Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “parallel I/O optimization”

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 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2011

Scientific Data Services -- A High-Performance I/O System with Array Semantics

As high-performance computing approaches exascale, the existing I/O system design is having trouble keeping pace in both performance and scalability. We propose to address this challenge by adopting database principles and techniques in parallel I/O systems. First, we propose to adopt an array data model because many scientific applications represent their data in arrays. This strategy follows a cardinal principle from database research, which separates the logical view from the physical layout of data. This high-level data model gives the underlying implementation more freedom to optimize the physical layout and to choose the most effective way of accessing the data. For example, knowing that a set of write operations is working on a single multi-dimensional array makes it possible to keep the subarrays in a log structure during the write operations and reassemble them later into another physical layout as resources permit. While maintaining the high-level view, the storage system could compress the user data to reduce the physical storage requirement, collocate data records that are frequently used together, or replicate data to increase availability and fault-tolerance. Additionally, the system could generate secondary data structures such as database indexes and summary statistics. We expect the proposed Scientific Data Services approach to create a “live” storage system that dynamically adjusts to user demands and evolves with the massively parallel storage hardware.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Efficient Parallelization of Irregular Applications on GPU Architectures

With the enlarging computation capacity of general Graphics Processing Units (GPUs), leveraging GPUs to accelerate parallel applications has become a critical topic in academia and industry. However, a wide range of irregular applications with the computation-/memory-intensive nature cannot easily achieve high GPU utilization. The challenges mainly involve the following aspects: first, data dependence leads to coarse-grained kernel and inefficient parallelism; second, heavy GPU memory usage may cause frequent memory evictions and extra overhead of I/O; third, specific computation patterns produce memory redundancies; last, workload balance and data reusability conjunctly benefit the overall performance, but there may exist a dynamic trade-off between them. Targeting these challenges, this dissertation proposes multiple optimizations to accelerate two real-world applications: many-body correlation functions to simulate nuclear physics in a large-scale scientific system; the other is the eALS-based matrix factorization recommendation system. To accelerate the calculations of many-body correlation functions, this dissertation presents three frameworks in GPU memory management and multi-GPU scheduling. Firstly, an optimized systematic GPU memory management framework, MemHC, utilizes a series of new memory reduction designs in GPU memory allocation, CPU/GPU communications, and GPU memory oversubscription. Secondly, an enhanced multi-GPU scheduling framework, MICCO, particularly by taking both data dimension (e.g., data reuse and data eviction) and computation dimension into account. MICCO designs a heuristic scheduling algorithm and a machine learning-based regression model to generate the optimal settings of a proposed new concept to manage the trade-off. Thirdly, a locality-aware multi-GPU scheduling framework. This scheduler leverages pipeline batch generation with a looking-ahead strategy by building local dependency graphs for memory transfer reduction and better data reuse, achieving up to 79.92% memory cost reduction and 1.67x speedup. To parallelize the eALS-based recommendation system, this dissertation proposes an efficient CPU/GPU heterogeneous recommendation system, HEALS. HEALS employs newly designed architecture-adaptive data formats to achieve load balance and good data locality on CPU and GPU. To mitigate the data dependence, HEALS presents a CPU/GPU collaboration model for both task parallelism and data parallelism with multiple kernel computation optimizations. In summary, this dissertation efficiently accelerates two typical irregular applications on GPUs by building four frameworks, including CPU/GPU collaboration, GPU memory management, and multi-GPU scheduling.

Wang, Qihan↗
DOE OSTI2025

Scaling Ultrahigh-Resolution E3SM Land Model for Leadership-Class Supercomputers

This paper presents advancements in scaling the ultrahigh-resolution E3SM Land Model (uELM) for deployment on leadership-class supercomputers, addressing the increased demand for km-scale Earth system modeling. By focusing on km-scale ELM simulations, we enhance predictive capabilities for climate interactions, facilitating improved responses to climate change impacts on energy systems, agriculture, and water resources. Our approach leverages innovative software architecture optimizations, sophisticated data handling techniques, and advanced parallel processing, achieving strong scalability on two leadership supercomputers (2400 nodes (105,600 cores) on Summit, and 1200 nodes (76,800 cores) on Frontier). Results from extensive scalability assessments on the Summit and Frontier also demonstrate outstanding I/O performance (close to 400 GB/s write throughput) and the model's ability to efficiently handle increasing computational demands. This study not only establishes uELM's capability for high-resolution simulations over vast geographical domains, but also sets a foundation for future Earth system modeling breakthroughs.

