Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Parallel Supercomputers”

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 163 records · Page 9

Provisioning a Multi-Tiered Data Staging Area for Extreme-Scale Machines

Massively parallel scientific applications, running on extreme-scale supercomputers, produce hundreds of terabytes of data per run, driving the need for storage solutions to improve their I/O performance. Traditional parallel file systems (PFS) in high performance computing (HPC) systems are unable to keep up with such high data rates, creating a storage wall. In this work, we present a novel multi-tiered storage architecture comprising hybrid node-local resources to construct a dynamic data staging area for extreme-scale machines. Such a staging ground serves as an impedance matching device between applications and the PFS. Our solution combines diverse resources (e.g., DRAM, SSD) in such a way as to approach the performance of the fastest component technology and the cost of the least expensive one. We have developed an automated provisioning algorithm that aids in meeting the checkpointing performance requirement of HPC applications, by using a least-cost storage configuration. We evaluate our approach using both an implementation on a large scale cluster and a simulation driven by six-years worth of Jaguar supercomputer job-logs, and show that our approach, by choosing an appropriate storage configuration, achieves 41.5% cost savings with only negligible impact on performance.

99 GENERAL AND MISCELLANEOUS↗

Designing Energy Efficient Communication Runtime Systems for Data Centric Programming Models

The insatiable demand of high performance computing is being driven by the most computationally intensive applications such as computational chemistry, climate modeling, nuclear physics, etc. The last couple of decades have observed a tremendous rise in supercomputers with architectures ranging from traditional clusters to system-on-a-chip in order to achieve the petaflop computing barrier. However, with advent of petaflop-plus computing, we have ushered in an era where power efficient system software stack is imperative for execution on exascale systems and beyond. At the same time, computationally intensive applications are exploring programming models beyond traditional message passing, as a combination of Partitioned Global Address Space (PGAS) languages and libraries, providing one-sided communication paradigm with put, get and accumulate primitives. To support the PGAS models, it is critical to design power efficient and high performance one-sided communication runtime system. In this paper, we design and implement PASCoL, a high performance power aware one-sided communication library using Aggregate Remote Memory Copy Interface (ARMCI), the communication runtime system of Global Arrays. For various communication primitives provided by ARMCI, we study the impact of Dynamic Voltage/Frequency Scaling (DVFS) and a combination of interrupt (blocking)/polling based mechanisms provided by most modern interconnects. We implement our design and evaluate it with synthetic benchmarks using an InfiniBand cluster. Our results indicate that PASCoL can achieve significant reduction in energy consumed per byte transfer without additional penalty for various one-sided communication primitives and various message sizes and data transfer patterns.

99 GENERAL AND MISCELLANEOUS↗

Probabilistic Communication and I/O Tracing with Deterministic Replay at Scale

With today's petascale supercomputers, applications often exhibit low efficiency, such as poor communication and I/O performance, that can be diagnosed by analysis tools. However, these tools either produce extremely large trace files that complicate performance analysis, or sacrifice accuracy to collect high-level statistical information using crude averaging. This work contributes Scala-H-Trace, which features more aggressive trace compression than any previous approach, particularly for applications that do not show strict regularity in SPMD behavior. Scala-H-Trace uses histograms expressing the probabilistic distribution of arbitrary communication and I/O parameters to capture variations. Yet, where other tools fail to scale, Scala-H-Trace guarantees trace files of near constant size, even for variable communication and I/O patterns, producing trace files orders of magnitudes smaller than using prior approaches. We demonstrate the ability to collect traces of applications running on thousands of processors with the potential to scale well beyond this level. We further present the first approach to deterministically replay such probabilistic traces (a) without deadlocks and (b) in a manner closely resembling the original applications. Our results show either near constant sized traces or only sub-linear increases in trace file sizes irrespective of the number of nodes utilized. Even with the aggressively compressed histogram-based traces, our replay times are within 12% to 15% of the runtime of original codes. Such concise traces resembling the behavior of production-style codes closely and our approach of deterministic replay of probabilistic traces are without precedence.

