 |
|
|
|
|
| FAQ |
|
|
General Questions
BOINC
General Questions
- What is The Lattice Project?
The Lattice Project is an effort by a community of scientists at the University of Maryland to develop and deploy a comprehensive Grid system for scientific analysis. See the overview for more information.
- When will The Lattice Project be completed?
As an ongoing research project, The Lattice Project will in many regards never be completed. However, in many ways its current state might be described as a production Grid system in which features are continually being added, tested, and improved. You can track the progress of services that have been added to the computational Grid by viewing the Grid Services page. In addition, any news and announcements will be posted as soon as they are available.
- Is there planned public participation?
People who want to contribute their computing resources can download a BOINC client that will retrieve work and run analyses on their PC. Like the traditional SETI@home project, we will maintain the infrastructure for forming teams, receiving credit for work done, and provide other community building features. For more information on The Lattice BOINC Project, visit our project web site, http://boinc.umiacs.umd.edu.
- What is the main goal of The Lattice Project?
Scientific research can be greatly enhanced with access to high performance computing resources. The computational environment in which many scientists work is substantively deficient, materially impeding progress in a number of fields. It is our goal to make sufficient computational and data resources available to an active community of scientific researchers.
- I have participated in other Distributed Computing projects, and I am not familiar with some of the technologies used here. What is Globus?
Globus is a toolkit for developing computational Grids. More information can be found on their web site.
- My organization would like to get involved in The Lattice Project. How may we do so?
Involvement could take many forms. If you're interested in helping develop The Lattice Project, integrating computational or data resources into the Grid, or contributing software to run on the Grid, please contact a member of the staff.
- I am confused about the types of jobs that will run on The Lattice Project. Is there one well defined scientific problem it is addressing?
The simple answer is no. The Lattice Project provides the framework for running many types of scientific applications as Grid services. This is also true for jobs that get sent to the BOINC pool. The Lattice "Project" that runs over BOINC will be comprised of multiple applications.
BOINC
- How is BOINC used in The Lattice Project?
The Berkeley Open Infrastructure for Network Computing, BOINC, is Grid middleware for uniting individual computers, typically desktops, into a Grid system. Recognizing the potential of desktop Grid systems, we use BOINC to vastly increase our base of computational resources. A job submitted to the Grid could get scheduled to run on a BOINC pool just as it could get scheduled to a more traditional resource, such as a University cluster.
- What kinds of jobs will the BOINC client run?
For a complete listing of our Grid services and their compatibility with BOINC, look here.
- The Lattice Project seems to differ from other BOINC based distributed computing projects. How does it compare to SETI@home?
The use of BOINC in The Lattice Project differs from that in other projects in several ways. Most BOINC projects run a single application or class of applications, where each workunit is similar in the type of analysis performed, but differs in the specific data analyzed. In this way SETI@home is a large, well defined project. The Lattice Project is really a large Grid computing system. The Lattice Project is heterogeneous in the types of scientific analyses performed, and the overarching scientific projects for which the analyses are a part. It is also heterogeneous in the computing models and computing resources used. As mentioned elsewhere, in The Lattice Project a BOINC pool (a collection of BOINC clients interacting with a BOINC server) is one of many resources to which specific analyses may be assigned. A BOINC pool takes its place in the Lattice Project Grid system with dedicated clusters, SMP (Symmetric Multiprocessor) computers, desktop machines running Condor, and other computing resources.
Given these characteristics of The Lattice Project and its use of BOINC, people participating through BOINC need to keep these differences in mind. Some specific characteristics to note are as follows.
- There will not always be workunits available.
- The type of workunits will vary greatly in characteristics such as type of analysis being performed and time to completion.
- Some work units will take a very long time (e.g., several days or more) to complete.
- Estimated time to completion, if given, is likely to be highly variable.
- Checkpointing is not available for some applications.
- Some workunits may be removed from the system for one reason or another. We list such workunits here.
In all these ways BOINC in The Lattice Project differs from SETI@home and other projects. Consequently, expectations based on experience with BOINC from SETI@home and other projects will not always be in keeping with actual experience participating in the Lattice Project. This is one of the challenges we face as a community of research scientists and people generously working together to extend public-participation Grid computing to a broader range of projects.
- Are the projects that you will be hosting all non-profit?
Yes, all of the analyses run on The Lattice Project are parts of academic research projects. Most of the research projects are funded by public agencies (e.g., in the United States by the National Science Foundation, National Aeronautics and Space Administration, National Institutes of Health), and institutional funds (e.g., University of Maryland). See the brief description of research projects using The Lattice Project for computation. The results of these studies will be be published.
|
|
 |
