The Linpack Benchmark

The LINPACK benchmark dates back to the early 1980s, when Jack Dongarra (then at Argonne National Laboratory) began collecting performance results of systems, based on their speed in solving a 100 x 100 linear system using Fortran routines from the LINPACK library. While a problem of this size is no longer a supercomputer-class exercise, it is still useful for assessing the computational performance of a single-processor system. In particular, it is a reasonable way to measure the performance...

Using PBS

From the user's perspective, a workload mangement system enables you to make more efficient use of your time by allowing you to specify the tasks you need run on the cluster. The system takes care of running these tasks and returning the results to you. If the cluster is full, then it holds your tasks and runs them when the resources are available. With PBS you create a batch job that you then submit to PBS. A batch job is a shell script containing the set of commands you want run on the...

Example of a Beowulf Node

The majority of Beowulfs (over five generations of systems in the past seven years) have employed microprocessors from Intel or AMD. This is because they have been among the least expensive systems to build, the system architectures are open providing a wealth of component choices, and the Linux operating system was first available on them. While not the fastest processors in peak performance, their overall capability has been good, and their price performance ratios are excellent. The most...

Hardware System Structure

The most visible and discussed aspects of cluster computing systems are their physical components and organization. These deliver the raw capabilities of the system, take up considerable room on the machine room floor, and yield their excellent price performance. The two principal subsystems of a Beowulf cluster are its constituent compute nodes and its interconnection network that integrates the nodes into a single system. These are discussed briefly below. The compute or processing nodes...