99 GENERAL AND MISCELLANEOUS↗

Modeling of wave processes in blocky media with porous and fluid-saturated interlayers

The wave processes in blocky media are analyzed by applying different mathematical models, wherein the elastic blocks interact with each other via pliant interlayers with the complex mechanical properties. Four versions of constitutive equations are considered. In the first version, an elastic interaction between the blocks is simulated within the framework of linear elasticity theory, and the model of elastic–plastic interlayers is constructed to take into account the appearance of irreversible deformation of interlayers at short time intervals. In the second one, the effects of viscoelastic shear in the interblock interlayers are taken into the consideration using the Poynting–Thomson rheological scheme. In the third option, the model of an elastic porous material is used in the interlayers, where the pores collapse if an abrupt compressive stress is applied. In the fourth case, the model of a fluid-saturated material with open pores is examined based on Biot's equations. The collapse of pores is modeled by the generalized rheological approach, wherein the mechanical properties of a material are simulated using four rheological elements. Three of them are the traditional elastic, viscous and plastic elements, the fourth element is the so-called rigid contact, which is used to describe the behavior of materials with the different resistance to tension and compression. It was shown that the thermodynamically consistent model is provided, which means that the energy balance equation is fulfilled for an entire blocky structure, where the kinetic and potential energy of the system is the sum of the kinetic and potential energies of the blocks and interlayers. Under numerical implementation of the interlayers models, the dissipationless finite difference Ivanov's method was used. The splitting method by spatial variables in the combination with the Godunov gap decay scheme was applied in the blocks. As a result, robust and stable computational algorithms are built and tested. Using MPI technology, the parallel software was designed for the modeling of wave processes in 2D setting. The numerical results are presented, discussed and future studies are outlined. - Highlights: • We model the propagation of waves in a blocky medium with compliant interlayers. • Models of porous and fluid saturated interlayers are proposed using Biot's approach. • A parallel computational algorithm is developed for supercomputers of cluster type. • Problems of instant rotation of central and boundary blocks are solved numerically. • Wave motions are compared in the cases of elastic and fluid-saturated interlayers.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Effective Ion Mobility Calculations for Macromolecules by Scattering off Electron Clouds

Broad commercialization and increasing resolving power of ion mobility spectrometry/mass spectrometry (IMS/MS) platforms has engendered an explosion of IMS applications to structural characterization of gas-phase biomolecules. That has renewed interest in more accurate and rapid ion mobility calculations needed to elicit ion geometries from the measurements. An approach based on scattering off electron density isosurfaces (SEDI) that mirrors the physics of molecular collisions was proven superior to the common methods involving atomic coordinates a decade ago, but has remained impractical for large ions because of extreme computational demands. Here, we accelerate SEDI by up to ~500 times using the fragment molecular orbital (FMO) approach for surface generation and the multiplexed scattering algorithm in conjunction with the new grid extrapolation procedure for cross section evaluations. Parallelization of the code on a supercomputer has produced major further speed gains, allowing SEDI calculations for proteins (defined by over a million surface points) with the precision of <0.1% in one minute. Initial tests reveal the anticipated dependences of mobility on the ion charge state and lower cross sections in view of reduced surface roughness. Present developments are expected to lead to broad application of SEDI in IMS studies of macromolecules, enabling more accurate and reliable structural assignments.

Alexeev, Yuri↗

Fast Fault Injection and Sensitivity Analysis for Collective Communications, In: 2015 IEEE International Conference on Cluster Computing

The collective communication operations, which are widely used in parallel applications for global communication and synchronization are critical for application's performance and scalability. However, how faulty collective communications impact the application and how errors propagate between the application processes is largely unexplored. One of the critical reasons for this situation is the lack of fast evaluation method to investigate the impacts of faulty collective operations. The traditional random fault injection methods relying on a large amount of fault injection tests to ensure statistical significance require a significant amount of resources and time. These methods result in prohibitive evaluation cost when applied to the collectives. In this paper, we introduce a novel tool named Fast Fault Injection and Sensitivity Analysis Tool (FastFIT) to conduct fast fault injection and characterize the application sensitivity to faulty collectives. The tool achieves fast exploration by reducing the exploration space and predicting the application sensitivity using Machine Learning (ML) techniques. A basis for these techniques are implicit correlations between MPI semantics, application context, critical application features, and application responses to faulty collective communications. The experimental results show that our approach reduces the fault injection points and tests by 97% for representative benchmarks (NAS Parallel Benchmarks (NPB)) and a realistic application (Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS)) on a production supercomputer. Further, we statistically generalize the application sensitivity to faulty collective communications for these workloads, and present correlation between application features and the sensitivity.

Computer Science↗

Hashing Strategies for the Cray XMT

Several strategies for hashing on a Cray XMT are experimentally compared. These strategies are designed to minimize memory contention. Results show good scalability and indicate under what circumstances each of the strategies should be used.

