Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “exascale computing”

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 775 records · Page 43
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Optimizing High Performance Markov Clustering for Pre-Exascale Architectures

HipMCL is a high-performance distributed memory implementation of the popular Markov Cluster Algorithm (MCL) and can cluster large-scale networks within hours using a few thousand CPU-equipped nodes. It relies on sparse matrix computations and heavily makes use of the sparse matrix-sparse matrix multiplication kernel (SpGEMM). The existing parallel algorithms in HipMCL are not scalable to Exascale architectures, both due to their communication costs dominating the runtime at large concurrencies and also due to their inability to take advantage of accelerators that are increasingly popular. In this work, we systematically remove scalability and performance bottlenecks of HipMCL. We enable GPUs by performing the expensive expansion phase of the MCL algorithm on GPU. Additionally, we propose a CPU-GPU joint distributed SpGEMM algorithm called pipelined Sparse SUMMA and integrate a probabilistic memory requirement estimator that is fast and accurate. Furthermore, we develop a new merging algorithm for the incremental processing of partial results produced by the GPUs, which improves the overlap efficiency and the peak memory usage. We also integrate a recent and faster algorithm for performing SpGEMM on CPUs. We validate our new algorithms and optimizations with extensive evaluations. With the enabling of the GPUs and integration of new algorithms, HipMCL is up to 12.4x faster, being able to cluster a network with 70 million proteins and 68 billion connections just under 15 minutes using 1024 nodes of ORNL's Summit supercomputer.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Development of a multi-layer canopy model for E3SM Land Model with support for heterogeneous computing

The vertical structure of vegetation canopies creates micro-climates. However, the land components of most Earth System Models, including the Energy Exascale Earth System Model (E3SM), typically neglect vertical canopy structure by using a single layer big-leaf representation to simulate water, CO 2 , and energy exchanges between the land and the atmosphere. In this study, we developed a Multi-Layer Canopy Model for the E3SM Land Model to resolve the micro-climate created by vegetation canopies. The model developed in this study re-implements the CLM-ml_v1 to support heterogeneous computing architectures consisting of CPUs and GPUs and includes three additional optimization-based stomatal conductance models. The use of Portable, Extensible Toolkit for Scientific Computation provides a speedup of 25–50 times on a GPU relative to a CPU. The numerical implementation of the model was verified against CLM-ml_v1 for a month-long simulation using data from the Ameriflux US-University of Michigan Biological Station site. Model structural uncertainty was explored by performing control simulations for five stomatal conductance models that exclude and include the control of plant hydrodynamics (PHD) on photosynthesis. The bias in simulated sensible and latent heat fluxes was lower when PHD was accounted for in the model. Additionally, six idealized simulations were performed to study the impact of three environmental variables (i.e. air temperature, atmospheric CO 2 , and soil moisture) on canopy processes (i.e. net CO 2 assimilation, leaf temperature, and leaf water potential). Increasing air temperature reduced net CO 2 assimilation and increased air temperature. Net CO 2 assimilation increased at higher atmospheric CO 2 , while decreasing soil moisture resulted in lower leaf water potential.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2018

Blackcomb 2: Hardware-Software Co-design for Non-Volatile Memory in Exascale Systems

Fast non-volatile memory (NVM) offers game-changing opportunities to rethink traditional system architectures; to address their energy inefficiencies; to leverage the new devices for greater performance; and to exploit the new capability of non-volatility to enhance system resilience, even in the face of larger system scale and degraded reliability of components. However, the devil is in the details, and to reap such potential benefits we must address challenges that arise when the contents of memory survive application failure, when software becomes responsible for managing the contents of CPU caches, and when memory accelerators blur the line between hardware and software. Building on the accomplishments of the Blackcomb Project, funded in 2010, we have identified, evaluated, and optimized promising NVM hardware and software technologies that are essential to provide the necessary memory capacity, performance, resilience, and energy efficiency in exascale systems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Porting the WAVEWATCH III (v6.07) wave action source terms to GPU

