SEARCH · Engineering Papers

Results for “High Performance Computing (HPC)”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 289 records · Page 16

All sources NASA NTRS DOE OSTI

DOE OSTI2023

Advanced Computing Annual Report 2022

National Renewable Energy Laboratory's (NREL) high-performance computing (HPC), efficient data center operation, and state-of-the-art data visualization powered over 300 clean energy projects in Fiscal Year 2022 (FY22). 555 scientists and engineers utilized HPC resources to assess and integrate data into their research. The report highlights the arrival of new HPC Kestrel in 2023, the contributions of the Insight Center, and researchers looking into bias in artificial intelligence, alongside key research collaborations using advanced computing.

advanced computing↗

DOE OSTI2013

McrEngine: A Scalable Checkpointing System Using Data-Aware Aggregation and Compression

High performance computing (HPC) systems use checkpoint-restart to tolerate failures. Typically, applications store their states in checkpoints on a parallel file system (PFS). As applications scale up, checkpoint-restart incurs high overheads due to contention for PFS resources. The high overheads force large-scale applications to reduce checkpoint frequency, which means more compute time is lost in the event of failure. We alleviate this problem through a scalable checkpoint-restart system, mcrEngine. McrEngine aggregates checkpoints from multiple application processes with knowledge of the data semantics available through widely-used I/O libraries, e.g., HDF5 and netCDF, and compresses them. Our novel scheme improves compressibility of checkpoints up to 115% over simple concatenation and compression. Our evaluation with large-scale application checkpoints show that mcrEngine reduces checkpointing overhead by up to 87% and restart overhead by up to 62% over a baseline with no aggregation or compression.

Islam, Tanzima Zerin↗

DOE OSTI2014

Computation Annual Report (2013)

Computation provides the intellectual electricity that helps make Lawrence Livermore National Laboratory (LLNL) a thriving and innovative research and development (R&D) institution. Our hardware, software, and discipline experts are integrated into almost every field of research at LLNL. Our in-house programs, software applications, and support initiatives help achieve mission-critical national-security objectives for the U.S. Departments of Energy, Homeland Security, and Defense, as well as other federal and state agencies. We deliver comprehensive support to a wide set of scientific disciplines that include physical sciences (materials modeling, fluid dynamics), biology (genomic sequencing), mathematics (scalable algorithms), and computer sciences (parallel processing, real-time control systems). This year, we have expanded our outreach with high performance computing (HPC) collaborations and made great strides in our capabilities related to mathematical computation, control systems, and computation infrastructure, many of them in concert with partners in industry and academia. The Computation workforce remains highly motivated and passionate about melding computational expertise with cutting-edge science and technology to enable breakthroughs. 2013 has once again proven that Computation provides the foundation and support that are essential for productivity at LLNL. Every scientific discipline and program leverages the tools and expertise provided by our staff. We remain at the forefront of HPC moving towards exascale and have developed the infrastructure and proficiency to make this an achievable goal. Our researchers repeatedly deliver outstanding research, attracting collaborations with the biggest names in academia and industry, while joining forces with other national laboratories to further push the bounds of what is scientifically possible. Computation’s mission is echoed by the President of the United States in his assertion that “our future depends on reaffirming America’s role as the world’s engine of scientific discovery and technological innovation.” Driven by this lofty goal, we employ our state-of-the-art computational capabilities to ensure LLNL’s position as a global science and technology leader.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Enabling machine learning-ready HPC ensembles with Merlin

With the growing complexity of computational and experimental facilities, many scientific researchers are turning to machine learning (ML) techniques to analyze large scale ensemble data. With complexities such as multi-component workflows, heterogeneous machine architectures, parallel file systems, and batch scheduling, care must be taken to facilitate this analysis in a high performance computing (HPC) environment. Here, we present Merlin, a workflow framework to enable large ML-friendly ensembles of scientific HPC simulations. By augmenting traditional HPC with distributed compute technologies, Merlin aims to lower the barrier for scientific subject matter experts to incorporate ML into their analysis. As a producer–consumer workflow model, Merlin enables multi-machine, cross-batch job, dynamically allocated yet persistent workflows capable of utilizing surge-compute resources. Key features of Merlin are a flexible HPC-centric interface, low per-task overhead, multi-tiered fault recovery, and a hierarchical sampling algorithm that allows for $\mathscr{O}$(N) task execution and $\mathscr{O}$(N ln N) task queuing to ensembles of millions of tasks. In addition to Merlin’s design, we test the algorithm’s performance in an HPC center and demonstrate the ability to enqueue 40 million simulations in 100 s, with a 30 millisecond per-task overhead that is independent of ensemble size. Finally, we describe some example applications that Merlin has enabled on leadership-class HPC resources, such as the ML-augmented optimization of nuclear fusion experiments and the calibration of infectious disease models to study the progression of and possible mitigation strategies for COVID-19.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Evaluation of LLVM Flang for Production HPC Applications and Modern Fortran Features