97 MATHEMATICS AND COMPUTING↗

User Manual - HydraGNN v5.0: Distributed Implementation of Multi-Tasking Graph Neural Networks

This document serves as the user manual for HydraGNN v5.0, a scalable graph neural network (GNN) architecture for simultaneous prediction of multiple target properties using multi-task learning (MTL). This version of HydraGNN has been developed primarily to support the development, training, and deployment of predictive graph-based deep learning (DL) models for atomistic materials modeling. HydraGNN is templated over 13 message-passing policies, including invariant models (GIN, PNA, PNAPlus, GAT, MFC, CGCNN, SAGE, SchNet, DimeNet) and equivariant models (EGNN, PNAEq, PAINN, MACE), and supports distributed training via distributed data parallelism (DDP), DeepSpeed, and Fully Sharded Data Parallelism (FSDP) on leadership-class supercomputers. Although HydraGNN can be applied to problems beyond atomistic materials modeling, its current use is confined to homogeneous graphs. Additional capabilities include machine-learned interatomic potentials with energy-conserving forces, General, Powerful, and Scalable Graph Transformer (GraphGPS) global attention, periodic boundary conditions, hyperparameter optimization, mixed-precision training, and uncertainty quantification.

97 MATHEMATICS AND COMPUTING↗

Viskores: Integrating Parallel Scientific Visualization Research into Applications

Viskores is a scientific visualization library that is the primary deployment of such algorithms to the parallel accelerated processors of modern DOE supercomputers. In this paper, we review the capabilities provided by Viskores and how these capabilities are leveraged by other software in the high-performance computing ecosystem. We discuss the Viskores data representation and pay particular attention to array management. Through this array management we describe how data is adapted between Viskores and other software along with strategies for converting dynamic, polymorphic objects to static representations better suited to GPU processing. We conclude with several examples of Viskores integrating with high-performance software that is used in production today.

Moreland, Ken [ORNL] (ORCID:0000000270513288)↗

Scale-up Unlearnable Examples Learning with High-performance Computing

Recent advancements in AI models, like ChatGPT, are structured to retain user interactions, which could inadvertently include sensitive healthcare data. In the healthcare field, particularly when radiologists use AI-driven diagnostic tools hosted on online platforms, there is a risk that medical imaging data may be repurposed for future AI training without explicit consent, spotlighting critical privacy and intellectual property concerns around healthcare data usage. Addressing these privacy challenges, a novel approach known as Unlearnable Examples (UEs) has been introduced, aiming to make data unlearnable to deep learning models. A prominent method within this area, called Unlearnable Clustering (UC), has shown improved UE performance with larger batch sizes but was previously limited by computational resources (e.g., a single workstation). To push the boundaries of UE performance with theoretically unlimited resources, we scaled up UC learning across various datasets using Distributed Data Parallel (DDP) training on the Summit supercomputer. Our goal was to examine UE efficacy at high-performance computing (HPC) levels to prevent unauthorized learning and enhance data security, particularly exploring the impact of batch size on UE’s unlearnability. Utilizing the robust computational capabilities of the Summit, extensive experiments were conducted on diverse datasets such as Pets, MedMNist, Flowers, and Flowers102. Our findings reveal that both overly large and overly small batch sizes can lead to performance instability and affect accuracy. However, the relationship between batch size and unlearnability varied across datasets, highlighting the necessity for tailored batch size strategies to achieve optimal data protection. The use of Summit’s high-performance GPUs, along with the efficiency of the DDP framework, facilitated rapid updates of model parameters and consistent training across nodes. Our results underscore the critical role of selecting appropriate batch sizes based on the specific characteristics of each dataset to prevent learning and ensure data security in deep learning applications. The source code is publicly available at https: // github. com/ hrlblab/ UE_ HPC .

Zhu, Yanfan [Vanderbilt University, Nashville, TN,↗

Large Data Visualization on Distributed Memory Mulit-GPU Clusters

Data sets of immense size are regularly generated on large scale computing resources. Even among more traditional methods for acquisition of volume data, such as MRI and CT scanners, data which is too large to be effectively visualization on standard workstations is now commonplace. One solution to this problem is to employ a 'visualization cluster,' a small to medium scale cluster dedicated to performing visualization and analysis of massive data sets generated on larger scale supercomputers. These clusters are designed to fit a different need than traditional supercomputers, and therefore their design mandates different hardware choices, such as increased memory, and more recently, graphics processing units (GPUs). While there has been much previous work on distributed memory visualization as well as GPU visualization, there is a relative dearth of algorithms which effectively use GPUs at a large scale in a distributed memory environment. In this work, we study a common visualization technique in a GPU-accelerated, distributed memory setting, and present performance characteristics when scaling to extremely large data sets.

