Search NASA⌕ Search

SEARCH · Search NASA

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2019

A Zoom-in Analysis of I/O Logs to Detect Root Causes of I/O Performance Bottlenecks

Scientific applications frequently spend a large fraction of their execution time in reading and writing data on parallel file systems. Identifying these I/O performance bottlenecks and attributing root causes are critical steps toward devising optimization strategies. Several existing studies analyze I/O logs of a set of benchmarks or applications that were run with controlled behaviors. However, there is still a lack of general approach that systematically identifies I/O performance bottlenecks for applications running "in the wild" on production systems. In this study, we have developed an analysis approach of "zooming in" from platform-wide to application-wide to job-level I/O logs for identifying I/O bottlenecks in arbitrary scientific applications. We analyze the logs collected on a Cray XC40 system in production over a two-month period. This study results in several insights for application developers to use in optimizing I/O behavior.

Wang, Teng↗
DOE OSTI2016

Parallel File System I/O Performance Testing On LANL Clusters

These are slides from a presentation on parallel file system I/O performance testing on LANL clusters. I/O is a known bottleneck for HPC applications. Performance optimization of I/O is often required. This summer project entailed integrating IOR under Pavilion and automating the results analysis. The slides cover the following topics: scope of the work, tools utilized, IOR-Pavilion test workflow, build script, IOR parameters, how parameters are passed to IOR, *run_ior: functionality, Python IOR-Output Parser, Splunk data format, Splunk dashboard and features, and future work.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

TunIO: An AI-powered Framework for Optimizing HPC I/O

I/O operations are a known performance bottleneck of HPC applications. To achieve good performance, users often employ an iterative multistage tuning process to find an optimal I/O stack configuration. However, an I/O stack contains multiple layers, such as high-level I/O libraries, I/O middleware, and parallel file systems, and each layer has many parameters. These parameters and layers are entangled and influenced by each other. The tuning process is time-consuming and complex. In this work, we present TunIO, an AI-powered I/O tuning framework that implements several techniques to balance the tuning cost and performance gain, including tuning the high-impact parameters first. Furthermore, TunIO analyzes the application source code to extract its I/O kernel while retaining all statements necessary to perform I/O. It utilizes a smart selection of high-impact configuration parameters of the given tuning objective. Finally, it uses a novel Reinforcement Learning (RL)-driven early stopping mechanism to balance the cost and performance gain. Experimental results show that TunIO leads to a reduction of up to ≈73% in tuning time while achieving the same performance gain when compared to H5Tuner. It achieves a significant performance gain/cost of 208.4 MBps/min (I/O bandwidth for each minute spent in tuning) over existing approaches under our testing.

Rajesh, Neeraj↗
DOE OSTI2019

Scalable Deep Learning via I/O Analysis and Optimization

Scalable deep neural network training has been gaining prominence because of the increasing importance of deep learning in a multitude of scientific and commercial domains. Consequently, a number of researchers have investigated techniques to optimize deep learning systems. Much of the prior work has focused on runtime and algorithmic enhancements to optimize the computation and communication. Despite these enhancements, however, deep learning systems still suffer from scalability limitations, particularly with respect to data I/O. This situation is especially true for training models where the computation can be effectively parallelized, leaving I/O as the major bottleneck. In fact, our analysis shows that I/O can take up to 90% of the total training time. Thus, in this article, we first analyze LMDB, the most widely used I/O subsystem of deep learning frameworks, to understand the causes of this I/O inefficiency. Based on our analysis, we propose LMDBIO—an optimized I/O plugin for scalable deep learning. LMDBIO includes six novel optimizations that together address the various shortcomings in existing I/O for deep learning. Lastly, our experimental results show that LMDBIO significantly outperforms LMDB in all cases and improves overall application performance by up to 65-fold on a 9,216-core system.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1996

The Galley Parallel File System

Most current multiprocessor file systems are designed to use multiple disks in parallel, using the high aggregate bandwidth to meet the growing I/0 requirements of parallel scientific applications. Many multiprocessor file systems provide applications with a conventional Unix-like interface, allowing the application to access multiple disks transparently. This interface conceals the parallelism within the file system, increasing the ease of programmability, but making it difficult or impossible for sophisticated programmers and libraries to use knowledge about their I/O needs to exploit that parallelism. In addition to providing an insufficient interface, most current multiprocessor file systems are optimized for a different workload than they are being asked to support. We introduce Galley, a new parallel file system that is intended to efficiently support realistic scientific multiprocessor workloads. We discuss Galley's file structure and application interface, as well as the performance advantages offered by that interface.

