Synchrotron-based high-pressure research in materials science Page: 4 of 12
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to UNT Digital Library by the UNT Libraries Government Documents Department.
Extracted Text
The following text was automatically extracted from the image on this page using optical character recognition software:
" allow description of the machine configuration to be as independent as possible from of the
descriptions of the components;
" be genuinely portable across platforms ranging from single-processor workstations to clusters
of SMPs;
" be able to interface to other distinct applications such as direct execution simulators and
visualization systems.
These requirements suggest factoring the simulator into three parts: component descriptions, config-
uration descriptions, and an underlying, reasonably generic, and reasonably light-weight simulation
system with which all porting issues are associated. An object-oriented approach facilitates these
goals.
It is clear that simulating systems of the size and complexity that we envision will require the
use of parallel simulation [6]. Furthermore, the parallel simulation substrate must support compo-
sition of simulations and be very efficient in its implementation. We concluded that a conservative
synchronization scheme would have the best chance of success for this application. The require-
ment of portability across a variety of platforms led us to a parallel simulation substrate that runs
on both shared memory and distributed memory machines. The Scalable Simulation Framework
(SSF) [7] and the implementation of this framework being developed at Dartmouth College, DaSSF
[8, 9], is our current choice.
3 Initial Prototype
To determine the suitability of DaSSF as a parallel simulation substrate a small prototype model
was built. The purpose was to allow us to become familiar with DaSSF and to gain experience in
constructing models. The prototype was to be a learning experience and feasibility study rather
than the basis for conducting a specific simulation study.
We used the results of the domain analysis to define a subset of components to implement in
the prototype. We chose not to model the processors and memory hierarchy of an SMP node in any
detail and instead to focus on modeling the interconnection network between nodes as a fat-tree
network with a circuit-switched routing protocol. For the workload, rather than model the message
traffic from any specific application, we chose to have each processor node emit messages with
exponentially distributed interarrival times. The message destination node is selected with uniform
probability, and the message size is exponentially distributed. The parameters for the distributions
are inputs to the simulation.
3.1 Requirements
Our requirements for the prototype were that it exercise all the essential components of DaSSF
and several of the DaSSF extensions to SSF. To investigate the scalability of DaSSF itself the
components of the prototype were to be such that they could could be easily configured into
arbitrarily large models. Since we were not conducting a real performance study, we were not
concerned with modeling our system components with high fidelity, but rather determining whether
DaSSF was an appropriate substrate for developing components with arbitrary levels of fidelity. A
small model that used all the features was desirable for rapid development. At the same time we
were also interested in the scaling properties of DaSSF since we will want to develop very large
models in the future. The data collection capabilities provided by DaSSF were another topic to be
investigated.3
Upcoming Pages
Here’s what’s next.
Search Inside
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Synchrotron-based high-pressure research in materials science, article, Date Unknown; [Los Alamos, New Mexico]. (https://digital.library.unt.edu/ark:/67531/metadc927910/m1/4/: accessed April 23, 2024), University of North Texas Libraries, UNT Digital Library, https://digital.library.unt.edu; crediting UNT Libraries Government Documents Department.