- What is a Compute Cluster?
- Different Types of Compute Jobs
- Building a Compute Cluster
- Computing Resources Needed
- Price Per CPU
- Optimal Solution Economics
- Beowulf Solution
- Beowulf Cluster on SPARC Hardware
- SUN Supported Beowulf Cluster
- How To Build Your Compute Cluster
- Advantages of a Sun Based Cluster
- Grid Computing
- Conclusion
- Compute Cluster Software
Computing Resources Needed
Different organizations require different computing resources; no optimal solution fits all. Some commercial applications, especially in the CFD and crash simulation areas are available in parallel versions (both multithreaded and MPI). If a parallel version of the application is available and the input data is such that the performance scales well, you can improve performance significantly by running the job in parallel. A CFD simulation runs seven times faster on eight CPUs compared to one CPU. Thus, the results are available overnight instead of in two days. In most organizations the majority of the jobs are single threaded. Some applications scale well to a modest number of CPUs; a few applications scale to a large number of CPUs. The exact numbers differ among organizations, but a similar distribution is often seen. FIGURE 1 shows a typical workload distribution.
FIGURE 1 Typical Workload Distribution GraphÑNo. of CPUs Per Job Versus No. of Jobs