Berkeley RISC

from Wikipedia, the free encyclopedia

The Berkeley RISC project was from 1980 to 1984 at the University of California, Berkeley, a state-run research project on behalf of the Defense Advanced Research Projects Agency (DARPA) under the direction of David A. Patterson for the development of RISC-based ( Reduced Instruction Set computer ) processor architectures . Commercial RISC microprocessor families such as Sun SPARC and AMD Am29000 subsequently developed from the project and its results . Later processor architectures such as DEC Alpha and the ARM architecture were strongly influenced by the results.

General

The idea of ​​the project was based on the observation that most machine programs only use a fraction of the hardware available in CISC architectures ( Complex Instruction Set Computer ). The restriction to a few and simple machine commands in the context of RISC enables a simpler and faster chip design with largely the same powerful software.

The first project line was called RISC I , also known as Gold , started in 1980 and was designed as a seminar on the subject of VLSI design . The first results were published in 1981. The RISC-I model processor consisted of 44,500 transistors and could process 31 machine instructions in 78 registers of 32 bits each. The control and control unit of the RISC I only took up around 6% of the chip area , while the CISC processors of the time required around 50% of the chip area for the control unit.

The second project line, RISC II , also called Blue , which started in parallel , ran more slowly until the end of the project. Due to the slower process, improvements over RISC I could be realized, including a pipeline and an expanded register set while reducing the chip size to 39,000 transistors. Furthermore, a new English instruction format expander was integrated, which made it possible to shorten certain machine commands in the external memory from 32 bits to 16 bits, with these machine commands being expanded internally to the full 32 bits. This made it possible to achieve a better code density in the memory. The ideas for this were later used, among other things, for the thumb instruction set of the ARM processors.

literature

  • National Research Council, Committee on Innovations in Computing (Ed.): Funding a Revolution: Government Support for Computing Research . National Academy Press, 1999, ISBN 978-0-309-06278-7 .

Individual evidence

  1. ^ Edwin D. Reilly: Milestones in Computer Science and Information Technology . Oryx Press, 2003, ISBN 978-1-57356-521-9 , pp. 50 .
  2. ↑ Carlo E. Sequin, David A. Patterson: Design and Implementation of RISC I. (PDF; 1.5 MB) University of Bristol, England, 1982, p. 25 , accessed on March 15, 2013 .