  |
|
|
|
The last two articles dealt with the clustering of workstations for the purpose of constructing a kind of supercomputer. The latter article described the effort in this arena being conducted jointly by Sun Microsystems and the University of Buffalo. If you happened to check out the UB on-line article which was referenced here, you might have noticed a portion of the piece devoted to the choice of the Linux operating system for the project. This is an important issue, not only for workstation clustering, but for scientific computing in general.
In considering which operating system to use, the scientist must have a handle on the problems he's examining, the software he intends to use to do research, and how much detailed control he needs to have in order for his project to be successfully completed. In general, researchers with a long history in computation or who come from a "hardcore" computing environment want to control what they're doing down to the finest detail. These people often need an operating system which provides just the basic tools needed for file manipulation. Most "bells and whistles" will either not be used at all or will actually be created by the scientists themselves. On the other hand, researchers with more of an experimental background often find that they prefer to have things laid out nicely before them, with a very friendly computing environment. In this case, a more interactive operating system is probably preferred, and any drop-off in performance will be more than compensated for in ease of use. In the coming weeks, we'll see how the more widely used operating systems fit into these niches, starting with the ubiquitous Microsoft Windows.
Go To Page: 1
The copyright of the article Operating Systems -- Looking out the Windows in Scientific Computing is owned by . Permission to republish Operating Systems -- Looking out the Windows in print or online must be granted by the author in writing.
|
|
|
|