Nieuwejaar, Nils↗
NASA NTRS2013

Profiling and Improving I/O Performance of a Large-Scale Climate Scientific Application

Exascale computing systems are soon to emerge, which will pose great challenges on the huge gap between computing and I/O performance. Many large-scale scientific applications play an important role in our daily life. The huge amounts of data generated by such applications require highly parallel and efficient I/O management policies. In this paper, we adopt a mission-critical scientific application, GEOS-5, as a case to profile and analyze the communication and I/O issues that are preventing applications from fully utilizing the underlying parallel storage systems. Through in-detail architectural and experimental characterization, we observe that current legacy I/O schemes incur significant network communication overheads and are unable to fully parallelize the data access, thus degrading applications' I/O performance and scalability. To address these inefficiencies, we redesign its I/O framework along with a set of parallel I/O techniques to achieve high scalability and performance. Evaluation results on the NASA discover cluster show that our optimization of GEOS-5 with ADIOS has led to significant performance improvements compared to the original GEOS-5 implementation.

Exascale↗
NASA NTRS2012

High-Performance, Multi-Node File Copies and Checksums for Clustered File Systems

Modern parallel file systems achieve high performance using a variety of techniques, such as striping files across multiple disks to increase aggregate I/O bandwidth and spreading disks across multiple servers to increase aggregate interconnect bandwidth. To achieve peak performance from such systems, it is typically necessary to utilize multiple concurrent readers/writers from multiple systems to overcome various singlesystem limitations, such as number of processors and network bandwidth. The standard cp and md5sum tools of GNU coreutils found on every modern Unix/Linux system, however, utilize a single execution thread on a single CPU core of a single system, and hence cannot take full advantage of the increased performance of clustered file systems. Mcp and msum are drop-in replacements for the standard cp and md5sum programs that utilize multiple types of parallelism and other optimizations to achieve maximum copy and checksum performance on clustered file systems. Multi-threading is used to ensure that nodes are kept as busy as possible. Read/write parallelism allows individual operations of a single copy to be overlapped using asynchronous I/O. Multinode cooperation allows different nodes to take part in the same copy/checksum. Split-file processing allows multiple threads to operate concurrently on the same file. Finally, hash trees allow inherently serial checksums to be performed in parallel. Mcp and msum provide significant performance improvements over standard cp and md5sum using multiple types of parallelism and other optimizations. The total speed-ups from all improvements are significant. Mcp improves cp performance over 27x, msum improves md5sum performance almost 19x, and the combination of mcp and msum improves verified copies via cp and md5sum by almost 22x. These improvements come in the form of drop-in replacements for cp and md5sum, so are easily used and are available for download as open source software at http://mutil.sourceforge.net.

Kolano, Paul Z.↗
DOE OSTI2019

Cataloging the visible universe through Bayesian inference in Julia at petascale

A key task in astronomy is to locate astronomical objects in images and to characterize them according to physical parameters such as brightness, color, and morphology. This task, known as cataloging, is challenging for several reasons: many astronomical objects are much dimmer than the sky background, labeled data is generally unavailable, overlapping astronomical objects must be resolved collectively, and the datasets are enormous – terabytes now, petabytes soon. In this work, we infer an astronomical catalog from 55 TB of imaging data using Celeste, a Bayesian variational inference code written entirely in the high-productivity programming language Julia. Using over 1.3 million threads on 650,000 Intel Xeon Phi cores of the Cori Phase II supercomputer, Celeste achieves a peak rate of 1.54 DP PFLOP/s. Celeste is able to jointly optimize parameters for 188 M stars and galaxies, loading and processing 178 TB across 8192 nodes in 14.6 min. To achieve this, Celeste exploits parallelism at multiple levels (cluster, node, and thread) and accelerates I/O through Cori’s burst buffer. Julia’s native performance enables Celeste to employ high-level constructs without resorting to hand-written or generated low-level code (C/C++/Fortran) while still achieving petascale performance.