Wang, Dali [ORNL] (ORCID:0000000168065108)↗
DOE OSTI2025

Optimizing Management of Persistent Data Structures in High-Performance Analytics

Large-scale data analytics workflows ingest massive input data into various data structures, including graphs and key-value datastores. These data structures undergo multiple transformations and computations and are typically reused in incremental and iterative analytics workflows. Persisting in-memory views of these data structures enables reusing them beyond the scope of a single program run while avoiding repetitive raw data ingestion overheads. Memory-mapped I/O enables persisting in-memory data structures without data serialization and deserialization overheads. However, memory-mapped I/O lacks the key feature of persisting consistent snapshots of these data structures for incremental ingestion and processing. The obstacles to efficient virtual memory snapshots using memory-mapped I/O include background writebacks outside the application’s control, and the significantly high storage footprint of such snapshots. To address these limitations, we present Privateer, a memory and storage management tool that enables storage-efficient virtual memory snapshotting while also optimizing snapshot I/O performance. Here, we integrated Privateer into Metall, a state-of-the-art persistent memory allocator for C++, and the Lightning Memory-Mapped Database (LMDB), a widely-used key-value datastore in data analytics and machine learning. Privateer optimized application performance by 1.22× when storing data structure snapshots to node-local storage, and up to 16.7× when storing snapshots to a parallel file system. Privateer also optimizes storage efficiency of incremental data structure snapshots by up to 11× using data deduplication and compression.

Computer science↗
DOE OSTI2019

Implementation of analytic gradients for CCSD and EOM-CCSD using Cholesky decomposition of the electron-repulsion integrals and their derivatives: Theory and benchmarks

We present a general formulation of analytic nuclear gradients for the coupled-cluster with single and double substitution (CCSD) and equation-of-motion (EOM) CCSD energies computed using Cholesky decomposition (CD) representations of the electron repulsion integrals. By rewriting the correlated energy and response equations such that the storage of the largest four-index intermediates is eliminated, CD leads to a significant reduction in disk storage requirements, reduced I/O penalties, and an improved parallel performance. CD thus extends the scope of the systems that can be treated by (EOM-)CCSD methods, although analytic gradients in the framework of CD are needed to extend the applicability of (EOM-)CCSD methods in the context of geometry optimizations. This paper presents a formulation of analytic (EOM-)CCSD gradient within the CD framework and reports on the salient details of the corresponding implementation. The accuracy and the capabilities of analytic CD-based (EOM-)CCSD gradients are illustrated by benchmark calculations and several illustrative examples.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2019

iez: Resource Contention Aware Load Balancing for Large-Scale Parallel File Systems

Parallel I/O performance is crucial to sustaining scientific applications on large-scale High-Performance Computing (HPC) systems. However, I/O load imbalance in the underlying distributed and shared storage systems can significantly reduce overall application performance. There are two conflicting challenges to mitigate this load imbalance: (i) optimizing systemwide data placement to maximize the bandwidth advantages of distributed storage servers, i.e., allocating I/O resources efficiently across applications and job runs; and (ii) optimizing client-centric data movement to minimize I/O load request latency between clients and servers, i.e., allocating I/O resources efficiently in service to a single application and job run. Moreover, existing approaches that require application changes limit wide-spread adoption in commercial or proprietary deployments. We propose iez, an “end-to-end control plane” where clients transparently and adaptively write to a set of selected I/O servers to achieve balanced data placement. Our control plane leverages realtime load information for distributed storage server global data placement while our design model leverages trace-based optimization techniques to minimize I/O load request latency between clients and servers. We evaluate our proposed system on an experimental cluster for two common use cases: synthetic I/O benchmark IOR for large sequential writes and a scientific application I/O kernel, HACC-I/O. Results show read and write performance improvements of up to 34% and 32%, respectively, compared to the state of the art.

Wadhwa, Bharti↗
DOE OSTI2018

Optimizing the performance of reactive molecular dynamics simulations for many-core architectures

Reactive molecular dynamics simulations are computationally demanding. Reaching spatial and temporal scales where interesting scientific phenomena can be observed requires efficient and scalable implementations on modern hardware. In this article, we focus on optimizing the performance of the widely used LAMMPS/ReaxC package for many-core architectures. As hybrid parallelism allows better leverage of the increasing on-node parallelism, we adopt thread parallelism in the construction of bonded and nonbonded lists and in the computation of complex ReaxFF interactions. To mitigate the I/O overheads due to large volumes of trajectory data produced and to save users the burden of post-processing, we also develop a novel in situ tool for molecular species analysis. We analyze the performance of the resulting ReaxC-OMP package on two different architectures: (i) Mira, an IBM Blue Gene/Q system and (ii) Cori-II, a Cray XC-40 sytem with Knights Landing processors. For Pentaerythritol tetranitrate (PETN) systems of sizes ranging from 32 thousand to 16.6 million particles, we observe speedups in the range of 1.5-4.5x. We observe sustained performance improvements for up to 262,144 cores (1,048,576 processes) of Mira and a weak scaling efficiency of 91.5% in large simulations containing 16.6 million particles. Finally, the in situ molecular species analysis tool incurs only insignificant overheads across various system sizes and runs configurations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