Abstract. Surface gravity waves play a critical role in several processes, including mixing, coastal inundation, and surface fluxes. Despite the growing literature on the importance of ocean surface waves, wind–wave processes have traditionally been excluded from Earth system models (ESMs) due to the high computational costs of running spectral wave models. The development of the Next Generation Ocean Model for the DOE’s (Department of Energy) E3SM (Energy Exascale Earth System Model) Project partly focuses on the inclusion of a wave model, WAVEWATCH III (WW3), into E3SM. WW3, which was originally developed for operational wave forecasting, needs to be computationally less expensive before it can be integrated into ESMs. To accomplish this, we take advantage of heterogeneous architectures at DOE leadership computing facilities and the increasing computing power of general-purpose graphics processing units (GPUs). This paper identifies the wave action source terms, W3SRCEMD, as the most computationally intensive module in WW3 and then accelerates them via GPU. Our experiments on two computing platforms, Kodiak (P100 GPU and Intel(R) Xeon(R) central processing unit, CPU, E5-2695 v4) and Summit (V100 GPU and IBM POWER9 CPU) show respective average speedups of 2× and 4× when mapping one Message Passing Interface (MPI) per GPU. An average speedup of 1.4× was achieved using all 42 CPU cores and 6 GPUs on a Summit node (with 7 MPI ranks per GPU). However, the GPU speedup over the 42 CPU cores remains relatively unchanged (∼ 1.3×) even when using 4 MPI ranks per GPU (24 ranks in total) and 3 MPI ranks per GPU (18 ranks in total). This corresponds to a 35 %–40 % decrease in both simulation time and usage of resources. Due to too many local scalars and arrays in the W3SRCEMD subroutine and the huge WW3 memory requirement, GPU performance is currently limited by the data transfer bandwidth between the CPU and the GPU. Ideally, OpenACC routine directives could be used to further improve performance. However, W3SRCEMD would require significant code refactoring to make this possible. We also discuss how the trade-off between the occupancy, register, and latency affects the GPU performance of WW3.

58 GEOSCIENCES↗
DOE OSTI2021

ExaSGD: 2021 Kernel Thrust Activities

The Kernel Thrust milestone ADSE22-214 covers the development of device-capable optimization algorithms and solvers technologies required by the ExaSGD project’s software stack in order to solve security-constrained alternating current optimal power flow (SC-ACOPF) problems on emerging exascale architectures. To this extent, in FY21 the main objective of the Kernel Thrust was (i) provide robust optimization solver(s) that run efficiently on hardware accelerator devices (i.e., NVIDIA and AMD GPUs) to perform intra-node computations and (ii) provide coarse-grain parallel optimization capabilities that exploit the decomposition opportunities present in the SC-ACOPF challenge problems to provide exascale-capable solvers.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Improving climate model coupling through a complete mesh representation: a case study with E3SM (v1) and MOAB (v5.x)

One of the fundamental factors contributing to the spatiotemporal inaccuracy in climate modeling is the mapping of solution field data between different discretizations and numerical grids used in the coupled component models. The typical climate computational workflow involves evaluation and serialization of the remapping weights during the preprocessing step, which is then consumed by the coupled driver infrastructure during simulation to compute field projections. Tools like Earth System Modeling Framework (ESMF) and TempestRemap offer capability to generate conservative remapping weights, while the Model Coupling Toolkit (MCT) that is utilized in many production climate models exposes functionality to make use of the operators to solve the coupled problem. However, such multistep processes present several hurdles in terms of the scientific workflow and impede research productivity. In order to overcome these limitations, we present a fully integrated infrastructure based on the Mesh Oriented datABase (MOAB) library, which allows for a complete description of the numerical grids and solution data used in each submodel. Through a scalable advancing-front intersection algorithm, the supermesh of the source and target grids are computed, which is then used to assemble the high-order, conservative, and monotonicity-preserving remapping weights between discretization specifications. The Fortran-compatible interfaces in MOAB are utilized to directly link the submodels in the Energy Exascale Earth System Model (E3SM) to enable online remapping strategies in order to simplify the coupled workflow process. We demonstrate the superior computational efficiency of the remapping algorithms in comparison with other state-of-the-science tools and present strong scaling results on large-scale machines for computing remapping weights between the spectral element atmosphere and finite volume discretizations on the polygonal ocean grids.