Regier, Jeffrey↗
DOE OSTI2016

Collaborative Project: Understanding Climate Model Biases in Tropical Atlantic and Their Impact on Simulations of Extreme Climate Events

Recent studies have revealed that among all the tropical oceans, the tropical Atlantic has experienced the most pronounced warming trend over the 20th century. Many extreme climate events affecting the U.S., such as hurricanes, severe precipitation and drought events, are influenced by conditions in the Gulf of Mexico and the Atlantic Ocean. It is therefore imperative to have accurate simulations of the climatic mean and variability in the Atlantic region to be able to make credible projections of future climate change affecting the U.S. and other countries adjoining the Atlantic Ocean. Unfortunately, almost all global climate models exhibit large biases in their simulations of tropical Atlantic climate. The atmospheric convection simulation errors in the Amazon region and the associated errors in the trade wind simulations are hypothesized to be a leading cause of the tropical Atlantic biases in climate models. As global climate models have resolutions that are too coarse to resolve some of the atmospheric and oceanic processes responsible for the model biases, we propose to use a high-resolution coupled regional climate model (CRCM) framework to address the tropical bias issue. We propose to combine the expertise in tropical coupled atmosphere-ocean modeling at Texas A&M University (TAMU) and the coupled land-atmosphere modeling expertise at Pacific Northwest National Laboratory (PNNL) to develop a comprehensive CRCM for the Atlantic sector within a general and flexible modeling framework. The atmospheric component of the CRCM will be the NCAR WRF model and the oceanic component will be the Rutgers/UCLA ROMS. For the land component, we will use CLM modified at PNNL to include more detailed representations of vegetation and soil hydrology processes. The combined TAMU-PNNL CRCM model will be used to simulate the Atlantic climate, and the associated land-atmosphere-ocean interactions at a horizontal resolution of 9 km or finer. A particular focus of the model development effort will be to optimize the performance of WRF and ROMS over several thousand of cores by focusing on both the parallel communication libraries and the I/O interfaces, in order to achieve the sustained throughput needed to perform simulations on such fine resolution grids. The CRCM model will be developed within the framework of the Coupler (CPL7) software that is part of the NCAR Community Earth System Model (CESM). Through efforts at PNNL and within the community, WRF and CLM have already been coupled via CPL7. Using the flux coupler approach for the whole CRCM model will allow us to flexibly couple WRF, ROMS, and CLM with each model running on its own grid at different resolutions. In addition, this framework will allow us to easily port parameterizations between CESM and the CRCM, and potentially allow partial coupling between the CESM and the CRCM. TAMU and PNNL will contribute cooperatively to this research endeavor. The TAMU team led by Chang and Saravanan has considerable experience in studying atmosphere-ocean interactions within tropical Atlantic sector and will focus on modeling issues that relate to coupling WRF and ROMS. The PNNL team led by Leung has extensive expertise in atmosphere-land interaction and will be responsible for improving the land surface parameterization. Both teams will jointly work on integrating WRF-ROMS and WRF-CLM to couple WRF, ROMS, and CLM through CPL7. Montuoro of the TAMU Supercomputing Center will be responsible for improving the MPI and Parallel IO interfaces of the CRCM. Both teams will contribute to the design and execution of the proposed numerical experiments and jointly perform analysis of the numerical experiments.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Unbalanced Parallel I/O: An Often-Neglected Side Effect of Lossy Scientific Data Compression

Lossy compression techniques have demonstrated promising results in significantly reducing the scientific data size while guaranteeing the compression error bounds. However, one important yet often neglected side effect of lossy scientific data compression is its impact on the performance of parallel I/O. Our key observation is that the compressed data size is often highly skewed across processes in lossy scientific compression. To understand this behavior, we conduct extensive experiments where we apply three lossy compressors MGARD, ZFP, and SZ, which are specifically designed and optimized for scientific data, to three real-world scientific applications Gray-Scott simulation, WarpX, and XGC. Our analysis result demonstrates that the size of the compressed data is always skewed even if the original data is evenly decomposed among processes. Such skewness widely exists in different scientific applications using different compressors as long as the information density of the data varies across processes. We then systematically study how this side effect of lossy scientific data compression impacts the performance of parallel I/O. We observe that the skewness in the sizes of the compressed data often leads to I/O imbalance, which can significantly reduce the efficiency of I/O bandwidth utilization if not properly handled. In addition, writing data concurrently to a single shared file through MPI-IO library is more sensitive to the unbalanced I/O loads. Therefore, we believe our research community should pay more attention to the unbalanced parallel I/O caused by lossy scientific data compression.

