Improving resilience of scientific software through a domain-specific approach
Not Available
Reguly, I. Z. (ORCID:0000000243854204)↗
SEARCH · Engineering Papers
Results for “Computer performance”
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.
The Kokkos OpenMPTarget Backend: Implementation and Lessons Learned
As the supercomputing landscape diversifies, solutions such as Kokkos to write vendor agnostic applications and libraries have risen in popularity. Kokkos provides a programming model designed for performance portability, which allows developers to write a single source implementation that can run efficiently on various architectures. At its heart, Kokkos maps parallel algorithms to architecture and vendor specific backends written in lower level programming models such as CUDA and HIP. Another approach to writing vendor agnostic parallel code is using OpenMP’s directives based approach, which lets developers annotate code to express parallelism. It is implemented at the compiler level and is supported by all major high performance computing vendors, as well as the primary Open Source toolchains GNU and LLVM. Since its inception, Kokkos has used OpenMP to parallelize on CPU architectures. In this paper, we explore leveraging OpenMP for a GPU backend and discuss the challenges we encountered when mapping the Kokkos APIs and semantics to OpenMP target constructs. As an exemplar workload we chose a simple conjugate gradient solver for sparse matrices. We find that performance on NVIDIA and AMD GPUs varies widely based on details of the implementation strategy and the chosen compiler. Furthermore, the performance of the OpenMP implementations decreases with increasing complexity of the investigated algorithms.
Gayatri, Rahulkumar↗
Exploring the capabilities of support vector machines in detecting silent data corruptions
As the exascale era approaches, the increasing capacity of high-performance computing (HPC) systems with targeted power and energy budget goals introduces significant challenges in reliability. Silent data corruptions (SDCs), or silent errors, are one of the major sources that corrupt the execution results of HPC applications without being detected. Here in this paper, we explore a set of novel SDC detectors – by leveraging epsilon-insensitive support vector machine regression – to detect SDCs that occur in HPC applications. The key contributions are threefold. (1) Our exploration takes temporal, spatial, and spatiotemporal features into account and analyzes different detectors based on different features. (2) We provide an in-depth study on the detection ability and performance with different parameters, and we optimize the detection range carefully. (3) Experiments with eight real-world HPC applications show that support-vector-machine-based detectors can achieve detection sensitivity (i.e., recall) up to 99% yet suffer a less than 1% false positive rate for most cases. Our detectors incur low performance overhead, 5% on average, for all benchmarks studied in this work.
97 MATHEMATICS AND COMPUTING↗
Multi-modal Energy-optimal Trip Scheduling in Real-time (METS-R) for Transportation Hubs (Final Report)
This report summarizes the work performed under the award number EE0008524. The project develops the Multi-modal Energy-optimal Trip Scheduling in Real-time (METS-R) platform as the next-generation transportation solution based on autonomous electric vehicles (AEV) serving passenger trips from and to urban transportation hubs, to substantially reduce transportation energy consumption. Extensive data collection and analyses were first conducted to understand the demand patterns and energy consumption of hub-based on-road trips. Then, a data-driven framework that consists of an analytical module and a simulation module was proposed. For the analytical module, five planning + operation tools were developed to support the planning and energy-efficient operations of urban AEV services: the charging station planning that robotically allocates charging supplies based on the stationary charging demand distribution; the transit planning and demand adaptive scheduling model that efficiently generates\ candidate transit routes from hubs to other places and dynamically adjusts the transit time table to fit the current demand; the online energy-efficient routing that learns the energy-optimal paths from observations of link-level energy consumption in real-time; the hub-based ridesharing that matches trip requests together with account for the uncertainty of future trip demand and vehicle supply; and finally, the integrated demand prediction and anomaly detection pipeline that leverages the flight/train time table and support other planning/operation tools. To demonstrate the performance of these tools, a scalable high-performance agent-based simulator was built. We divided the urban space into multiple service zones where each zone was considered as an agent for passenger generation and vehicle charging. Two types of AEV agents were coded to model two types of mobility services: AEV taxi and AEV transit. For the AEV taxi, the team implemented the functions of pickup/drop-off passengers, energy-efficient routing, ridesharing, fleet rebalancing, and recharging. For the AEV bus, the team implemented the functions of demand-adaptive route scheduling, passenger boarding, and recharging. A high-performance computing framework was introduced to receive various profiling information (such as link energy updates, vehicle speed) from the simulator instances and communicate the operational commands back to the instances. The numerical experiments show that each of the proposed operational algorithms can reduce energy consumption and improve system efficiency. Furthermore, there exists the need to collectively consider multiple planning + operational strategies as multiple strategies can influence each other in terms of performance impacts. Recommendations for future work related to AEV planning and simulation are discussed.