58 GEOSCIENCES↗
DOE OSTI2020

Hypergraph Analytics of Domain Name System Relationships

We report on the use of novel mathematical methods in hypergraph analytice over a large quantity of DNS data. Hypergraphs generalize graphs, as used in network science, to better model complex multiway relations in cyber data. Specifically, casting DNS data from Georgia Tech's ActiveDNS repository as hypergraphs allows us to fully represent the interactions between {\em collections} of domains and IP addresses. To facilitate large-scale analytics, we fielded an analytical pipeline of two capabilities. HyperNetX (HNX) is a Python package for the exploration and visualization of hypergraphs, acting as a frontend. For the backend, the Chapel HyperGraph Library (CHGL) is a library for high performance hypergraph analytics written in the exascale programming language Chapel. CHGL was used to process gigascale DNS data, performing compute-intensive calculations for data reduction and segmentation. Identified portions are then sent to HNX for both exploratory analysis and knowledge discovery targeting known tactics, techniques, and procedures.

Joslyn, Cliff A.↗
DOE OSTI2022

Low-synch Gram–Schmidt with delayed reorthogonalization for Krylov solvers

The parallel strong-scaling of iterative methods is often determined by the number of global reductions at each iteration. Low-synch Gram-Schmidt algorithms are applied here to the Arnoldi algorithm to reduce the number of global reductions and therefore to improve the parallel strong-scaling of iterative solvers for nonsymmetric matrices such as the GMRES and the Krylov-Schur iterative methods. In the Arnoldi context, the factorization is "left-looking" and processes one column at a time. Among the methods for generating an orthogonal basis for the Arnoldi algorithm, the classical Gram-Schmidt algorithm, with reorthogonalization (CGS2) requires three global reductions per iteration. A new variant of CGS2 that requires only one reduction per iteration is presented and applied to the Arnoldi algorithm. Delayed CGS2 (DCGS2) employs the minimum number of global reductions per iteration (one) for a one-column at-a-time algorithm. The main idea behind the new algorithm is to group global reductions by rearranging the order of operations. DCGS2 must be carefully integrated into an Arnoldi expansion or a GMRES solver. Numerical stability experiments assess robustness for Krylov-Schur eigenvalue computations. Performance experiments on the ORNL Summit supercomputer then establish the superiority of DCGS2 over CGS2.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

PLEXUS: A Pattern-Oriented Runtime System Architecture for Resilient Extreme-Scale High-Performance Computing Systems

For high-performance computing (HPC) system designers and users, meeting the myriad challenges of next-generation exascale supercomputing systems requires rethinking their approach to application and system software design. Among these challenges, providing resiliency and stability to the scientific applications in the presence of high fault rates requires new approaches to software architecture and design. As HPC systems become increasingly complex, they require intricate solutions for detection and mitigation for various modes of faults and errors that occur in these large-scale systems, as well as solutions for failure recovery. These resiliency solutions often interact with and affect other system properties, including application scalability, power and energy efficiency. Therefore, resilience solutions for HPC systems must be thoughtfully engineered and deployed.In previous work, we developed the concept of resilience design patterns, which consist of templated solutions based on well-established techniques for detection, mitigation and recovery. In this paper, we use these patterns as the foundation to propose new approaches to designing runtime systems for HPC systems. The instantiation of these patterns within a runtime system enables flexible and adaptable end-to-end resiliency solutions for HPC environments. The paper describes the architecture of the runtime system, named Plexus, and the strategies for dynamically composing and adapting pattern instances under runtime control. This runtime-based approach enables actively balancing the cost-benefit trade-off between performance overhead and protection coverage of the resilience solutions. Based on a prototype implementation of PLEXUS, we demonstrate the resiliency and performance gains achieved by the pattern-based runtime system for a parallel linear solver application.

