Grid workflow: A flexible framework for fault tolerance in the grid. - Page 26 |
Save page Remove page | Previous | 26 of 180 | Next |
|
small (250x250 max)
medium (500x500 max)
Large (1000x1000 max)
Extra Large
large ( > 500x500)
Full Resolution
All (PDF)
|
This page
All
|
Following are two typical scientific applications in experimental study in structured
biology and problem solving in linear system, respectively. The examples emphasize the
coordinated and controlled used of multiple dedicated high-end computing resources when
scientists performing their daily work to solve their domain-specific problems.
• Scientists perform a series of tasks by acessing diverce types of computing resources
to accomplish their scientific experiments: a remote instrumental device for 3-D
raw data acquisition, one or more remote supercomputer(s) for reconstructed data
acquisition from the raw data set, and a special-purpose graphics computer for
visualization of the reconstructed data. Scientists need to acquire dedicated high-end
computers and instruments in advance, and statically configure them for the
experiments.
• In order to solve large linear equation problems, researchers are required to do
experimentation and testing with a sequence of computational steps such as analyzing
the given matrix, applying a reordering or scaling transformation, selecting
and applying a preconditional, and applying a solver among various alternatives.
Each computation might be required to explicitly or transparently access remote
numerical systems installed on specialized high-end computing resources.
Note that the above examples are often called loosely coupled applications in following
senses; first, no communication between activities/tasks (i.e., often called jobs in HPDC
context) is required during the execution (i.e., message exchange between tasks occurs
only during initialization and termination [102] of the tasks execution); second, there
are no strict constraints on the component resources in terms of different programming
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Object Description
Description
| Title | Grid workflow: A flexible framework for fault tolerance in the grid. - Page 26 |
| Repository email | cisadmin@lib.usc.edu |
| Full text | Following are two typical scientific applications in experimental study in structured biology and problem solving in linear system, respectively. The examples emphasize the coordinated and controlled used of multiple dedicated high-end computing resources when scientists performing their daily work to solve their domain-specific problems. • Scientists perform a series of tasks by acessing diverce types of computing resources to accomplish their scientific experiments: a remote instrumental device for 3-D raw data acquisition, one or more remote supercomputer(s) for reconstructed data acquisition from the raw data set, and a special-purpose graphics computer for visualization of the reconstructed data. Scientists need to acquire dedicated high-end computers and instruments in advance, and statically configure them for the experiments. • In order to solve large linear equation problems, researchers are required to do experimentation and testing with a sequence of computational steps such as analyzing the given matrix, applying a reordering or scaling transformation, selecting and applying a preconditional, and applying a solver among various alternatives. Each computation might be required to explicitly or transparently access remote numerical systems installed on specialized high-end computing resources. Note that the above examples are often called loosely coupled applications in following senses; first, no communication between activities/tasks (i.e., often called jobs in HPDC context) is required during the execution (i.e., message exchange between tasks occurs only during initialization and termination [102] of the tasks execution); second, there are no strict constraints on the component resources in terms of different programming 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. |
