The Ultimate Computer,
Abstract
Whoever builds a computer with ultimate capability must push hardware to the bounds of its physical limitations, find better mathematical approaches, and employ the most efficient program implementation. Multistream machines or large memories could markedly improve speed, but for real gains, one must also look to a more sophisticated exploitation of components. To this end, analysis shows that component-speed improvements by factors of several hundred to a few thousand can be expected before physical limitations are encountered. Changes in machine architecture along the lines of parallel operation--together with the improved components--could lead one to anticipate computers at least 1,000-fold faster than any now operating. Whether one should build such computers depends on the priorities that are set. The possibilities are described in the report.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1972
- Accession Number
- AD0748924
Entities
People
- Willis H. Ware
Organizations
- RAND Corporation