressible solver in the Pele suite, and detail its capabilities, including complex geometry representation, chemistry integration, and discretization. We present a comparison of development efforts using both OpenACC and AMReX's C++ performance portability framework for execution on multiple GPU architectures. We discuss relevant details that have allowed PeleC to achieve high performance and scalability. PeleC's performance characteristics are measured through relevant simulations on multiple supercomputers. The success of PeleC's design for exascale is exhibited through demonstration of a 160 billion cell simulation and weak scaling onto 100\% of Summit, an NVIDIA-based GPU supercomputer at Oak Ridge National Laboratory. Our results provide confidence that PeleC will enable future combustion science simulations with unprecedented fidelity.

combustion↗
NASA NTRS1987

Analytical determination of propeller performance degradation due to ice accretion

A computer code capable of computing the propeller performance for clean, glaze, or rime ice propeller configurations to determine the performance degradation resulting from a given icing encounter has been developed. The inviscid, incompressible flowfield at each specified propeller radial location is first computed using the Theodorsen method. A droplet trajectory computation then calculates the droplet impingement points and airfoil collection efficiency for each radial location. User-selectable empirical correlations are available for determining the aerodynamic penalties due to ice accretion. Propeller performance is finally computed using strip analysis for either the clean or iced propeller. In the iced mode, the thrust and torque coefficient equations are modified by the drag and lift coefficient increments due to ice to obtain the appropriate iced values. Comparison with available experimental propeller icing data shows generally good agreement. The code's capability of properly predicting the thrust coefficient, power coefficient, and propeller efficiency of an iced propeller is shown to be dependent on the choice of empirical correlation employed as well as on the proper specification of the radial icing extent and propeller blade angle.

Miller, T. L.↗
DOE OSTI2022

PeleC: An adaptive mesh refinement solver for compressible reacting flows

Reacting flow simulations for combustion applications require extensive computing capabilities. Leveraging the AMReX library, the Pele suite of combustion simulation tools targets the largest supercomputers available and future exascale machines. We introduce PeleC, the compressible solver in the Pele suite, and detail its capabilities, including complex geometry representation, chemistry integration, and discretization. We present a comparison of development efforts using both OpenACC and AMReX’s C++ performance portability framework for execution on multiple GPU architectures. We discuss relevant details that have allowed PeleC to achieve high performance and scalability. PeleC’s performance characteristics are measured through relevant simulations on multiple supercomputers. The success of PeleC’s design for exascale is exhibited through demonstration of a 160 billion cell simulation and weak scaling onto 100% of Summit, an NVIDIA-based GPU supercomputer at Oak Ridge National Laboratory. Our results provide confidence that PeleC will enable future combustion science simulations with unprecedented fidelity.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Feature Analysis, Tracking, and Data Reduction: An Application to Multiphase Reactor Simulation MFiX-Exa for In-Situ Use Case

As we enter the exascale computing regime, powerful supercomputers continue to produce much higher amounts of data than what can be stored for offline data processing. To utilize such high compute capabilities on these machines, much of the data processing needs to happen in situ, when the full high-resolution data is available at the supercomputer memory. In this article, we discuss our MFiX-Exa simulation, which models multiphase flow by tracking a very large number of particles through the simulation domain. In one of the use cases, the carbon particles interact with air to produce carbon dioxide bubbles from the reactor. These bubbles are of primary interest to the domain experts for these simulations. For this particle-based simulation, we propose a streaming technique that can be deployed in situ to efficiently identify the bubbles, track them over time, and use them to down-sample the data with minimal loss in these features.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Improving the Accessibility and Usability of Geothermal Information with Data Lakes and Data Pipelines on the Geothermal Data Repository

The Geothermal Data Repository (GDR) provides universal access to data and information resulting from research and development activities funded by the Department of Energy (DOE). The GDR has extended this universal access to big data through integration with data lakes developed by the Open Energy Data Initiative (OEDI). Previously, large datasets such as seismic waveform or distributed acoustic sensing (DAS) data could only be accessed by institutions with high performance data storage and compute capabilities, effectively limiting the accessibility of big data to national labs, larger universities, and major corporations. Moreover, the time and resources needed to transport big data and configure them can produce additional barriers to use. Many of the standard formats used for structured data models (also known as content models) are incapable of handling big data and can introduce additional usability problems, often requiring data to be reformatted prior to use. This paper will explore how recent integrations between the GDR and the OEDI data lake have improved the accessibility and usability of geothermal data in a big way, making the data available to a broader audience, and enabling collaborative analysis and innovation across the greater geothermal industry.

