Insight the FRIENDSHIP-Framework
The constant further development of the FRIENDSHIP-Framework has top priority at FRIENDSHIP SYSTEMS. We work on quality software solutions for practical, convenient and resource saving engineering which are inspired by the needs of practicioners. One of the major improvements currently under way is grid-based computation. Through this, massive time and cost saving in the design process is achievable.
What is the idea behind grid-based computing? Briefly, it means spreading the execution of computations over a number of processing units at a given time so as to parallely produce results. Practically, this means that you can use your whole IT infrastructure like your office workstations and your server machines for the parallel execution of computations steered from within the FRIENDSHIP-Framework. The application will enable you to use all your CPU cores for computationally intensive tasks. Grid-computing increases your productivity and produces faster results. Applied to the FRIENDSHIP-Framework, clustered computation calculates CFD optimization and other operations yet faster.
An example illustrates this: Given you wish to compute a SHIPFLOW or Fluent calculation of about 2,000 variants within the FRIENDSHIP-Framework. Say one computation takes 10 minutes. If you use your single desk computer only, you would get the results in about 14 days. Now, if you allocate resources, e.g. a rack with 80 kernels, distributed resource management software (DRM) may spread the computations among the kernel cluster. Reduced by factor 80, you receive the desired variants in 4 hours. Imagine the massive savings of time and engineering power! Chiefly, grid-based computing is applicable for you in three ways:
- use of kernel cluster
- use of office computer hardware
- use of computing capacity in single computer
All three uses would require an instance which distributes open tasks among the computing resources, i.e. grid resources (also see illustration below). This instance is a program or software for Distributed Resource Management (DRM). For the most part, it is installed on the main computer which manages the computation, the "gridmaster". Through DRM the gridmaster chooses open resources for grid computation. This happens through either specialized software like Condor or through Secure Shell (SSH) protocol which connects remote computers. The latter is frequently used for the distribution of tasks among office workstations. A DRM version of SSH is currently developed at FRIENDSHIP SYSTEMS. As this version is developed in line with the FRIENDSHIP-Framework, it is straightforward to use and spares you extensive administration. Hence, you save a massive invest in administrative power for the set up of DRM. Even without a kernel cluster or additional (external) workstations does grid-based computing significantly boost your work capacity. Why so? DRM can initiate computations on the different kernels of one computer. This means that you can use three of your say four kernels for parallel execution of tasks and leave the fourth free for other operations like the main application running. Without grid-based computation, you can execute merely one task at a time as you are not able to employ the other three kernels. Altogether, grid-based computing helps you to make most of your resources, most of all of your engineering capacity, i.e. manpower and wit.
Your benefits You receive excellent geometries in less time and cost to make the most of your advantages. Those you gain from applying the FRIENDSHIP-Framework are significantly reduced development time, first class geometry generation and optimal surface properties for better performance. Plus: grid-computation as provided with the FRIENDSHIP-Framework is singular in the market! Through this new feature you are able to bring your ideas to life and to market first. Headstarting means market relevant advance and a favorable position against your competitors. You are the first to come off the marks.
The next major upgrade of the FRIENDSHIP-Framework is planned to provide grid-based computing. It is scheduled for fall 2009.
For further information please contact Mr. Claus Abt.
Illustration of grid-based computing: the FRIENDSHIP-Framework addresses available grid resources (like number of processors, memory or other license restricted software (hardware bounded solvers)) through distributed resource management software (commercial, open source, inhouse programs) and distributes extensive computations among several resources parallely at a given time.