Hukerikar, Saurabh↗
NASA NTRS2019

Testing Scientific Software: Challenges and Remedies

Over the past two decades, the emergence of highly effective software testing frameworks has greatly simplified the development and use of unit tests and has led to new software development paradigms such as test driven development (TDD). However, scientific computing introduces a number of unique testing challenges, including numerical algorithms, distributed parallelism, and exascale environments. This presentation will begin with a brief introduction to unit testing, testing frameworks, and some simple examples using pFUnit, a unit testing framework for Fortran + MPI. I will then take a closer look at several of the obstacles one faces when testing technical software and suggest methodologies that can mitigate these difficulties.

Clune, Tom↗
DOE OSTI2017

Application Modernization at LLNL and the Sierra Center of Excellence

We repport that in 2014, Lawrence Livermore National Laboratory began acquisition of Sierra, a pre-exascale system from IBM and Nvidia. It marks a significant shift in direction for LLNL by introducing the concept of heterogeneous computing via GPUs. LLNL’s mission requires application teams to prepare for this paradigm shift. Thus, the Sierra procurement required a proposed Center of Excellence that would align the expertise of the chosen vendors with laboratory personnel that represent the application developers, system software, and tool providers in a concentrated effort to prepare the laboratory’s codes in advance of the system transitioning to production in 2018. Finally, this article presents LLNL’s overall application strategy, with a focus on how LLNL is collaborating with IBM and Nvidia to ensure a successful transition of its mission-oriented applications into the exascale era.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Machine Learning and Understanding for Intelligent Extreme Scale Scientific Computing and Discovery. DOE Workshop Report, January 7–9, 2015, Rockville, MD

Large-scale parallel simulations and data analysis drive scientific discovery across many disciplines. To drive larger and more detailed simulations, and deal with larger data volumes, exascale machines capable of 1018 operations per second are expected within the next five to ten years. However, the complexity of developing and adapting modern simulation codes for these architectures is increasing rapidly. Workloads on exascale machines will be billion-way parallel. Exploiting the full capability of the machine will require carefully assigning application tasks to cores, accelerators, deep memories, and other heterogeneous compute resources, while simultaneously optimizing for time to solution, data movement, power, and resilience. Optimizations that improve performance on one machine may slow down another. Worse, applications themselves are dauntingly complex. Production simulations comprise millions of lines of code and use sophisticated, adaptive algorithms whose performance is input-dependent. Complex workflows can couple multi-physics simulation with data preprocessing and post-processing modules.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

ESnet Requirements Review Program Through the IRI Lens: A Meta-Analysis of Workflow Patterns Across DOE Office of Science Programs (Final Report)