29 ENERGY PLANNING, POLICY, AND ECONOMY↗
Using Discrete Event Simulation for Programming Model Exploration at Extreme-Scale: Macroscale Components for the Structural Simulation Toolkit (SST).
Discrete event simulation provides a powerful mechanism for designing and testing new extreme- scale programming models for high-performance computing. Rather than debug, run, and wait for results on an actual system, design can first iterate through a simulator. This is particularly useful when test beds cannot be used, i.e. to explore hardware or scales that do not yet exist or are inaccessible. Here we detail the macroscale components of the structural simulation toolkit (SST). Instead of depending on trace replay or state machines, the simulator is architected to execute real code on real software stacks. Our particular user-space threading framework allows massive scales to be simulated even on small clusters. The link between the discrete event core and the threading framework allows interesting performance metrics like call graphs to be collected from a simulated run. Performance analysis via simulation can thus become an important phase in extreme-scale programming model and runtime system design via the SST macroscale components.
97 MATHEMATICS AND COMPUTING↗
Exploiting Redundancy and Application Scalability for Cost-Effective, Time-Constrained Execution of HPC Applications on Amazon EC2
The use of clouds to execute high-performance computing (HPC) applications has greatly increased recently. Clouds provide several potential advantages over traditional supercomputers and in-house clusters. The most popular cloud is currently Amazon EC2, which provides fixed-cost and variable-cost, auction-based options. The auction market trades lower cost for potential interruptions that necessitate checkpointing; if the market price exceeds the bid price, a node is taken away from the user without warning. We explore techniques to maximize performance per dollar given a time constraint within which an application must complete. Specifically, we design and implement multiple techniques to reduce expected cost by exploiting redundancy in the EC2 auction market. We then design an adaptive algorithm that selects a scheduling algorithm and determines the bid price. We show that our adaptive algorithm executes programs up to seven times cheaper than using the on-demand market and up to 44 percent cheaper than the best non-redundant, auction-market algorithm. We extend our adaptive algorithm to incorporate application scalability characteristics for further cost savings. In conclusion, we show that the adaptive algorithm informed with scalability characteristics of applications achieves up to 56 percent cost savings compared to the expected cost for the base adaptive algorithm run at a fixed, user-defined scale.
97 MATHEMATICS AND COMPUTING↗
Summer Internship Report: ARA2 Benchmarking
Over the past decade, the RISC-V Instruction Set Architecture (ISA) has emerged as a significant player in both academic and industrial processor design due to its open-source nature, modular extension system, and versatility across domains ranging from microcontrollers to high-performance computing (HPC). One of its most important recent advancements is the RISC-V Vector Extension (RVV), which enables explicit data-level parallelism through vector registers and vectorized instructions. Unlike traditional SIMD (Single Instruction, Multiple Data) architectures that fix vector lengths at design time, RVV uses the concept of VLEN (vector register length) as a hardware-independent parameter and allows software to adapt dynamically to the available vector width. This flexible approach ensures portability across implementations while enabling scalable performance. The ARA2 core is a parameterizable RISC-V vector processor developed at the Integrated Systems Lab at ETH Zürich and the University of Bologna. Designed as a tightly-coupled accelerator to a scalar RISC-V core, ARA2 implements the RVV 1.0 specification and offers tunable architectural parameters such as the number of vector lanes, VLEN, and cache sizes.
97 MATHEMATICS AND COMPUTING↗
Algorithm sensitivity analysis and parameter tuning for tissue image segmentation pipelines
Motivation: Sensitivity analysis and parameter tuning are important processes in large-scale image analysis. They are very costly because the image analysis workflows are required to be executed several times to systematically correlate output variations with parameter changes or to tune parameters. An integrated solution with minimum user interaction that uses effective methodologies and high performance computing is required to scale these studies to large imaging datasets and expensive analysis workflows. Results: The experiments with two segmentation workflows show that the proposed approach can (i) quickly identify and prune parameters that are non-influential; (ii) search a small fraction (about 100 points) of the parameter search space with billions to trillions of points and improve the quality of segmentation results (Dice and Jaccard metrics) by as much as 1.42 x compared to the results from the default parameters; (iii) attain good scalability on a high performance cluster with several effective optimizations. Conclusions: Our work demonstrates the feasibility of performing sensitivity analyses, parameter studies and auto-tuning with large datasets. The proposed framework can enable the quantification of error estimations and output variations in image segmentation pipelines.