Wang, Xinying↗
DOE OSTI2017

High-Performance I/O: HDF5 for Lattice QCD

Practitioners of lattice QCD/QFT have been some of the primary pioneer users of the state-of-the-art high-performance-computing systems, and contribute towards the stress tests of such new machines as soon as they become available. As with all aspects of high-performance-computing, I/O is becoming an increasingly specialized component of these systems. In order to take advantage of the latest available high-performance I/O infrastructure, to ensure reliability and backwards compatibility of data files, and to help unify the data structures used in lattice codes, we have incorporated parallel HDF5 I/O into the SciDAC supported USQCD software stack. Here we present the design and implementation of this I/O framework. Our HDF5 implementation outperforms optimized QIO at the 10-20% level and leaves room for further improvement by utilizing appropriate dataset chunking.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

HEPnOS: a Specialized Data Service for High Energy Physics Analysis

In this paper, we present HEPnOS, a distributed data service for managing data produced by high-energy physics (HEP) experiments. Using HEPnOS, HEP applications can use HPC resources more effciently than traditional fle-based applications. The fle-based model leads to a rigid, chunk-based allocation of computational resources and limits the number of cores that can be used concurrently by an HEP application. The fundamental problem is that organizing domain-specifc data into fles inadvertently introduces a single, artifcial, confated tuning parameter that puts key optimization goals into confict: larger fle sizes reduce metadata overhead and thus improve I/O effciency, but smaller fle sizes provide more opportunity for workfow parallelism and load balancing. In this work, we introduce a domain-specifc data service that decouples that constraint so that data can be accessed and processed in its natural granularity while still maintaining I/O effciency. By removing the constraints introduced by fle handling we are able to obtain better scaling and make effcient use of more cores for processing a fxed-sized data sample. We demonstrate the improved scalability by using an application developed in the fle-based paradigm and comparing it to a version modifed to use HEPnOS.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2021

Optimizing Error-Bounded Lossy Compression for Scientific Data by Dynamic Spline Interpolation

Today's scientific simulations are producing vast volumes of data that cannot be stored and transferred efficiently because of limited storage capacity, parallel I/O bandwidth, and network bandwidth. The situation is getting worse over time because of the ever-increasing gap between relatively slow data transfer speed and fast-growing computation power in modern supercomputers. Error-bounded lossy compression is becoming one of the most critical techniques for resolving the big scientific data issue, in that it can significantly reduce the scientific data volume while guaranteeing that the reconstructed data is valid for users because of its compression-error-bounding feature. In this paper, we present a novel error-bounded lossy compressor based on a state-of-the-art prediction-based compression framework. Our solution exhibits substantially better compression quality than all of the existing error-bounded lossy compressors, with comparable compression speed. Specifically, our contribution is threefold. (1) We provide an in-depth analysis of why the best-existing prediction-based lossy compressor can only minimally improve the compression quality. (2) We propose a dynamic spline interpolation approach with a series of optimization strategies that can significantly improve the data prediction accuracy, substantially improving the compression quality in turn. (3) We perform a thorough evaluation using six real-world scientific simulation datasets across different science domains to evaluate our solution vs. all other related works. Experiments show that the compression ratio of our solution is higher than that of the second-best lossy compressor by 20%similar to 460% with the same error bound in most of the cases.

Zhao, Kai↗
DOE OSTI2018

Machine Learning Based Parallel I/O Predictive Modeling: A Case Study on Lustre File Systems

Parallel I/O hardware and software infrastructure is a key contributor to performance variability for applications running on large-scale HPC systems. This variability confounds efforts to predict application performance for characterization, modeling, optimization, and job scheduling. We propose a modeling approach that improves predictive ability by explicitly treating the variability and by leveraging the sensitivity of application parameters on performance to group applications with similar characteristics. We develop a Gaussian process-based machine learning algorithm to model I/O performance and its variability as a function of application and file system characteristics. We demonstrate the effectiveness of the proposed approach using data collected from the Edison system at the National Energy Research Scientific Computing Center. The results show that the proposed sensitivity-based models are better at prediction when compared with application-partitioned or unpartitioned models. We highlight modeling techniques that are robust to the outliers that can occur in production parallel file systems. Using the developed metrics and modeling approach, we provide insights into the file system metrics that have a significant impact on I/O performance.