Writing ROOT Data in Parallel with TBufferMerger

Experiments at the Large Hadron Collider (LHC) produce tens of petabytes of new data in ROOT format per year that need to be processed and analysed. In the next decade, following the planned upgrades of the LHC and its detectors, this data production rate is expected to increase at least ten-fold. Therefore, optimizing the ROOT I/O subsystem is of critical importance to the success of the LHC physics programme. This contribution presents ROOT’s approach for writing data from multiple threads to a single output file in an efficient way. Technical aspects of the implementation—the TBufferMerger class—and programming model examples are described. Measurements of runtime performance and the overall improvement relative to the case of serial data writing are also discussed.

47 OTHER INSTRUMENTATION↗
DOE OSTI2024

Boosting RDataFrame performance with transparent bulk event processing

RDataFrame is ROOT’s high-level interface for Python and C++ data analysis. Since it first became available, RDataFrame adoption has grown steadily and it is now poised to be a major component of analysis software pipelines for LHC Run 3 and beyond. Thanks to its design inspired by declarative programming principles, RDataFrame enables the development of highperformance, highly parallel analyses without requiring expert knowledge of multi-threading and I/O: user logic is expressed in terms of self-contained, small computation kernels tied together by a high-level API. This design completely decouples analysis logic from its actual execution, and opens several interesting avenues for workflow optimization. In particular, in this work we explore the benefits of moving internal data processing from an event-by-event to a bulkby-bulk loop. This refactoring dramatically reduces the framework’s runtime overheads; in collaboration with the I/O layer it improves data access patterns; it exposes information that optimizing compilers might use to auto-vectorize the invocation of user-defined computations; finally, while existing user-facing interfaces remain unaffected, it becomes possible to additionally offer interfaces that explicitly expose bulks of events, useful e.g. for the injection of GPU kernels into the analysis workflow. In order to inform similar future R&D, design challenges will be presented, as well as an investigation of the relevant timememory trade-off backed by novel performance benchmarks.

Guiraud, Enrico↗
DOE OSTI2023

Parallel Memory-Independent Communication Bounds for SYRK

In this paper, we focus on the parallel communication cost of multiplying a matrix with its transpose, known as a symmetric rank-k update (SYRK). SYRK requires half the computation of general matrix multiplication because of the symmetry of the output matrix. Recent work (Beaumont et al., SPAA '22) has demonstrated that the sequential I/O complexity of SYRK is also a constant factor smaller than that of general matrix multiplication. Inspired by this progress, we establish memory-independent parallel communication lower bounds for SYRK with smaller constants than general matrix multiplication, and we show that these constants are tight by presenting communication-optimal algorithms. The crux of the lower bound proof relies on extending a key geometric inequality to symmetric computations and analytically solving a constrained nonlinear optimization problem. Here, the optimal algorithms use a triangular blocking scheme for parallel distribution of the symmetric output matrix and corresponding computation.

Communication costs↗
DOE OSTI2024

Improving I/O-aware Workflow Scheduling via Data Flow Characterization and trade-off Analysis

The scientific computing paradigm has transitioned from compute-intensive to I/O-intensive and memory-intensive in the past decade, especially when data-driven science has become common practice. Numerous empirical I/O-aware scheduling optimizations have been developed by incorporating I/O capacity and bandwidth as constraints into scheduling. Unfortunately, there is a lack of data flow (I/O) characterization tool and an understanding of trade-offs between concurrency, locality, and I/O bandwidth. To bridge the gap, this work 1) presents a set of descriptors to characterize, organize, and visualize I/O profiles, including flow size, I/O bandwidth, and operation count, which group data flows by I/O types, tasks, and files; 2) proposes an I/O Roofline model-based trade-off analysis to find the optimal trade-off between flow operational intensity, concurrency, and flow performance. The I/O descriptors generate useful insights into complicated I/O behaviors, suggesting distinct concurrency, storage, and scheduling to be used by types, tasks, and files. The proposed trade-off analysis guides scheduling decisions that generate resource assignment with the best flow parallelism. We evaluate our I/O-aware scheduling methodology on a highly I/O-intensive workflow–1000 Genomes. The experimental results demonstrate speedups of up to 2.4× compared to the state-of-the- art methods.