Biochemistry & Molecular Biology↗
Solving the electronic structure problem for over 100000 atoms in real space
Using a real-space high-order finite-difference approach, we investigate the electronic structure of large spherical silicon nanoclusters. Within Kohn-Sham density functional theory and using pseudopotentials, we report the self-consistent field convergence of a system with over 100000 atoms: a Si 107,641 H 9,084 nanocluster with a diameter of 16 nm. Our approach uses Chebyshev-filtered subspace iteration to speed up the convergence of the eigenspace, and blockwise Hilbert space-filling curves to speed up sparse matrix-vector multiplications, all of which are implemented in the parsec code. For the largest system, we utilized 2048 nodes (114 688 cores) on the Frontera machine in the Texas Advanced Computing Center. Our quantitative analysis of the electronic structure shows how it gradually approaches its bulk counterpart as a function of nanocluster size. The band gap is enlarged due to quantum confinement in nanoclusters, but decreases as the system size increases, as expected. In conclusion, our work serves as a proof of concept for the capacity of the real-space approach in efficiently parallelizing very large calculations using high-performance computer platforms, which can straightforwardly be replicated in other systems with more than 10 5 atoms.
0-dimensional systems↗
Cross Domain Deterrence: Livermore Technical Report, 2014-2016
Lawrence Livermore National Laboratory (LLNL) is an original collaborator on the project titled “Deterring Complex Threats: The Effects of Asymmetry, Interdependence, and Multi-polarity on International Strategy,” (CDD Project) led by the UC Institute on Global Conflict and Cooperation at UCSD under PIs Jon Lindsay and Erik Gartzke , and funded through the DoD Minerva Research Initiative. In addition to participating in workshops and facilitating interaction among UC social scientists, LLNL is leading the computational modeling effort and assisting with empirical case studies to probe the viability of analytic, modeling and data analysis concepts. This report summarizes LLNL work on the CDD Project to date, primarily in Project Years 1-2, corresponding to Federal fiscal year 2015. LLNL brings two unique domains of expertise to bear on this Project: (1) access to scientific expertise on the technical dimensions of emerging threat technology, and (2) high performance computing (HPC) expertise, required for analyzing the complexity of bargaining interactions in the envisioned threat models. In addition, we have a small group of researchers trained as social scientists who are intimately familiar with the International Relations research. We find that pairing simulation scientists, who are typically trained in computer science, with domain experts, social scientists in this case, is the most effective route to developing powerful new simulation tools capable of representing domain concepts accurately and answering challenging questions in the field.
97 MATHEMATICS AND COMPUTING↗
Grid-based minimization at scale: Feldman-Cousins corrections for SBN
We present a computational model for the construction of Feldman-Cousins (FC) corrections frequently used in High Energy Physics (HEP) analysis. The program contains a grid-based minimization and is written in C++. Our algorithms exploit vectorization through Eigen3, yielding a single-core speed-up of 350 compared to the original implementation, and achieve MPI data parallelism by using DIY. We demonstrate the application to scale very well at High Performance Computing (HPC) sites. We use HDF5 in conjunction with HighFive to write results of the calculation to file.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
chatHPC: Empowering HPC users with large language models
The ever-growing number of pre-trained large language models (LLMs) across scientific domains presents a challenge for application developers. While these models offer vast potential, fine-tuning them with custom data, aligning them for specific tasks, and evaluating their performance remain crucial steps for effective utilization. However, applying these techniques to models with tens of billions of parameters can take days or even weeks on modern workstations, making the cumulative cost of model comparison and evaluation a significant barrier to LLM-based application development. To address this challenge, we introduce an end-to-end pipeline specifically designed for building conversational and programmable AI agents on high performance computing (HPC) platforms. Our comprehensive pipeline encompasses: model pre-training, fine-tuning, web and API service deployment, along with crucial evaluations for lexical coherence, semantic accuracy, hallucination detection, and privacy considerations. Here, we demonstrate our pipeline through the development of chatHPC, a chatbot for HPC question answering and script generation. Leveraging our scalable pipeline, we achieve end-to-end LLM alignment in under an hour on the Frontier supercomputer. We propose a novel self-improved, self-instruction method for instruction set generation, investigate scaling and fine-tuning strategies, and conduct a systematic evaluation of model performance. The established practices within chatHPC will serve as a valuable guidance for future LLM-based application development on HPC platforms.