In 2025, LLVM released its first Flang Fortran compiler version considered ready for widespread evaluation. We know of no published assessment of Flang compiling a workload- derived portfolio of high-performance computing (HPC) applications. We address this gap using workload data from the National Energy Research Scientific Computing Center (NERSC), which supports more than 10,000 scientists on approximately 1,000 projects. The NERSC workload analyses identify many Fortran components in heavily used applications. We selected 10 such packages with available source code. We compiled them with Flang 22.1.3 on NERSC’s Perlmutter system. Six compiled without code modifications, though some required build-system changes. Three compiled after minor source edits, mostly to address Fortran standard violations. One built only without OpenMP enabled. We evaluated seven additional packages selected for their use of, or enablement of, standard Fortran parallel features: multi-image execution and do concurrent. Six such codes compiled with most or all unit tests passing.

Rasmussen, Katherine↗

DOE OSTI2013

Parallel Implementation of Power System Dynamic Simulation

Dynamic simulation of power system transient stability is important for planning, monitoring, operation, and control of electrical power systems. However, modeling the system dynamics and network involves the computationally intensive time-domain solution of numerous differential and algebraic equations (DAE). This results in a transient stability implementation that may not maintain the real-time constraints of an online security assessment. This paper presents a parallel implementation of the dynamic simulation on a high-performance computing (HPC) platform using parallel simulation algorithms and computation architectures. It enables the simulation to run even faster than real time, enabling the “look-ahead” capability of upcoming stability problems in the power grid.

dynamic simulation↗

DOE OSTI2022

DRAS: Deep Reinforcement Learning for Cluster Scheduling in High Performance Computing

Cluster schedulers are crucial in high-performance computing (HPC). They determine when and which user jobs should be allocated to available system resources. Existing cluster scheduling heuristics are developed by human experts based on their experience with specific HPC systems and workloads. However, the increasing complexity of computing systems and the highly dynamic nature of application workloads have placed tremendous burden on manually designed and tuned scheduling heuristics. More aggressive optimization and automation are needed for cluster scheduling in HPC. In this work, we present an automated HPC scheduling agent named DRAS (Deep Reinforcement Agent for Scheduling) by leveraging deep reinforcement learning. DRAS is built on a hierarchical neural network incorporating special HPC scheduling features such as resource reservation and backfilling. An efficient training strategy is presented to enable DRAS to rapidly learn the target environment. Once being provided a specific scheduling objective given by the system manager, DRAS automatically learns to improve its policy through interaction with the scheduling environment and dynamically adjusts its policy as workload changes. We implement DRAS into a HPC scheduling platform called CQGym. CQGym provides a common platform allowing users to flexibly evaluate DRAS and other scheduling methods such as heuristic and optimization methods. Furthermore, the experiments using CQGym with different production workloads demonstrate that DRAS outperforms the existing heuristic and optimization approaches by up to 50%.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2016

The impact of the U.S. supercomputing initiative will be global

Last July, President Obama issued an executive order that created a coordinated federal strategy for HPC research, development, and deployment called the U.S. National Strategic Computing Initiative (NSCI). However, this bold, necessary step toward building the next generation of supercomputers has inaugurated a new era for U.S. high performance computing (HPC).

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2015

Data Transfer Study HPSS Archiving

The movement of the large amounts of data produced by codes run in a High Performance Computing (HPC) environment can be a bottleneck for project workflows. To balance filesystem capacity and performance requirements, HPC centers enforce data management policies to purge old files to make room for new computation and analysis results. Users at Oak Ridge Leadership Computing Facility (OLCF) and many other HPC user facilities must archive data to avoid data loss during purges, therefore the time associated with data movement for archiving is something that all users must consider. This study observed the difference in transfer speed from the originating location on the Lustre filesystem to the more permanent High Performance Storage System (HPSS). The tests were done with a number of different transfer methods for files that spanned a variety of sizes and compositions that reflect OLCF user data. This data will be used to help users of Titan and other Cray supercomputers plan their workflow and data transfers so that they are most efficient for their project. We will also discuss best practice for maintaining data at shared user facilities.

