Critical Problems in Very Large Scale Computer Systems
Abstract
The research vehicle for this contract is the largest p ossible computer that could be conceived for the mid to late 1990s. The technical challenges of such a machine serve as the guiding stimulus for the research carried out and reported here. We imagine this machine to occupy a 14-story building, to cost upwards of $1,000,000,000 and to be so colossal that the nation could only afford one or two of them. The available chip technology and machine size are consistent with a million billion FLOPS (that's 10 to 15th) and a million billion Bytes of memory. It will dissipate 50 megawatts of power using CMOS technology. Communication across the machine will be much slower than computation at a node. The architecture, software, interconnect technology, packaging, and operating system are unknown. This investigation deals with hardware technology, software techniques, programming algorithms, communications, processing elements, and applications. The study will determine the plausibility (not feasibility) of such a machine. Progress in these various areas are highlighted in the individual sections below. Keywords: Circuits, Systems software, Communications topology, Routing algorithms.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 31, 1989
- Accession Number
- ADA208070
Entities
People
- Anant Agarwal
- Bill Dally
- F. Thomson Leiserson
- Jacob K. White
- John L. Wyatt Jr.
- Paul Penfield Jr.
- Srinivas Devadas
- Thomas F. Knight Jr.
Organizations
- Massachusetts Institute of Technology