The Department of Energy (DOE) ensures America’s security and prosperity by addressing its energy, environmental, and nuclear challenges through transformative science and technology solutions. The DOE’s Office of Science (SC) delivers groundbreaking scientific discoveries and major scientific tools that transform our understanding of nature and advance the energy, economic, and national security of the United States. The SC’s programs advance DOE mission science across a wide range of disciplines and have developed the research infrastructure needed to remain at the forefront of scientific discovery. The DOE SC’s world-class research infrastructure — exemplified by the 28 SC scientific user facilities — provides the research community with premier observational, experimental, computational, and network capabilities. Each user facility is designed to provide unique capabilities to advance core DOE mission science for its sponsor SC program and to stimulate a rich discovery and innovation ecosystem. Research communities gather and flourish around each user facility, bringing together diverse perspectives. A hallmark of many facilities is the large population of students, postdoctoral researchers, and early-career scientists who contribute as full-fledged users. These facility staff and users collaborate over years to devise new approaches to utilizing the user facility’s core capabilities. The history of the SC user facilities has many examples of wildly inventive researchers challenging operational orthodoxy to pioneer new vistas of discovery; for example, the use of the synchrotron X-ray light sources for study of proteins and other large biological molecules. This continual reinvention of the practice of science — as users and staff forge novel approaches expressed in research workflows — unlocks new discoveries and propels scientific progress. Within this research ecosystem, the high-performance computing (HPC) and networking user facilities stewarded by SC’s Advanced Scientific Computing Research (ASCR) program play a dynamic cross-cutting role, enabling complex workflows demanding high performance data, networking, and computing solutions. The DOE SC’s three HPC user facilities and the Energy Sciences Network (ESnet) high-performance research network serve all of the SC’s programs as well as the global research community. Argonne Leadership Computing Facility (ALCF), the National Energy Research Scientific Computing Center (NERSC), and Oak Ridge Leadership Computing Facility (OLCF) conceive, build, and provide access to a range of supercomputing, advanced computing, and large-scale data-infrastructure platforms, while ESnet interconnects DOE SC research infrastructure and enables seamless exchange of scientific data. All four facilities operate testbeds to expand the frontiers of computing and networking research. Together, the ASCR facilities enterprise seeks to understand and meet the needs and requirements across SC and DOE domain science programs and priority efforts, highlighted by the formal requirements reviews (RRs) methodology. In recent years, the research communities around the SC user facilities have begun experimenting with and demanding solutions integrated with HPC and data infrastructure. This rise of integrated-science approaches is documented in many community and high-level government reports. At the dawn of the era of exascale science and the acceleration of artificial intelligence (AI) innovation, there is a broad need for integrated computational, data, and networking solutions. In response to these drivers, DOE has developed a vision for an Integrated Research Infrastructure (IRI): To empower researchers to meld DOE’s world-class research tools, infrastructure, and user facilities seamlessly and securely in novel ways to radically accelerate discovery and innovation.

42 ENGINEERING↗
DOE OSTI2023

Experiences Detecting Defective Hardware in Exascale Supercomputers

In May 2022, the newest supercomputer to top the TOP 500 list was Frontier at Oak Ridge National Laboratory, demonstrating the capability of computing more than 1.1 quintillion (1018) floating-point calculations every second. Driving this ground-breaking rate of computing is Frontier’s more than 37,000 graphics processing units (GPUs) and 9,408 central processing units (CPUs). In total, Frontier contains more than 60 million parts. At this scale, the smallest margin of error may generate hundreds of hardware errors across the system. These errors are capable of directly hindering world-class science performed on Frontier if not found. In this work, we describe and evaluate two strategies for finding hardware-level faults in Frontier’s 9,408 compute nodes. There are two strategies developed: the first uses the Slurm scheduler to scavenge available compute time to run the node screen, the second builds upon the lessons learned in the first strategy and enforces a weekly screen of each node. Using June 2023 as a case study, we find that the first scheduling strategy consumed more than ten times the resources as the second scheduling strategy, but successfully detected five hardware defects in Frontier. We summarize the lessons learned while developing and running a node screen on the world’s first exascale supercomputer.

Hagerty, Nick↗
DOE OSTI2025

In situ multi-tier auto-ignition detection applied to dual-fuel combustion simulations

