A5073745784- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Embedded Systems Design Techniques
- Advanced Data Storage Technologies
- Interconnection Networks and Systems
- Scientific Computing and Data Management
- Digital Filter Design and Implementation
- Numerical Methods and Algorithms
- Low-power high-performance VLSI design
- Neural Networks and Applications
- Cloud Computing and Resource Management
- VLSI and FPGA Design Techniques
- Advanced Database Systems and Queries
- Social and Educational Sciences
- Matrix Theory and Algorithms
- Advanced Memory and Neural Computing
- Cellular Automata and Applications
- Peer-to-Peer Network Technologies
- Access Control and Trust
- Engineering Applied Research
- DNA and Biological Computing
- Cooperative Communication and Network Coding
- Blind Source Separation Techniques
- Advanced Optimization Algorithms Research
- Algorithms and Data Compression
University of Houston
2010-2024
University of Chicago
2016
Tekniska Högskolans Studentkår
2013
Norwegian Womens Public Health Association
2013
Khyber Teaching Hospital
2011
KTH Royal Institute of Technology
2004-2008
Uppsala University
2008
University of California, Santa Barbara
2004
Domus Medica
1997
Harvard University Press
1994
Advances in networking technologies will soon make it possible to use the global information infrastructure a qualitatively different way—as computational as well an resource. As described recent book The Grid: Blueprint for New Computing Infrastructure, this Grid connect nation’s computers, databases, instruments, and people seamless web of computing distributed intelligence, which can be used on-demand fashion problem-solving resource many fields human endeavor—and, particular, science...
In this work, we describe new strategies for scheduling and executing workflow applications on grid resources using the GrADS [Ken Kennedy et al., 2002] infrastructure. Workflow is based heuristic that use application component performance models. The executed a novel strategy to bind launch onto heterogeneous resources. We apply these in context of EMAN, bio-imaging application, grid. results our experiments show model based, in-advance 1.5 2.2 times better makespan than other existing...
To facilitate implementation of deep neural networks on embedded systems keeping memory and computation requirements low is critical, particularly for real-time mobile use. In this work, we propose a SqueezeNet inspired version U-Net image segmentation that achieves 12X reduction in model size to 32MB, 3.2X Multiplication Accumulation operations (MACs) from 287 billion ops 88 inference the CamVid data set while preserving accuracy. Our proposed Squeeze efficient both MACs performance results...
Summary form only given. The goal of the Grid Application Development Software (GrADS) project is to provide programming tools and an execution environment ease program development for grid. We present recent extensions GrADS software framework: 1. A new approach scheduling workflow computations, applied a 3D image reconstruction application; 2. simple stop/migrate/restart rescheduling grid applications, QR 3. process-swapping rescheduling, N-body simulation. Experiments validating these...
<b>Objectives</b> To estimate how many people object to storage of biological samples collected in health care Sweden and their use research withdraw previous consent. <b>Design</b> Cross sectional study register data. <b>Setting</b> Biobanks used Swedish care, 2005-6. <b>Population</b> Data on refusal consent were obtained for 1.4 million biobank per year from 20 21 counties. <b>Main outcome measures</b> Rates preliminary consent, confirmed refusal, withdrawal <b>Results</b> Patients...
Increased complexity of memory systems to ameliorate the gap between speed processors and has made it increasingly harder for compilers optimize an arbitrary code within a palatable amount time. With emergence multicore (CMP), multiprocessor (SMP) hybrid shared architectures, achieving high e ciency is becoming even more challenging. To address challenge achieve in performance critical applications, domain speci c frameworks have been developed that aid scheduling computations. We portable...
In this paper, we present an Open Grid Services Architecture (OGSA)-based decentralized allocation enforcement system, developed with emphasis on a consistent data model and easy integration into existing scheduling, workload management software at six independent high-performance computing centers forming known as SweGrid. The Swedish National Allocations Committee (SNAC) allocates resource quotas these to research projects requiring substantial computer time. Our the SweGrid Accounting...
Summary Important computational physics problems are often large‐scale in nature, and it is highly desirable to have robust high performing frameworks that can quickly address these problems. However, no trivial task determine whether a framework efficiently or scalable. The aim of this paper present various strategies for better understanding the performance any parallel solving PDEs. issues negatively impact time‐to‐solution discussed, we propose spectrum analysis enhance one's critical...
In this paper we present an adaptive and portable software library for the fast Fourier transform (FFT). The consists of a number composable blocks code called codelets, each computing part transform. actual FFT algorithm used by is determined at run-time selecting fastest strategy among all possible strategies, given available size. We also efficient automatic method generating modules using special-purpose compiler. generator written in C it generates codelets. shown to be flexible...
Moore's law, the doubling of transistors per unit area for each CMOS technology generation, is expected to continue throughout decade, while Dennard voltage scaling resulting in constant power stopped about a decade ago. The semiconductor industry's response loss and consequent challenges managing distribution dissipation has been leveled off clock rates, die performance gain reduced from factor 2.8 1.4 multi-core processor dies with increased cache sizes. Increased caches sizes offers...
Abstract The SweGrid Accounting System (SGAS) allocates capacity in collaborative Grid environments by coordinating enforcement of Grid‐wide usage limits as a means to offer guarantees and prevent overuse. SGAS employs credit‐based allocation model where is granted projects via quota allowances that can be spent across the resources. resources collectively enforce these soft, real‐time manner. built on service‐oriented principles with strong focus interoperability Web services standards....
We present the SweGrid Accounting System (SGAS) — a decentralized and standards-based system for Grid resource allocation enforcement that has been developed with an emphasis on uniform data model easy integration into existing scheduling workload management software. The tested at six high-performance computing centers comprising computational resource, addresses need soft, real-time quota across clusters. SGAS framework is based state-of-the-art Web services technologies. openness ubiquity...