Guo, Luanzheng [BATTELLE (PACIFIC NW LAB)]↗
DOE OSTI2013

Acceleration of the matrix multiplication of Radiance three phase daylighting simulations with parallel computing on heterogeneous hardware of personal computer

Building designers are increasingly relying on complex fenestration systems to reduce energy consumed for lighting and HVAC in low energy buildings. Radiance, a lighting simulation program, has been used to conduct daylighting simulations for complex fenestration systems. Depending on the configurations, the simulation can take hours or even days using a personal computer. This paper describes how to accelerate the matrix multiplication portion of a Radiance three-phase daylight simulation by conducting parallel computing on heterogeneous hardware of a personal computer. The algorithm was optimized and the computational part was implemented in parallel using OpenCL. The speed of new approach was evaluated using various daylighting simulation cases on a multicore central processing unit and a graphics processing unit. Based on the measurements and analysis of the time usage for the Radiance daylighting simulation, further speedups can be achieved by using fast I/O devices and storing the data in a binary format.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2017

A Unit Testing Framework for Scientific Legacy Code

Large-scale scientific applications play important roles in supporting research. However, it is often very expensive and time-consuming to make changes to, maintain and evolve the scientific code due to its complexity and poor programming skills of researchers. Therefore, in order to visualize scientific code architecture to optimize software design, understand undocumented source code, and analyze software flow and functionality, we first introduce a unit testing framework (UTF). Then, because such infrastructure's performance is very crucial in practical use since the scientific legacy applications simulate instances in a long period of time, we improve the UTF by applying Message Passing based Parallelization and parallel I/O operations. Furthermore, due to the scientific code has enormous state data and the I/O capacity on the server is limited, we apply in situ data analysis method to encounter fewer resource limitations, and adopt signal processing to greatly reduce data transfer. Last, we demonstrated the correctness and high-efficiency of our framework for legacy Earth model on Titan supercomputer.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

Damsel: A Data Model Storage Library for Exascale Science

Computational science applications have been described as having one of seven motifs (the “seven dwarfs”), each having a particular pattern of computation and communication. From a storage and I/O perspective, these applications can also be grouped into a number of data model motifs describing the way data is organized and accessed during simulation, analysis, and visualization. Major storage data models developed in the 1990s, such as Network Common Data Format (netCDF) and Hierarchical Data Format (HDF) projects, created support for more complex data models. Development of both netCDF and HDF5 was influenced by multi-dimensional dataset storage requirements, but their access models and formats were designed with sequential storage in mind (e.g., a POSIX I/O model). Although these and other high-level I/O libraries have had a beneficial impact on large parallel applications, they do not always attain a high percentage of peak I/O performance due to fundamental design limitations, and they do not address the full range of current and future computational science data models. The goal of this project is to enable exascale computational science applications to interact conveniently and efficiently with storage through abstractions that match their data models. The project consists of three major activities: (1) identifying major data model motifs in computational science applications and developing representative benchmarks; (2) developing a data model storage library, called Damsel, that supports these motifs, provides efficient storage data layouts, incorporates optimizations to enable exascale operation, and is tolerant to failures; and (3) productizing Damsel and working with computational scientists to encourage adoption of this library by the scientific community. The product of this project, Damsel library, is openly available for download from http://cucis.ece.northwestern.edu/projects/DAMSEL. Several case studies and application programming interface reference are also available to assist new users to learn to use the library.

97 MATHEMATICS AND COMPUTING↗
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 OSTI2012

Active Flash: Performance-Energy Tradeoffs for Out-of-Core Processing on Non-Volatile Memory Devices