Big Data↗

DOE OSTI2020

The Installation of Direct Water-Cooling Systems to Reduce Cooling Energy Requirements for High-Performance Computing Centers

A large cluster of High-Performance Computing (HPC) equipment at the Lawrence Livermore National Laboratory in California was retrofitted with an Asetek cooling system. The Asetek system is a hybrid scheme with water-cooled cold plates on high-heat-producing components in the information technology (IT) equipment, and with the remainder of the heat being removed by conventional air-cooling systems. In order to determine energy savings of the Asetek system, data were gathered and analyzed two ways: using top-down statistical models, and bottom-up engineering models. The cluster, “Cabernet”, rejected its heat into a facilities cooling water loop which in turn rejected the heat into the same chilled water system serving the computer-room air handlers (CRAHs) that provided the air-based cooling for the room. Because the “before” and “after” cases both reject their heat into the chilled water system, the only savings is due to reduction in CRAH fan power. The top-down analysis showed a 4% overall energy savings for the data center (power usage effectiveness (PUE) —the ratio of total data center energy to IT energy— dropped from 1.60 to 1.53, lower is better); the bottom-up analysis showed a 3% overall energy savings (PUE from 1.70 to 1.66) and an 11% savings for the Cabernet system by itself (partial PUE of 1.51). Greater savings, on the order of 15-20%, would be possible if the chilled water system was not used for rejecting the heat from the Asetek system. About 37% of the heat from the Cab system was rejected to the cooling water, lower than at other installations.

Earni, Shankar↗

DOE OSTI2023

Comparing the Performance of Julia on CPUs versus GPUs and Julia-MPI versus Fortran-MPI: a case study with MPAS-Ocean (Version 7.1)

Abstract. Some programming languages are easy to develop at the cost of slow execution, while others are fast at runtime but much more difficult to write. Julia is a programming language that aims to be the best of both worlds – a development and production language at the same time. To test Julia's utility in scientific high-performance computing (HPC), we built an unstructured-mesh shallow water model in Julia and compared it against an established Fortran-MPI ocean model, the Model for Prediction Across Scales–Ocean (MPAS-Ocean), as well as a Python shallow water code. Three versions of the Julia shallow water code were created: for single-core CPU, graphics processing unit (GPU), and Message Passing Interface (MPI) CPU clusters. Comparing identical simulations revealed that our first version of the Julia model was 13 times faster than Python using NumPy, where both used an unthreaded single-core CPU. Further Julia optimizations, including static typing and removing implicit memory allocations, provided an additional 10–20× speed-up of the single-core CPU Julia model. The GPU-accelerated Julia code was almost identical in terms of performance to the MPI parallelized code on 64 processes, an unexpected result for such different architectures. Parallelized Julia-MPI performance was identical to Fortran-MPI MPAS-Ocean for low processor counts and ranges from 2× faster to 2× slower for higher processor counts. Our experience is that Julia development is fast and convenient for prototyping but that Julia requires further investment and expertise to be competitive with compiled codes. We provide advice on Julia code optimization for HPC systems.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

Machine Learning-Based Extreme Data Reduction for Prompt Supernova Pointing at DUNE

One of the goals of the Deep Underground Neutrino Experiment (DUNE) is to use the massive underground liquid argon time projection chamber (LArTPC) detectors at its far site for multimessenger astronomy (MMA), in the detection of neutrinos from core-collapse supernovae (SNe). Its current baseline trigger strategy detects activity in the detector that is consistent with supernova (SN) neutrinos and saves the raw data for further offline analysis but provides no prompt pointing information crucial for optical follow-ups by other observatories. This approach is based on the assumption that prompt pointing determination using raw data is computationally prohibitive. In this article, we demonstrate a proof-of-concept based on applying extreme data reduction on the buffered SN data in the DUNE data acquisition (DAQ) system’s front-end computers using a machine learning (ML) workflow. This reduces the data by ~5 orders of magnitude, allowing a full track reconstruction to be carried out quickly on a single server. The total time to perform the ML-based data reduction and the full track reconstruction is less than the time to transfer the SN data back to Fermilab or a high-performance computing (HPC) center. This shows that prompt processing of raw SN data is possible and, in fact, trivial once the data have been reduced to reject radiological backgrounds, paving the way to a high-quality SN pointing trigger that is based on fully reconstructed data instead of trigger primitives (TPs).

