Lisp machine
A Lisp Machine (short often LispM ) is a computer whose processor is optimized for the execution of LISP programs and its operating system is written in LISP. In addition, written and typical applications in Lisp. Lisp machines offered a comfortable development environment. Some Lisp machines were also used without a development environment for the use of Lisp applications. Lisp machines were designed in the 1970s and 1980s in order to perform tasks in the field of artificial intelligence (AI ) better. Also for animation, some machines were used. The number of Lisp machines used was very low. There are estimates 5000-7000 pieces.
Despite this small number, some concepts of today's computer systems were tested on Lisp machines and first used productively, such as color graphics, window systems, computer mice, computer networks, hypertext, incremental compilation, and the concept of single-user workstations.
Due to the great advances in processor technology in the 1990s, which allowed also sufficiently fast general microprocessors, and the AI winter overthrew the Lisp machine manufacturers in a crisis and ended the production of Lisp machines.
Technical idea behind the Lisp Machine
For research projects on Artificial Intelligence Lisp programming language was used during the 1960s and 1970s, mostly. The available computers were optimized to handle programs in languages such as assembler or Fortran with as little memory usage and computation time. The execution of Lisp programs required, however, for that time considerable resources. One reason for this was the complexity of Lisp programs and the large amounts of data that have been processed in AI applications.
Lisp uses dynamic typing and dynamic memory management (garbage collection ). The Lisp machines often provide generic operations. The machine operation accepted at runtime any number types. The processor determines the types of the arguments, this checked for applicability, lead conversions, selects the appropriate addition and then applies them. For this purpose, the data words have been tagged with type information ( met ). The type checking could be done in parallel and was much faster than a software implementation. Typical word lengths for Lisp machines are 32 bit (eg TI Explorer Microprocessor ), 36 bit ( Symbolics 3600) or 40 bit ( Symbolics Ivory ). With a word length of 36 bits and data fit with 32-bit and using 4- bit in a data word.
In addition, virtual memory was introduced and the garbage collection supported by the hardware. In commercial Lisp machines and entire Lisp functions have been implemented in hardware.
In order to provide users with as much computing power, Lisp machines were designed as single-user computer ( with support for bitmap screens, keyboard, mouse, network interface, hard drives, tape drives, and a variety of expansion slots). This was unusual for that time, were used in the mainframe through terminals as a multiuser system. In order to allow collaborative work among multiple users, Lisp machines were designed with the ability to form computer networks ( first Chaosnet, and later Ethernet), which was also unusual for the time.
In the area of hypertext Lisp machines were then leader. The Lisp Machine manufacturer Symbolics documentation system won several awards.
On Lisp machines not only run Lisp programs. There are also compilers for example, C, Pascal, Fortran, Ada, and Prolog. These compilers are usually written in Lisp and may also be used interactively as Lisp compiler.
History
Beginnings at the MIT AI Lab and Xerox
1973 began Richard Greenblatt and Thomas Knight with the development of a prototype for a machine that should run optimized Lisp code. The first machine on the Knight wrote his master's thesis was, CONS machine ( after the cons Lisp function ) called and completed in 1976. The CONS machine had a 24-bit architecture and still needed a PDP-10 for operation. After a presentation of the machine in 1978 at a conference on AI DARPA began to finance the project, and companies have expressed their interest in acquiring a Lisp Machine. This led to the development of the CADR machine ( according to the Lisp function cadr ), were produced by the 25 pieces. The great interest in Lisp machines led to the founding of a company has been planned for commercialization.
Parallel to the development at MIT developed BBN Technologies own Lisp Machine ( Jericho ), but this was never commercialized. The disappointing team was then recruited by Xerox and developed at Xerox PARC in 1979 a Lisp machine named Dolphin. The Xerox Lisp machines based on Interlisp, in contrast to the MIT machines based on MacLisp.
Commercialization: Symbolics Inc., Lisp Machines Inc.
In 1979, an argument arose between Russell Noftsker and Greenblatt about the business model of the company. Noftsker wanted to build a traditional company, while Greenblatt especially strove with one of the hacker ethic of the MIT AI Lab, to be agreed business model that should refrain from risk capital. Since Noftsker, who had left the AI Lab in 1973 to work in the private sector, already had experience in the commercial world and it also came from other reasons to dispute between Greenblatt and several employees at the AI Lab, managed Noftsker many employees, among others, Thomas Knight to win for his plans and he founded Symbolics Inc. Greenblatt initially remained passive and was very upset about Noftsker. Control Data Corporation ( CDC) have shown great interest to acquire an MIT CADR machine, and Alexander Jacobson, a consultant from CDC, therefore Greenblatt drove them to finally start his own company. Greenblatt then founded Lisp Machines Inc. ( LMI short ). 1980/1981 brought the Symbolics LM -2 on the market, which was a repackaged MIT CADR machine. LMI also brought out an MIT CADR machine, the LMI CADR machine.
The competition between LMI and Symbolics meant that the employees of both companies had to leave the AI Lab. Only Richard Stallman and Marvin Minsky remained. In addition, LMI and Symbolics had their technology and software has been licensed from MIT and MIT for granted a right to use their changes, but Symbolics refused to allow the WITH the changes in the original prototype and the software to integrate, so that LMI was no goal. This angered Stallman, who thereby became the advocate of free software. Stallman used the access at MIT Lisp machines, to reconstruct the changes and provide LMI available. However, LMI kept well sealed their own changes to the Lisp Machine.
LMI licensed their Lisp machines to the company Texas Instruments, which produced I / II based on the LMI Lambda machines with Explorer.
End of the Lisp Machine
In the late 1980s and early 1990s broke the already small market Lisp machines together. LMI was already insolvent in 1986 and an attempt to revive the company as GigaMos system failed because of legal problems of the investor. Xerox had relatively early stop the development of additional Lisp machines.
Reasons for the collapse are many. On the one hand, the market was very small. Speculations are based on 5000-7000 machines. This ensured that the producers could invest less money in the technical development of the Lisp machines, while the manufacturers of conventional computer developed better and better procedures and caught up with Lisp machines soon to speed and even outdated. Companies like Lucid Inc. and Franz Inc. began Lispumgebungen for Micro Computer for sale. The porting of the Symbolics operating system software generation from 1992 to a DEC Tru64 UNIX / Alpha system was three times as fast as the fastest Lisp Machine.
In addition, the exaggerated expectations unfulfilled at the Artificial Intelligence, which is why the funds for many AI research projects have been cut ( the so-called AI winter). In particular, the massive reduction in funding for the SDI project (also called Star Wars project) hit the market hard. Many AI research projects ( particularly in the area of expert systems ) had been funded through SDI agent. Thus, the most important market for Lisp machines broke.
Manufacturer of Lisp machines
- Symbolics
- Lisp Machines Inc. ( LMI short )
- Integrated Inference Machines ( short IIM )
- Texas Instruments
- Fujitsu
- Xerox