Scalability.
Saved in:
| Title: | Scalability. |
|---|---|
| Source: | Communications of the ACM; Aug1992, Vol. 35 Issue 8, p32-35, 4p |
| Subject Terms: | SYSTEMS design, SCALABILITY, MONTE Carlo method software, LOCAL area networks, MICROCOMPUTER workstations (Computers), THREADS (Computer programs) |
| Abstract: | The article discusses about scalability of computers that include size scalability, generation (time) scalability and problem scalability. Size scalability simply means that a very large computer such as the ultracomputer can be built. Typical definitions fails to recognize cost, efficiency, and whether such a large-scale computer is affordable in a reasonable timescale. A size scalable computer is designed from a small number of basic components, with no single bottleneck component, so the computer can be incrementally expanded over its designed scaling range, delivering linear incremental performance for a well-defined set of applications. A generation scalable computer can be implemented in a new technology, and thus take advantage of increased circuit and packaging technology. Problem scalability defines whether an application is feasible on a computer with given granularity. In practical terms, problem scalability means that a program can be made large enough to operate efficiently on a computer with a given granularity. Problems such as Monte Carlo simulation and "ray tracing" are "perfectly parallel" since their threads of computation almost never come together. Ordinary workstations provide some size and generation scalability but are LAN-limited. |
| Database: | Complementary Index |
Be the first to leave a comment!
Full Text Finder
Nájsť tento článok vo Web of Science