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2025

I/O in Machine Learning Applications on HPC Systems: A 360-degree Survey

Growing interest in Artificial Intelligence (AI) has resulted in a surge in demand for faster methods of Machine Learning (ML) model training and inference. This demand for speed has prompted the use of high performance computing (HPC) systems that excel in managing distributed workloads. Because data is the main fuel for AI applications, the performance of the storage and I/O subsystem of HPC systems is critical. In the past, HPC applications accessed large portions of data written by simulations or experiments or ingested data for visualizations or analysis tasks. ML workloads perform small reads spread across a large number of random files. This shift of I/O access patterns poses several challenges to modern parallel storage systems. In this paper, we survey I/O in ML applications on HPC systems, and target literature within a 6-year time window from 2019 to 2024. We define the scope of the survey, provide an overview of the common phases of ML, review available profilers and benchmarks, examine the I/O patterns encountered during offline data preparation, training, and inference, and explore I/O optimizations utilized in modern ML frameworks and proposed in recent literature. Lastly, we seek to expose research gaps that could spawn further R&D.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

A Scalable Pipeline for Gigapixel Whole Slide Imaging Analysis on Leadership Class HPC Systems

Whole Slide Imaging (WSI) captures microscopic details of a patient's histopathological features at multiple res-olutions organized across different levels. Images produced by WSI are gigapixel-sized, and saving a single image in memory requires a few gigabytes which is scarce since a complicated model occupies tens of gigabytes. Performing a simple met-ric operation on these large images is also expensive. High-performance computing (HPC) can help us quickly analyze such large images using distributed training of complex deep learning models. One popular approach in analyzing these images is to divide a WSI image into smaller tiles (patches) and then train a simpler model with these reduced-sized but large numbers of patches. However, we need to solve three pre-processing challenges efficiently for pursuing this patch-based approach. 1) Creating small patches from a high-resolution image can result in a high number (hundreds of thousands per image) of patches. Storing and processing these images can be challenging due to a large number of I/O and arithmetic operations. To reduce I/Oand memory accesses, an optimal balance between the size and number of patches must exist to reduce I/O and memory accesses. 2) WSI images may have tiny annotated regions for cancer tissue and a significant portion with normal and fatty tissues; correct patch sampling should avoid dataset imbalance. 3) storing and retrieving many patches to and from disk storage might incur I/O latency while training a deep learning model. An efficient distributed data loader should reduce I/O latency during the training and inference steps. This paper explores these three challenges and provides empirical and algorithmic solutions deployed on the Summit supercomputer hosted at the Oak Ridge Leadership Computing Facility.

Dash, Sajal↗

DOE OSTI2018

Hardware Evaluation: Abstract Machine Models and Proxy Architectures for Exascale Computing

The strategic and economic impacts of high performance computing (HPC) systems, over the last several decades, have enabled dramatic improvements in manufacturing, design, development, and research, across almost every sector of the U.S. economy, including genetic analysis for agriculture and medicine, oil exploration, electronics, aircraft, and automotive design, and of course national defense technologies. HPC has repeatedly compounded reductions in time-to-solution and time-to-market in each of these areas. In recognition of these strategic and economic impacts, the Department of Energy (DOE) created the Exascale Computing Project (ECP) as a broad-based technology project focusing on simultaneously co-evolving high-performance computing architecture, system software, and application software. While many details of the exascale architectures are undefined, communicating the form and direction of this co-evolution of advances, in each of these three components, across the project, presents a need for a common language through which these advances can be shared. In this document we present a series of abstract representations designed to allow application developers to focus on the aspects of the machine that are important or relevant to performance and code structure. These abstract machine models (AMMs) describe the proposed architectures at the component and system level and are intended as communication aids between application developers and hardware architects during the co-design process. In addition to providing a common language for communication, the constraints of protecting the intellectual property of ECP’s vendor partners requires that the architectural advances be presented in an abstract sense, rather than exposing the specifics of an individual vendor’s designs, while still providing enough detail such that application developers will be able to reason about the performance trade-offs in the design space of their applications.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2016