I/O performance variability↗
DOE OSTI2019

Applying Machine Learning to Understand Write Performance of Large-scale Parallel Filesystems

In high-performance computing (HPC), I/O performance prediction offers the potential to improve the efficiency of scientific computing. In particular, accurate prediction can make runtime estimates more precise, guide users toward optimal checkpoint strategies, and better inform facility provisioning and scheduling policies. HPC I/O performance is notoriously difficult to predict and model, however, in large part because of inherent variability and a lack of transparency in the behaviors of constituent storage system components. In this work we seek to advance the state of the art in HPC I/O performance prediction by (1) modeling the mean performance to address high variability, (2) deriving model features from write patterns, system architecture and system configurations, and (3) employing Lasso regression model to improve model accuracy. We demonstrate the efficacy of our approach by applying it to a crucial subset of common HPC I/O motifs, namely, file-per-process checkpoint write workloads. We conduct experiments on two distinct production HPC platforms—Titan at the Oak Ridge Leadership Computing Facility and Cetus at the Argonne Leadership Computing Facility—to train and evaluate our models. We find that we can attain ≤ 30% relative error for 92.79% and 99.64% of the samples in our test set on these platforms, respectively.

Xie, Bing↗
NASA NTRS1990

Multi-level Hierarchical Poly Tree computer architectures

Based on the concept of hierarchical substructuring, this paper develops an optimal multi-level Hierarchical Poly Tree (HPT) parallel computer architecture scheme which is applicable to the solution of finite element and difference simulations. Emphasis is given to minimizing computational effort, in-core/out-of-core memory requirements, and the data transfer between processors. In addition, a simplified communications network that reduces the number of I/O channels between processors is presented. HPT configurations that yield optimal superlinearities are also demonstrated. Moreover, to generalize the scope of applicability, special attention is given to developing: (1) multi-level reduction trees which provide an orderly/optimal procedure by which model densification/simplification can be achieved, as well as (2) methodologies enabling processor grading that yields architectures with varying types of multi-level granularity.

Padovan, Joe↗
NASA NTRSDate not supplied

Rapid Lightweight Firmware Architecture of the Mobile Metamaterial Internal Co-Integrator Robot

The Mobile Metamaterial Internal Co-Integrator (MMIC-I) is a structure assembly and servicing robot for in-space servicing, assembly, and manufacturing of primary structures and infrastructure. MMIC-I is a battery-powered crawling robot that can travel through periodic structures such as trusses and open framework mechanical metamaterials. It does this through sequences of component extension, contraction, and gripping. This paper provides a detailed discussion of MMIC-I’s lightweight and rapidly developed firmware architecture, to enable demonstration of robot locomotion, secondary operations, and communications with a central command source. The rationale for the lightweight rapid development approach is to allow for assessment of long term system requirements in parallel with the mechatronics development, including optimization of system and subsystem power densities, to inform a future choice of flight ready software frameworks. MMIC-I system computing and I/O requirements are much lower than what is provided by proven baseline computing hardware for existing flight ready software frameworks such as the core Flight System, F prime, and the Robot Operating System. Development of earth gravity ground demonstration of the robotic systems is greatly benefited by limited power and mass factors for computing hardware. Here, we implement inter-process communication, commanding, and telemetry with the Espressif ESP32 module running the Arduino OS.

Damiana Catanoso↗
DOE OSTI2017

Experiments and Analyses of Data Transfers Over Wide-Area Dedicated Connections

Dedicated wide-area network connections are increasingly employed in high-performance computing and big data scenarios. One might expect the performance and dynamics of data transfers over such connections to be easy to analyze due to the lack of competing traffic. However, non-linear transport dynamics and end-system complexities (e.g., multi-core hosts and distributed filesystems) can in fact make analysis surprisingly challenging. We present extensive measurements of memory-to-memory and disk-to-disk file transfers over 10 Gbps physical and emulated connections with 0–366 ms round trip times (RTTs). For memory-to-memory transfers, profiles of both TCP and UDT throughput as a function of RTT show concave and convex regions; large buffer sizes and more parallel flows lead to wider concave regions, which are highly desirable. TCP and UDT both also display complex throughput dynamics, as indicated by their Poincare maps and Lyapunov exponents. For disk-to-disk transfers, we determine that high throughput can be achieved via a combination of parallel I/O threads, parallel network threads, and direct I/O mode. Our measurements also show that Lustre filesystems can be mounted over long-haul connections using LNet routers, although challenges remain in jointly optimizing file I/O and transport method parameters to achieve peak throughput.

Rao, Nageswara S.↗