97 MATHEMATICS AND COMPUTING↗
Using OpenMP for HEP Framework Algorithm Scheduling [Slides]
The OpenMP standard is the primary mechanism used at high performance computing facilities to allow intra-process parallelization. In contrast, many HEP specific software (such as CMSSW, GaudiHive, and ROOT) make use of Intel's Threading Building Blocks (TBB) library to accomplish the same goal. In this talk we will discuss our work to compare TBB and OpenMP when used for scheduling algorithms to be run by a HEP style data processing framework (i.e. running hundreds of interdependent algorithms at most once for each event read from the detector). This includes both scheduling of different algorithms to be run concurrently as well as scheduling concurrent work within one algorithm. As part of the discussion we present an overview of the OpenMP threading model. We also explain how we used OpenMP when creating a simplified HEP-like processing framework. Using that simplified framework, and a similar one written using TBB, we will present performance comparisons between TBB and different compiler versions of OpenMP.
73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
Computationally efficient optimization models for preliminary distillation column design and separation energy targeting
We propose versatile shortcut distillation column and separation energy targeting models that are well-suited for superstructure-based process synthesis. The models are based on a novel reformulation of the Underwood equations to address systems where the components that are present in the feed can vary due to zero flow rates of some components. Also, we propose valid constraints, resulting in a significant enhancement of the computational performance of the models. Furthermore, the proposed distillation column model can automatically identify adequate key components and the energy requirement of a desired separation task, while considering a wide range of types of separations including non-sharp/sloppy splits. Also, the proposed separation energy targeting model can be used to estimate an energy requirement target for the separation of a mixture without finding detail network configurations. Due to their versatility and computational efficiency, the proposed models can be readily used for and expand the scope of superstructure-based process synthesis approaches.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
MaDaTS: Managing Data on Tiered Storage for Scientific Workflows
Scientific workflows are increasingly used in High Performance Computing (HPC) environments to manage complex simulation and analyses, often consuming and generating large amounts of data. However, workflow tools have limited support for managing the input, output and intermediate data. The data elements of a workflow are often managed by the user through scripts or other ad-hoc mechanisms. Technology advances for future HPC systems is redefining the memory and storage subsystem by introducing additional tiers to improve the I/O performance of data-intensive applications. These architectural changes introduce additional complexities to managing data for scientific workflows. Thus, we need to manage the scientific workflow data across the tiered storage system on HPC machines. In this paper, we present the design and implementation of MaDaTS (Managing Data on Tiered Storage for Scientific Workflows), a software architecture that manages data for scientific workflows. We introduce Virtual Data Space (VDS), an abstraction of the data in a workflow that hides the complexities of the underlying storage system while allowing users to control data management strategies. We evaluate the data management strategies with real scientific and synthetic workflows, and demonstrate the capabilities of MaDaTS. Our experiments demonstrate the flexibility, performance and scalability gains of MaDaTS as compared to the traditional approach of managing data in scientific workflows.
Ghoshal, Devarshi↗
Benchmarking Monte Carlo codes for the modelling of low-energy neutron production target reactions
The increasing adoption of accelerator-based neutron sources (ABNS) for applications including neutron capture therapy (NCT) research has highlighted the need for accurate simulation tools. Precise modelling of the neutron production target is crucial to ensure that simulated predictions of neutron beam characteristics used for subsequent beam shaping assembly design are reliable. This work presents a comprehensive benchmarking of four widely-used Monte Carlo codes - Geant4, PHITS, FLUKA (CERN), and MCNP - for modelling low-energy neutron production target reactions. Using their recommended physics models and cross-section libraries, we evaluate each code’s performance in simulating four beam-target reactions: 7 Li(p,n) 7 Be, 9 Be(p,n) 9 B, 9 Be(d,n) 10 B, and C(d,n)N. Predictions of neutron yield, angular distributions, and energy spectra are compared against available thick target experimental data. Results show varying levels of agreement between the codes depending on the reaction type, energy range, and beam characteristics. Geant4, MCNP and PHITS are the overall best performing codes for the simulation of total neutron yield and yield in the forward direction across most reactions. Across energies where experimental benchmarks exist, inter-code discrepancies in total and forward-directed yield are typically 10 to 30%, with larger deviations at near-threshold incident ion energies. PHITS provides the best overall reproduction of experimental spectra, particularly for the 9 Be(p,n) 9 B reaction. Additionally, PHITS demonstrates superior computational performance for most reactions. These findings provide valuable guidance for ABNS design, highlighting the strengths and limitations of each code for the simulation of low-energy neutron production reactions.
