A5044270964- Advanced Data Storage Technologies
- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Distributed systems and fault tolerance
- Particle physics theoretical and experimental studies
- High-Energy Particle Collisions Research
- Utopian, Dystopian, and Speculative Fiction
- Gothic Literature and Media Analysis
- Digital Games and Media
- Cloud Computing and Resource Management
- Environmental Conservation and Management
- Particle Detector Development and Performance
- Scientific Computing and Data Management
- Algorithms and Data Compression
- Caching and Content Delivery
- Embedded Systems Design Techniques
- Quantum Chromodynamics and Particle Interactions
- Sexuality, Behavior, and Technology
- Asian Culture and Media Studies
- Computer Graphics and Visualization Techniques
- Political Economy and Marxism
- Remote Sensing and LiDAR Applications
- Gender, Feminism, and Media
- Service-Oriented Architecture and Web Services
- Simulation Techniques and Applications
Argonne National Laboratory
2015-2025
Illinois Institute of Technology
2016
Office of Scientific and Technical Information
2012
Alpha Star (United States)
2000
Oklahoma State University
1982
University of Minnesota
1973
Dataset storage, exchange, and access play a critical role in scientific applications. For such purposes netCDF serves as portable, efficient file format programming interface, which is popular numerous application domains. However, the original interface does not provide an mechanism for parallel data storage access. In this work, we present new writing reading datasets. This derived with minimal changes from serial but defines semantics tailored high performance. The underlying I/O...
Developing and tuning computational science applications to run on extreme scale systems are increasingly complicated processes. Challenges such as managing memory access message-passing behavior made easier by tools designed specifically aid in these Tools that can help users better understand the of their application with respect I/O have not yet reached level utility necessary play a central role development tuning. This deficiency tool set means we poor understanding how specific...
Computational science applications are driving a demand for increasingly powerful storage systems. While many techniques available capturing the I/O behavior of individual application trial runs and specific components system, continuous characterization production system remains daunting challenge systems with hundreds thousands compute cores multiple petabytes storage. As result, these often designed without clear understanding diverse computational workloads they will support. In this...
Current leadership-class machines suffer from a significant imbalance between their computational power and I/O bandwidth. While Moore's law ensures that the of high-performance computing systems increases with every generation, same is not true for subsystems. The scalability challenges faced by existing parallel file respect to increasing number clients, coupled minimalistic compute node kernels running on these machines, call new paradigm meet requirements data-intensive scientific...
Today's top high performance computing systems run applications with hundreds of thousands processes, contain storage nodes, and must meet massive I/O requirements for capacity performance. These leadership-class face daunting challenges to deploying scalable systems. In this paper we present a case study the scalability on Intrepid, IBM Blue Gene/P system at Argonne Leadership Computing Facility. Listed in 5 fastest supercomputers 2008, Intrepid runs computational science intensive demands...
Computational science applications are driving a demand for increasingly powerful storage systems. While many techniques available capturing the I/O behavior of individual application trial runs and specific components system, continuous characterization production system remains daunting challenge systems with hundreds thousands compute cores multiple petabytes storage. As result, these often designed without clear understanding diverse computational workloads they will support.In this...
SUMMARY Exploding dataset sizes from extreme‐scale scientific simulations necessitates efficient data management and reduction schemes to mitigate I/O costs. With the discrepancy between bandwidth computational power, scientists are forced capture infrequently, thereby making collection an inherently lossy process. Although compression can be effective solution, random nature of real‐valued datasets renders lossless routines ineffective. These techniques also impose significant overhead...
Efficient handling of large volumes data is a necessity for exascale scientific applications and database systems. To address the growing imbalance between amount available storage being produced by high speed (FLOPS) processors on system, must be compressed to reduce total placed file General-purpose loss less compression frameworks, such as zlib bzlib2, are commonly used datasets requiring compression. Quite often, however, many sets compress poorly, referred hard-to-compress datasets, due...