In this abstract, we study the performance and energy tradeoffs involved in migrating data analysis into the flash device, a process we refer to as Active Flash. The Active Flash paradigm is similar to 'active disks', which has received considerable attention. Active Flash allows us to move processing closer to data, thereby minimizing data movement costs and reducing power consumption. It enables true out-of-core computation. The conventional definition of out-of-core solvers refers to an approach to process data that is too large to fit in the main memory and, consequently, requires access to disk. However, in Active Flash, processing outside the host CPU literally frees the core and achieves real 'out-of-core' analysis. Moving analysis to data has long been desirable, not just at this level, but at all levels of the system hierarchy. However, this requires a detailed study on the tradeoffs involved in achieving analysis turnaround under an acceptable energy envelope. To this end, we first need to evaluate if there is enough computing power on the flash device to warrant such an exploration. Flash processors require decent computing power to run the internal logic pertaining to the Flash Translation Layer (FTL), which is responsible for operations such as address translation, garbage collection (GC) and wear-leveling. Modern SSDs are composed of multiple packages and several flash chips within a package. The packages are connected using multiple I/O channels to offer high I/O bandwidth. SSD computing power is also expected to be high enough to exploit such inherent internal parallelism within the drive to increase the bandwidth and to handle fast I/O requests. More recently, SSD devices are being equipped with powerful processing units and are even embedded with multicore CPUs (e.g. ARM Cortex-A9 embedded processor is advertised to reach 2GHz frequency and deliver 5000 DMIPS; OCZ RevoDrive X2 SSD has 4 SandForce controllers, each with 780MHz max frequency Tensilica core). Efforts that take advantage of the available computing cycles on the processors on SSDs to run auxiliary tasks other than actual I/O requests are beginning to emerge. Kim et al. investigate database scan operations in the context of processing on the SSDs, and propose dedicated hardware logic to speed up scans. Also, cluster architectures have been explored, which consist of low-power embedded CPUs coupled with small local flash to achieve fast, parallel access to data. Processor utilization on SSD is highly dependent on workloads and, therefore, they can be idle during periods with no I/O accesses. We propose to use the available processing capability on the SSD to run tasks that can be offloaded from the host. This paper makes the following contributions: (1) We have investigated Active Flash and its potential to optimize the total energy cost, including power consumption on the host and the flash device; (2) We have developed analytical models to analyze the performance-energy tradeoffs for Active Flash, by treating the SSD as a blackbox, this is particularly valuable due to the proprietary nature of the SSD internal hardware; and (3) We have enhanced a well-known SSD simulator (from MSR) to implement 'on-the-fly' data compression using Active Flash. Our results provide a window into striking a balance between energy consumption and application performance.

42 ENGINEERING↗
DOE OSTI2015

An Approach to Enhance pnetCDF Performance in Environmental Modeling Applications

I/O has been considered as a bottleneck in parallel applications. The software package, pnetCDF which works with parallel file systems, was developed to address this issue and provide parallel I/O capability. This study examines the performance of a novel approach which performs data aggregation along either row or column dimension of the spatial domain, and then applies the pnetCDF parallel I/O paradigm. The test was done with three different domain sizes which represents small, moderately large and large data domains, using a small scale Community Multi-scale Air Quality model (CMAQ) mocked up code. The examination includes comparing I/O performance with traditional serial I/O technique, straight application of pnetCDF, and the data aggregation along row and column dimension before applying pnetCDF. After the comparison, “optimal” I/O configurations for this new novel approach were quantified. Data aggregation along the row dimension (pnetCDFcr) works better than along the column dimension (pnetCDFcc) although it may perform slightly worse than straight the pnetCDF method with a small number of processors. When the number of processors becomes larger, pnetCDFcr out performs pnetCDF significantly. If the number of processors keeps increasing, pnetCDF reaches a point that the performance is even worse than the serial I/O technique. This new approach has also been tested on a real application where it performs two times better than the straight pnetCDF paradigm.

I/O↗
DOE OSTI2015

Cluster-to-cluster data transfer with data compression over wide-area networks

The recent emergence of ultra high-speed networks up to 100 Gb/s has posed numerous challenges and has led to many investigations on efficient protocols to saturate 100 Gb/s links. However, end-to-end data transfers involve many components, not only protocols, affecting overall transfer performance. These components include disk I/O subsystem, additional computation associated with data streams, and network adapters. For example, achievable bandwidth by TCP may not be implementable if disk I/O or CPU becomes a bottleneck in end-to-end data transfer. In this paper, we first model all the system components involved in end-to-end data transfer as a graph. We then formulate the problem whose goal is to achieve maximum data transfer throughput using parallel data flows. We also propose a variable data flow GridFFP XIO stack to improve data transfer with data compression. Our contributions lie in how to optimize data transfers considering all the system components involved rather than in accurately modeling all the system components involved. Our proposed formulations and solutions are evaluated through experiments on the ESnet 100G testbed and a wide-area cluster-to-cluster testbed. The experimental results on the ESnet 100G testbed show that our approach is several times faster than Globus Online-8 x faster for datasets with many 10 MB files and 3-4 x faster for other datasets of larger size files. The experimental results on the cluster-to-cluster testbed show that our variable data flow approach is up to 4 x faster than a normal cluster data transfer.

Disk-to-Disk Transfer↗