43 PARTICLE ACCELERATORS↗
CYRANO3 the EDF fuel code performance: Global overview and recent developments on MOX fuel
CYRANO3 is the thermal mechanical industrial code developed and used by EDF to simulate the nuclear fuel rod performance in pile, and during the disposal storage period as well. This code has already been used successfully for the last ten years by the EDF engineering teams in justifying the overall reloading options covering various fuels: UO{sub 2}, UO{sub 2} + gadolinium, MOX, UO{sub 2} plus additives, and various cladding as well proposed by the nuclear fuel suppliers. Its particularity is related to a construction around a standard one-dimensional thermal and mechanical finite elements kernel with an appropriate architecture to plug in any new functionalities or models. Since the beginning it is developed as a module able integrating within a core calculation chain. Other constrain is to always guaranty a running time consistent with the calculation of a core within a reasonable time. The objective is to provide the EDF designers with a convivial and qualified code, with robust algorithms and able to quickly evolve with computer performances. Algorithms are able to account for adiabatic transients and are robust enough to avoid any undesirable overflows. The permanent challenge is indeed to provide with the more descriptive simulations not degrading to much the running time. Developments are obviously performed within the strict Quality Insurance rules. The code is then associated to a large documentation base, including model specification, realization, validation and test procedures. Maintenance is managed through a system of anomaly and improvement forms. Computers are still evolving quickly. Therefore, after an architecture adaptation, the CYRANO3 code is now integrated in the PLEIADES platform as a PWR industrial application. The platform is an exchange space allowing sharing functionalities, models and properties. In this paper, we will give a global overview of the capabilities of CYRANO3 code to simulate the thermomechanical behavior of fuel rods, and we will focus on recent developments realized on MOX fuel. In CYRANO3 code, heterogeneous models (neutronic model, micromechanical behavior law and fission gas model) are developed for MOX fuels and integrated in order to account for differential evolutions between phases under irradiation. The methodology and these models are presented in this paper. (authors)
11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
Resilience Design Patterns: A Structured Approach to Resilience at Extreme Scale
Reliability is a serious concern for future extreme-scale high-performance computing (HPC) systems. Projections based on the current generation of HPC systems and technology roadmaps suggest the prevalence of very high fault rates in future systems. While the HPC community has developed various resilience solutions, application-level techniques as well as system-based solutions, the solution space remains fragmented. There are no formal methods and metrics to integrate the various HPC resilience techniques into composite solutions, nor are there methods to holistically evaluate the adequacy and efficacy of such solutions in terms of their protection coverage, and their performance \& power efficiency characteristics. Additionally, few of the current approaches are portable to newer architectures and software environments that will be deployed on future systems. In this paper, we develop a structured approach to the design, evaluation and optimization of HPC resilience using the concept of design patterns. A design pattern is a general repeatable solution to a commonly occurring problem. We identify the problems caused by various types of faults, errors and failures in HPC systems and the techniques used to deal with these events. Each well-known solution that addresses a specific HPC resilience challenge is described in the form of a pattern. We develop a complete catalog of such resilience design patterns, which may be used by system architects, system software and tools developers, application programmers, as well as users and operators as essential building blocks when designing and deploying resilience solutions. We also develop a design framework that enhances a designer's understanding the opportunities for integrating multiple patterns across layers of the system stack and the important constraints during implementation of the individual patterns. It is also useful for defining mechanisms and interfaces to coordinate flexible fault management across hardware and software components. The resilience patterns and the design framework also enable exploration and evaluation of design alternatives and support optimization of the cost-benefit trade-offs among performance, protection coverage, and power consumption of resilience solutions. Here, the overall goal of this work is to establish a systematic methodology for the design and evaluation of resilience technologies in extreme-scale HPC systems that keep scientific applications running to a correct solution in a timely and cost-efficient manner despite frequent faults, errors, and failures of various types.
97 MATHEMATICS AND COMPUTING↗