The advent of exascale supercomputers heralds a new era scientific discovery, yet it introduces significant architectural challenges that must be overcome for MPI applications to fully exploit its potential. Among these is the adoption heterogeneous architectures, particularly integration GPUs accelerate computation. Additionally, complexity multithreaded programming models has also become critical factor in achieving performance at scale. efficient utilization hardware acceleration...
Efficient analytics of scientific data from extreme-scale simulations is quickly becoming a top-notch priority. The increasing simulation output sizes demand for paradigm shift in how conducted. In this paper, we argue that query-driven over compressed---rather than original, full-size---data promising strategy order to meet storage-and-I/O-bound application challenges. As proof-of-principle, propose parallel query processing engine, called ISABELA-QA designed and optimized knowledge priors...
Parallel I/O plays a crucial role for most data-intensive applications running on massively parallel systems like Blue Gene/L that provides the promise of delivering enormous computational capability. We designed and implemented highly scalable file architecture Gene/L, which leverages benefit hierarchical functional partitioning design system software with separate cores. The exploits scalability aspect GPFS (General File System) at backend, while using MPI as an interface between...
In addition to their role as simulation engines, modern supercomputers can be harnessed for scientific visualization. Their extensive concurrency, parallel storage systems, and high-performance interconnects mitigate the expanding size complexity of datasets prepare in situ visualization these data. ongoing research into testing volume rendering on IBM Blue Gene/P (BG/P), we measure performance disk I/O, rendering, compositing large datasets, evaluate bottlenecks with respect system-specific...
Concurrent accesses to the shared storage resources in current HPC machines lead severe performance degradation caused by I/O contention. In this study, we identify some key challenges efficiently handling interleaved data accesses, and propose a system-wide solution optimize global performance. We implemented tested several scheduling policies, including prioritizing specific applications leveraging burst buffers defer conflicting from another application and/or directing requests different...
We present a scalable technique for the simulation of collider events with multijet final states, based on an improved parton-level event file format and $\mathrm{I}/\mathrm{O}$. The method is implemented both leading- next-to-leading-order QCD calculations. perform comprehensive analysis computing performance validate our new framework using Higgs-boson plus production up to seven jets. make resulting code base available public use.
From the novels of Anne Rice to The Lost Boys, from Terminator cyberpunk science fiction, vampires and cyborgs have become strikingly visible figures within American popular culture, especially youth culture. In Consuming Youth, Rob Latham explains why, showing how film, other media deploy these ambiguous monsters embody work through implications a capitalist system in which both consume are consumed. Inspired by Marx's use cyborg vampire as metaphor for objectification physical labor...
The FLASH code is a computational science tool for simulating and studying thermonuclear reactions. program periodically outputs large checkpoint files (to resume calculation from particular point in time) smaller plot (for visualization analysis). Initial experiments on BlueGene/P spent excessive time input/output (I/O), making it difficult to do actual science. Our investigation of I/O revealed several locations the software stack where we could make improvements. Fixing data corruption...
The optimization of parallel I/O has become challenging because the increasing storage hierarchy, performance variability shared systems, and number factors in hardware software stacks that impact performance. In this paper, we perform an in-depth study complexity involved autotuning modeling, including architecture, stack, noise. We propose a novel hybrid model combining analytical models for communication operations black-box individual operations. experimental results show approach...
Accurate analysis of HPC storage system designs is contingent on the use I/O workloads that are truly representative expected use. However, analyses generally bound to specific workload modeling techniques such as synthetic benchmarks or trace replay mechanisms, despite fact no single technique appropriate for all cases. In this work, we present design IOWA, a novel abstraction allows arbitrary consumer components obtain from range diverse input sources. Thus, researchers can choose...
There are two popular parallel I/O programming styles used by modern scientific computational applications: unique-file and shared-file. Unique-file usually gives satisfactory performance, but its major drawback is that managing a large number of files can overwhelm the task post-simulation data processing. Shared-file produces fewer allows arrays partitioned among processes to be saved in canonical order. As processors on machines increases into thousands more, problem size turn global...