An asynchronous traversal engine for graph-based rich metadata management

Rich metadata in high-performance computing (HPC) systems contains extended information about users, jobs, data files, and their relationships. Property graphs are a promising data model to represent heterogeneous rich metadata flexibly. Specifically, a property graph can use vertices to represent different entities and edges to record the relationships between vertices with unique annotations. The high-volume HPC use case, with millions of entities and relationships, naturally requires an out-of-core distributed property graph database, which must support live updates (to ingest production information in real time), low-latency point queries (for frequent metadata operations such as permission checking), and large-scale traversals (for provenance data mining). Among these needs, large-scale property graph traversals are particularly challenging for distributed graph storage systems. Most existing graph systems implement a "level synchronous" breadth-first search algorithm that relies on global synchronization in each traversal step. This performs well in many problem domains; but a rich metadata management system is characterized by imbalanced graphs, long traversal lengths, and concurrent workloads, each of which has the potential to introduce or exacerbate stragglers (i.e., abnormally slow steps or servers in a graph traversal) that lead to low overall throughput for synchronous traversal algorithms. Previous research indicated that the straggler problem can be mitigated by using asynchronous traversal algorithms, and many graph-processing frameworks have successfully demonstrated this approach. Such systems require the graph to be loaded into a separate batch-processing framework instead of being iteratively accessed, however. In this work, we investigate a general asynchronous graph traversal engine that can operate atop a rich metadata graph in its native format. We outline a traversal-aware query language and key optimizations (traversal-affiliate caching and execution merging) necessary for efficient performance. We further explore the effect of different graph partitioning strategies on the traversal performance for both synchronous and asynchronous traversal engines. Our experiments show that the asynchronous graph traversal engine is more efficient than its synchronous counterpart in the case of HPC rich metadata processing, where more servers are involved and larger traversals are needed. Furthermore, the asynchronous traversal engine is more adaptive to different graph partitioning strategies.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

DOE OSTI2018

High-Performance Computing and Analysis-Led Development of High Efficiency Dilute Opposed Piston Gasoline Engine

Pinnacle is developing a multicylinder 1.2 L gasoline engine for automotive applications using high-performance computing (HPC) and analysis methods. Pinnacle and Oak Ridge National Laboratory executed large-scale multidimensional combustion analyses at the Oak Ridge Leadership Computing Facility to thoroughly explore the design space. These HPC-led investigations show high fuel efficiency (~46% gross indicated efficiency) may be achieved by operating with extremely high charge dilution levels of exhaust gas recirculation (EGR) at a light load key drive cycle condition (2000 RPM, 3 bar brake mean effective pressure (BMEP)), while simultaneously attaining high levels of fuel conversion efficiency and low NO x emissions. In this extremely dilute environment, the flame propagation event is supported by turbulence and bulk in-cylinder charge motion brought about by modulation of inlet port flow. This arrangement produces a load and speed adjustable amalgamation of swirl and counter-rotating tumble which provides the turbulence required to support stable low-temperature combustion. At higher load conditions, the engine may operate at more traditional combustion modes to generate competitive power. In this work, the numerical results from these HPC simulations are presented. Finally, further HPC simulations and test validations are underway and will be reported in future publications.

42 ENGINEERING↗

DOE OSTI2021

HAM: Hotspot-Aware Manager for Improving Communications with 3D-Stacked Memory

merging High-Performance Computing (HPC) workloads, such as graph analytics, machine learning, and big data science, are data-intensive. Data-intensive workloads usually present fine-grained memory accesses with limited or no data locality, and thus incur frequent cache misses and low utilization of memory bandwidth. 3D-stacked memory devices such as Hybrid Memory Cube (HMC) and High Bandwidth Memory (HBM) can provide significantly higher bandwidth than conventional memory modules. However, the traditional interfaces and optimization methods for JEDEC DDR devices do not allow to fully exploit the potential performance of 3D-stacked memory with the massive amount of irregular memory accesses of data-intensive applications. In this paper, we propose a novel Hotspot-Aware Manager (HAM) infrastructure for 3D-stacked memory devices capable of optimizing memory access streams via request aggregation, hotspot detection, and in-memory prefetching. %and an associated hotspot-aware page policy. We present the HAM design and implementation, and simulate it on a system using RISC-V embedded cores with attached HMC devices. We extensively evaluate HAM with over 12 benchmarks and applications representing diverse irregular memory access patterns. The results show that, on average, HAM reduces redundant requests by 37.51\% and increases the prefetch buffer hit rate by 4.2 times, compared to a baseline streaming prefetcher. On the selected benchmark set, HAM provides performance gains of 21.81\% in average (up to 34.28\%) and power savings of 35.07\% over a standard 3D-stacked memory.

Wang, Xi↗