On the Horizon: Fast Chips Quickly

Article Abstract:

Silicon compilers for conversion of a behavioral description to the the chip layout are now being used commercially, but are still far from implementing any VLSI chip in a matter of hours. The simplest silicon compilers use gate arrays or standard cells and are not capable of making design decisions about architecture and floor- planning. Today's silicon compilers start from a functional description, sometimes combined with structural information, and turn that description into an integrated circuit. The circuit is built-up of parameterized cells that are generated by small procedures. The quality of these cells is usually weak compared to those designed by humans. Some compilers attempt to use a type of expert system to generate high-level architectures, instead of leaving that work to the human designer. A big problem is the lack of standards for data format among compilers. There is disagreement today on what kind of description a program must accept and what kind of layout it must produce to be called a silicon compiler. Diagrams illustrate the silicon compilation process. Crucial technological terms are defined.

Integrated circuits, Processor architectures, Circuit design, Integrated circuit fabrication, Computer aided design, Expert systems, Gate arrays, CAD/CAM, Computer-Aided Design, Processor Architecture, Pipelining, Silicon Chips, Compiler/decompiler, Standard Cells

---

Designing the Next Generation

Article Abstract:

Advances in many fields are required for a new generation of computers to evolve that are faster, more intelligent, and more accessible than today's computers. Artificial intelligence breakthroughs such as expert systems, language understanding, and speech and picture recognition will be a key to future technology. Trees, square and cubic arrays, and data flow structures are possible parallel architectures that would provide the speed necessary for new applications. Such architectures require advances in VLSI circuit design and fabrication. Improving programmer productivity is another goal of future-generation researchers. Future computer networks and communications systems will be complex to tie all of the advanced systems together.

Technological forecasting, Software engineering, Industrial research, Artificial intelligence, Parallel processing, Research and Development, Future Technologies, Future of Computing

---

The Limits to Simulation

Article Abstract:

For some areas of simulation, no increase in computing power will improve simulations beyond a certain point. This is because the problems involve mathematically ill-conditioned behavior such as turbulence and nonlinearity. Turbulence causes problems in flow studies, plasma physics, and economic models. The concept of strange attractors has been helpful in mathematical studies. Besides economic modeling, weather forecasting and aerodynamics studies are also plagued by uncertainties. Strange attractor computations are shown in graphs and photographs show the growth of magnetic islands in a toroidal plasma.

Economics, Aerodynamics, Physics, Meteorology, Stability, Modeling, Data modeling software, Simulation, Limitations, Plasmas

Similar abstracts:

• Abstracts: Can I have sloping limits? The origin of specifications. When do I recalculate my limits?
• Abstracts: Motoring into the 21st Century. Bearing up to extreme conditions. Engine spins the barrel
• Abstracts: Hard on the outside, soft in the middle. VR comes out of the screen. Lubrication comes in from the cold
• Abstracts: The Professionals: Experts in Short Supply. Networks: Wiring a Nation. Militarization: Peace or War?
• Abstracts: Detroit Unveils Sophisticated Electronics. The Problems and Promises of High-Definition Television. How Data-Driven Automation Got its Start at IBM

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.