A General-purpose Distributed Programming System using Data-parallel Streams

Graph component labeling, which is a subset of the general graph coloring problem, is a computationally expensive operation in many important applications and simulations. A number of data-parallel algorithmic variations to the component labeling problem are possible and we explore their use with general purpose graphical processing units (GPGPUs) and with the CUDA GPU programming language. We discuss implementation issues and performance results on CPUs and GPUs using CUDA. We evaluated our system with real-world graphs. We show how to consider different architectural features of the GPU and the host CPUs and achieve high performance.

Download Full-text

Design and Build Reporting Acceptance Test Procedure (ATP) for Web-Based BTS Installation

Jurnal Jartel Jurnal Jaringan Telekomunikasi ◽

10.33795/jartel.v3i2.219 ◽

2016 ◽

Vol 3 (2) ◽

pp. 40-46

Author(s):

Yanna Maharastri

Keyword(s):

System Design ◽

Test Procedure ◽

General Purpose ◽

Acceptance Test ◽

Work Time ◽

Web Based ◽

Service Testing ◽

Report System ◽

Report Data ◽

Using Data

The general purpose of making an ATP (Acceptance Test Procedure) report is to test whether the system that has been worked on is in accordance with the function specifications (validation). The ATP report used by an agency is infrastructure, equipment installation, etc. The manufacture of ATP for BTS installation on telecommunication subcons is manual which can reduce a lot of work time and data discrepancies. Therefore, a web-based ATP report system was designed to be able to perform reports with proper coordinate validation prior to installation and can also save work time. The web-based ATP report can match the important points of BTS installation. System planning starts from data collection and analysis to be used as a web-based report. Data transmission will be accommodated on the server and will be stored in the data base using My SQL. After the design is complete, it can be seen that with the analysis stage of the system design and system design using Data Flow Diagrams (DFD), input output design, responses from each each user with the function of each form content. for user engineers there are 97.2% agree from some menu functions, for user documentation there are 97.5% agree with the system, for user owners there are 91.65% agree and also QoS (Quality Of Service) testing is carried out for several cellular operators namely Im3 ooredoo , XL, Telkomsel and also the Polynema wifi network.

Download Full-text

The Use of Data Explorer as a 3-D Reconstruction Tool for Microscopy Data Sets

Microscopy and Microanalysis ◽

10.1017/s143192760001254x ◽

1997 ◽

Vol 3 (S2) ◽

pp. 1131-1132

Author(s):

Jansma P.L ◽

M.A. Landis ◽

L.C. Hansen ◽

N.C. Merchant ◽

N.J. Vickers ◽

...

Keyword(s):

Volume Rendering ◽

Laser Scanning ◽

Three Dimensional ◽

Building Blocks ◽

General Purpose ◽

Data Sets ◽

Laser Scanning Confocal Microscope ◽

Use Of Data ◽

Using Data ◽

Microscopy Data

We are using Data Explorer (DX), a general-purpose, interactive visualization program developed by IBM, to perform three-dimensional reconstructions of neural structures from microscopic or optical sections. We use the program on a Silicon Graphics workstation; it also can run on Sun, IBM RS/6000, and Hewlett Packard workstations. DX comprises modular building blocks that the user assembles into data-flow networks for specific uses. Many modules come with the program, but others, written by users (including ourselves), are continually being added and are available at the DX ftp site, http://www.tc.cornell.edu/DXhttp://www.nice.org.uk/page.aspx?o=43210.Initally, our efforts were aimed at developing methods for isosurface- and volume-rendering of structures visible in three-dimensional stacks of optical sections of insect brains gathered on our Bio-Rad MRC-600 laser scanning confocal microscope. We also wanted to be able to merge two 3-D data sets (collected on two different photomultiplier channels) and to display them at various angles of view.

Download Full-text

AN EVALUATION OF MULTIPLE FEED-FORWARD NETWORKS ON GPUs

International Journal of Neural Systems ◽

10.1142/s0129065711002638 ◽

2011 ◽

Vol 21 (01) ◽

pp. 31-47 ◽

Cited By ~ 14

Author(s):

NOEL LOPES ◽

BERNARDETE RIBEIRO

Keyword(s):

Graphics Processing Unit ◽

Parallel Implementation ◽

Low Cost ◽

Back Propagation ◽

General Purpose ◽

Training System ◽

Graphics Hardware ◽

Processing Unit ◽

Data Parallel ◽

Graphics Processing

The Graphics Processing Unit (GPU) originally designed for rendering graphics and which is difficult to program for other tasks, has since evolved into a device suitable for general-purpose computations. As a result graphics hardware has become progressively more attractive yielding unprecedented performance at a relatively low cost. Thus, it is the ideal candidate to accelerate a wide variety of data parallel tasks in many fields such as in Machine Learning (ML). As problems become more and more demanding, parallel implementations of learning algorithms are crucial for a useful application. In particular, the implementation of Neural Networks (NNs) in GPUs can significantly reduce the long training times during the learning process. In this paper we present a GPU parallel implementation of the Back-Propagation (BP) and Multiple Back-Propagation (MBP) algorithms, and describe the GPU kernels needed for this task. The results obtained on well-known benchmarks show faster training times and improved performances as compared to the implementation in traditional hardware, due to maximized floating-point throughput and memory bandwidth. Moreover, a preliminary GPU based Autonomous Training System (ATS) is developed which aims at automatically finding high-quality NNs-based solutions for a given problem.

Download Full-text