97 MATHEMATICS AND COMPUTING↗

Effects of floating-point non-associativity on numerical computations on massively multithreaded systems

Floating-point addition and multiplication are not necessarily associative. When performing those operations over large numbers of operands with different magnitudes, the order in which individual operations are performed can affect the final result. On massively multithreaded systems, when performing parallel reductions, the non-deterministic nature of numerical operation interleaving can lead to non-deterministic numerical results. We have investigated the effect of this problem on the convergence of a conjugate gradient calculation used as part of a power grid analysis application.

97 MATHEMATICS AND COMPUTING↗

Enabling a Highly-Scalable Global Address Space Model for Petascale Computing

Over the past decade, the trajectory to the petascale has been built on increased complexity and scale of the underlying parallel architectures. Meanwhile, software de- velopers have struggled to provide tools that maintain the productivity of computational science teams using these new systems. In this regard, Global Address Space (GAS) programming models provide a straightforward and easy to use addressing model, which can lead to improved produc- tivity. However, the scalability of GAS depends directly on the design and implementation of the runtime system on the target petascale distributed-memory architecture. In this paper, we describe the design, implementation, and optimization of the Aggregate Remote Memory Copy Interface (ARMCI) runtime library on the Cray XT5 2.3 PetaFLOPs computer at Oak Ridge National Laboratory. We optimized our implementation with the flow intimation technique that we have introduced in this paper. Our optimized ARMCI implementation improves scalability of both the Global Arrays (GA) programming model and a real-world chemistry application NWChem from small jobs up through 180,000 cores.

97 MATHEMATICS AND COMPUTING↗

DataStager: scalable data staging services for petascale applications

Known challenges for petascale machines are that (1) the costs of I/O for high performance applications can be substantial, especially for output tasks like checkpointing, and (2) noise from I/O actions can inject undesirable delays into the runtimes of such codes on individual compute nodes. This paper introduces the flexible 'DataStager' framework for data staging and alternative services within that jointly address (1) and (2). Data staging services moving output data from compute nodes to staging or I/O nodes prior to storage are used to reduce I/O overheads on applications total processing times, and explicit management of data staging offers reduced perturbation when extracting output data from a petascale machine's compute partition. Experimental evaluations of DataStager on the Cray XT machine at Oak Ridge National Laboratory establish both the necessity of intelligent data staging and the high performance of our approach, using the GTC fusion modeling code and benchmarks running on 1000+ processors.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Modern gyrokinetic particle-in-cell simulation of fusion plasmas on top supercomputers

The Gyrokinetic Toroidal Code at Princeton (GTC-P) is a highly scalable and portable particle-in-cell (PIC) code. It solves the 5D Vlasov-Poisson equation featuring efficient utilization of modern parallel computer architectures at the petascale and beyond. Motivated by the goal of developing a modern code capable of dealing with the physics challenge of increasing problem size with sufficient resolution, new thread-level optimizations have been introduced as well as a key additional domain decomposition. GTC-P's multiple levels of parallelism, including inter-node 2D domain decomposition and particle decomposition, as well as intra-node shared memory partition and vectorization have enabled pushing the scalability of the PIC method to extreme computational scales. In this paper, we describe the methods developed to build a highly parallelized PIC code across a broad range of supercomputer designs. This particularly includes implementations on heterogeneous systems using NVIDIA GPU accelerators and Intel Xeon Phi (MIC) co-processors and performance comparisons with state-of-the-art homogeneous HPC systems such as Blue Gene/Q. New discovery science capabilities in the magnetic fusion energy application domain are enabled, including investigations of Ion-Temperature-Gradient (ITG) driven turbulence simulations with unprecedented spatial resolution and long temporal duration. Performance studies with realistic fusion experimental parameters are carried out on multiple supercomputing systems spanning a wide range of cache capacities, cache-sharing configurations, memory bandwidth, interconnects and network topologies. Furthermore, these performance comparisons using a realistic discovery-science-capable domain application code provide valuable insights on optimization techniques across one of the broadest sets of current high-end computing platforms worldwide.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Multi-task Parallelism for Robust Pre-training of Graph Foundation Models on Multi-source, Multi-fidelity Atomistic Modeling Data