access↗
DOE OSTI2021

Modeling of the Molten Salt Reactor Experiment with SCALE

A SCALE model was developed for the Molten Salt Reactor Experiment (MSRE) benchmark that was recently added to the International Handbook of Evaluated Reactor Physics Benchmark Experiments. This SCALE model served as a basis for criticality calculations and nuclear data sensitivity and uncertainty analyses with the Monte Carlo code Shift and the TSUNAMI computational capabilities in the SCALE code system. The focus of this work is the assessment of the impact of nuclear data on the calculated eigenvalue results in support of the discussion of differences between the calculated and the experimental eigenvalue result. The differences in the eigenvalues obtained using the ENDF/B-VII.0, ENDF/B-VII.1, and ENDF/B-VIII.0 nuclear data libraries cover a relatively small range of ~230 pcm. Since eigenvalue sensitivity of the MSRE is dominated by the neutron multiplicity and neutron capture of 235 U and elastic scattering in graphite, relevant changes in the ENDF/B libraries for nuclear reactions (such as carbon capture) that caused large differences in other graphite-moderated systems did not have a significant impact. Propagation of nuclear data uncertainty results in an eigenvalue uncertainty of ~700 pcm with the major contributors being 235 U neutron multiplicity, graphite elastic scattering, and 7Li neutron capture. All calculations resulted in large differences of ~2000 pcm in eigenvalue compared to the benchmark experimental value. Several potential contributors to this difference—including uncertainties and gaps in the knowledge of the material, geometry, and nuclear data—were identified. Simplified models of the full MSRE core were developed, and similarity assessments were conduced with the full MSRE core model. It was found that simplified models can serve as adequate surrogates of the full-core model such that they can be used for performing selected nuclear data performance assessments with a lower computational burden.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2020

Multi-GPU immersed boundary method hemodynamics simulations

Large-scale simulations of blood flow that resolve the 3D deformation of each comprising cell are increasingly popular owing to algorithmic developments in conjunction with advances in compute capability. Among different approaches for modeling cell-resolved hemodynamics, fluid structure interaction (FSI) algorithms based on the immersed boundary method are frequently employed for coupling separate solvers for the background fluid and the cells within one framework. GPUs can accelerate these simulations; however, both current pre-exascale and future exascale CPU-GPU heterogeneous systems face communication challenges critical to performance and scalability. In this paper, we describe, to our knowledge, the largest distributed GPU-accelerated FSI simulations of high hematocrit cell-resolved flows with over 17 million red blood cells. We compare scaling on a fat node system with six GPUs per node and on a system with a single GPU per node. Through comparison between the CPU- and GPU-based implementations, we identify the costs of data movement in multiscale multi-grid FSI simulations on heterogeneous systems and show it to be the greatest performance bottleneck on the GPU.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Characterization and identification of HPC applications at leadership computing facility

High Performance Computing (HPC) is an important method for scientific discovery via large-scale simulation, data analysis, or artificial intelligence. Leadership-class supercomputers are expensive, but essential to run large HPC applications. The Petascale era of supercomputers began in 2008, with the first machines achieving performance in excess of one petaflops, and with the advent of new supercomputers in 2021 (e.g., Aurora, Frontier), the Exascale era will soon begin. However, the high theoretical computing capability (i.e., peak FLOPS) of a machine is not the only meaningful target when designing a supercomputer, as the resources demand of applications varies. A deep understanding of the characterization of applications that run on a leadership supercomputer is one of the most important ways for planning its design, development and operation. In order to improve our understanding of HPC applications, user demands and resource usage characteristics, we perform correlative analysis of various logs for different subsystems of a leadership supercomputer. This analysis reveals surprising, sometimes counter-intuitive patterns, which, in some cases, conflicts with existing assumptions, and have important implications for future system designs as well as supercomputer operations. For example, our analysis shows that while the applications spend significant time on MPI, most applications spend very little time on file I/O. Combined analysis of hardware event logs and task failure logs show that the probability of a hardware FATAL event causing task failure is low. Combined analysis of control system logs and file I/O logs reveals that pure POSIX I/O is used more widely than higher level parallel I/O. Based on holistic insights of the application gained through combined and co-analysis of multiple logs from different perspectives and general intuition, we engineer features to "fingerprint" HPC applications. We use t-SNE (a machine learning technique for dimensionality reduction) to validate the explainability of our features and finally train machine learning models to identify HPC applications or group those with similar characteristic. To the best of our knowledge, this is the first work that combines logs on file I/O, computing, and inter-node communication for insightful analysis of HPC applications in production.

Liu, Zhengchun↗