Here we use an anomaly detection methodology that is centered on analyzing fourth-order joint moments (co-kurtosis), particularly focusing on its application in auto-ignition of combustion problems with large numbers of species. Unsupervised anomaly detection is challenging to generalize across problem types and domains. A recent technique, centered on analyzing information in the fourth-order joint moment co-kurtosis, has shown promise, especially for high-dimensional scientific data. In this work we present developments to the co-kurtosis based anomaly detection method needed to make it effective and scalable for large-scale distributed scientific data, such as those generated by massively parallel simulations. An in situ co-kurtosis algorithm is employed as the anomaly detection method for identifying ignition kernels in simulations of turbulent combustion. Here, we extend an existing methodology which identifies regions of the domain where anomalies are present, and add another tier of anomaly detection where the individual samples contributing to the anomaly are identified. We apply this algorithm on-the-fly to a variety of turbulent reacting flow problems and compare it to the widely used (but significantly more expensive) chemical explosive mode analysis (CEMA). We demonstrate the ability of the method to detect and identify the onset of low and high temperature ignition which can be used for computational steering, as chemical and combustion anomalies occur intermittently at spatio-temporal locations unknown a priori. Finally, we apply our lightweight in situ algorithm to an exascale high-fidelity simulation with a total of 2.4 Trillion degrees of freedom, performed using an adaptive mesh refinement solver. Furthermore, through a scalability analysis, we show that the relative computational cost of this in-situ anomaly detection algorithm compared to an iteration of the reacting flow solver is negligible.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Numerically Relevant Timescales in the MG2 Microphysics Model

Climate models rely on parameterizations of a variety of processes in the atmospheric physics, but a common concern is that the temporal resolution is too coarse to consistently resolve the behavior that individual parameterizations are designed to capture. This study examines timescales numerically derived from the Morrison-Gettelman (MG2) microphysics as implemented within the Energy Exascale Earth System Model, version 1 (E3SMv1). Numerically relevant timescales in MG2 are derived by computing the eigenspectrum of its Jacobian. These timescales are found to often be smaller than the default 5 min time step used for MG2. The fast timescales are then heuristically connected to individual microphysics processes. By substepping a few particular rain processes within MG2, the time discretization error for those processes was considerably reduced with minimal additional expense to the overall microphysics. While this improvement has a substantial effect on the target processes and on the vertical distribution of stratiform-derived rain within E3SMv1, the overall model climate is found to not be sensitive to MG2 time step. Here, we hypothesize that this is because the surface climate does not depend strongly on certain process rates, especially MG2's rain evaporation rate.

58 GEOSCIENCES↗
DOE OSTI2024

An Evaluation of the Effect of Network Cost Optimization for Leadership Class Supercomputers

Dragonfly-based networks are an extensively deployed network topology in large-scale high-performance computing due to their cost-effectiveness and efficiency. The US will soon have three Exascale supercomputers for leadership class workloads deployed using dragonfly networks. Compared to indirect networks of similar scale, the dragonfly network has considerably reduced cable lengths, cable counts, and switch counts, resulting in significant network cost savings for a given system size, however, these cost reductions result in reduced global minimal paths and more challenging routing. Additionally, large scale dragonfly networks often require a taper at the global link level, resulting in less bisection bandwidth than is achievable in other traditional non-blocking topologies of equivalent scale. While dragonfly networks have been extensively studied, they have yet to be fully evaluated in an extreme scale (i.e., exascale) system that targets capability workloads. In this paper, we present the results of the first large scale evaluation of a dragonfly network on an exascale system (Frontier) and compare its behavior to a similar scale fat-tree network on a previous generation TOP500 system (Summit). This evaluation aims to determine the effect of network cost optimizations by measuring a tapered topology’s impact on capability workloads. Our evaluation is based on a collection of synthetic microbenchmarks, mini-apps, and full scale applications. It compares the scaling efficiencies of each benchmark between the dragonfly-based Frontier and the fat-tree-based Summit systems. Our results show that a dragonfly network is $\sim \mathbf{3 0 \%}$ more cost efficient than a fat-tree topology, which amortizes to $\sim 3 \%$ of an exascale system cost. Furthermore, while tapered dragonfly networks impose significant tradeoffs, the impacts are not as broad as initially thought and are mostly seen in applications with global communication patterns, particularly all-to-all (e.g., FFT-based algorithms), but also local communication patterns (e.g., nearest-neighbor algorithms) that are sensitive to network performance variability.

Khan, Awais↗