Graph foundation models using graph neural networks promise sustainable, efficient atomistic modeling. To tackle challenges of processing multi-source, multi-fidelity data during pre-training, recent studies employ multi-task learning, in which shared message passing layers initially process input atomistic structures regardless of source, then route them to multiple decoding heads that predict data-specific outputs. This approach stabilizes pre-training and enhances a model’s transferability to unexplored chemical regions. Preliminary results on approximately four million structures are encouraging, yet questions remain about generalizability to larger, more diverse datasets and scalability on supercomputers. We propose a multi-task parallelism method that distributes each head across computing resources with GPU acceleration. Implemented in the open-source HydraGNN architecture, our method was trained on over 24 million structures from five datasets and tested on the Perlmutter, Aurora, and Frontier supercomputers, demonstrating efficient scaling on all three highly heterogeneous super-computing architectures.

Lupo Pasini, Massimiliano [ORNL] (ORCID:0000000249↗

h5bench: A unified benchmark suite for evaluating HDF5 I/O performance on pre‐exascale platforms

Summary Parallel I/O is a critical technique for moving data between compute and storage subsystems of supercomputers. With massive amounts of data produced or consumed by compute nodes, high‐performant parallel I/O is essential. I/O benchmarks play an important role in this process; however, there is a scarcity of I/O benchmarks representative of current workloads on HPC systems. Toward creating representative I/O kernels from real‐world applications, we have created h5bench , a set of I/O kernels that exercise hierarchical data format version 5 (HDF5) I/O on parallel file systems in numerous dimensions. Our focus on HDF5 is due to the parallel I/O library's heavy usage in various scientific applications running on supercomputing systems. The various tests benchmarked in the h5bench suite include I/O operations (read and write), data locality (arrays of basic data types and arrays of structures), array dimensionality (one‐dimensional arrays, two‐dimensional meshes, three‐dimensional cubes), I/O modes (synchronous and asynchronous). In this paper, we present the observed performance of h5bench executed along several of these dimensions on existing supercomputers (Cori and Summit) and pre‐exascale platforms (Perlmutter, Theta, and Polaris). h5bench measurements can be used to identify performance bottlenecks and their root causes and evaluate I/O optimizations. As the I/O patterns of h5bench are diverse and capture the I/O behaviors of various HPC applications, this study will be helpful to the broader supercomputing and I/O community.

97 MATHEMATICS AND COMPUTING↗

Final Scientific Report: A Scalable Development Environment for Peta-Scale Computing

This document is the final scientific report of the project DE-SC000120 (A scalable Development Environment for Peta-Scale Computing). The objective of this project is the extension of the Parallel Tools Platform (PTP) for applying it to peta-scale systems. PTP is an integrated development environment for parallel applications. It comprises code analysis, performance tuning, parallel debugging and system monitoring. The contribution of the Juelich Supercomputing Centre (JSC) aims to provide a scalable solution for system monitoring of supercomputers. This includes the development of a new communication protocol for exchanging status data between the target remote system and the client running PTP. The communication has to work for high latency. PTP needs to be implemented robustly and should hide the complexity of the supercomputer's architecture in order to provide a transparent access to various remote systems via a uniform user interface. This simplifies the porting of applications to different systems, because PTP functions as abstraction layer between parallel application developer and compute resources. The common requirement for all PTP components is that they have to interact with the remote supercomputer. E.g. applications are built remotely and performance tools are attached to job submissions and their output data resides on the remote system. Status data has to be collected by evaluating outputs of the remote job scheduler and the parallel debugger needs to control an application executed on the supercomputer. The challenge is to provide this functionality for peta-scale systems in real-time. The client server architecture of the established monitoring application LLview, developed by the JSC, can be applied to PTP's system monitoring. LLview provides a well-arranged overview of the supercomputer's current status. A set of statistics, a list of running and queued jobs as well as a node display mapping running jobs to their compute resources form the user display of LLview. These monitoring features have to be integrated into the development environment. Besides showing the current status PTP's monitoring also needs to allow for submitting and canceling user jobs. Monitoring peta-scale systems especially deals with presenting the large amount of status data in a useful manner. Users require to select arbitrary levels of detail. The monitoring views have to provide a quick overview of the system state, but also need to allow for zooming into specific parts of the system, into which the user is interested in. At present, the major batch systems running on supercomputers are PBS, TORQUE, ALPS and LoadLeveler, which have to be supported by both the monitoring and the job controlling component. Finally, PTP needs to be designed as generic as possible, so that it can be extended for future batch systems.

97 MATHEMATICS AND COMPUTING↗