Copyright Symbolics Inc.
October 1985
Cambridge, Massachusetts
Used with permission.

Symbolics manufactures high-performance symbol processing computers and associated software. Symbol processing (or symbolic processing) includes computation with symbols, relationships, and graphical objects as well as numbers, characters, and bits. Symbolic representations can more closely model a given problem as it exists in the real world. By making complex problems easier to represent, symbol processing languages enhance programmer productivity. The coupling of symbol processing software and hardware in the 1980s has considerably broadened the scope of problems that can be solved by computers.

The Symbolics 3600 family is a line of 36-bit single-user computers designed for high-productivity software development and for the execution of large symbolic programs. 3600-family processors give the user all the computational power associated with multi-user timesharing computers in a dedicated workstation. This is accomplished via a new and unique machine architecture that supports high-speed symbol processing operations directly in hardware. For example, every word in a Symbolics computer's virtual memory is tagged with data type bits - hence the name tagged architecture to describe 3600-family processors. The processor reads these bits to prevent illegal operations. As an added benefit, tag bits reduce the need for data type declarations in programs.

The Symbolics-Lisp system software constitutes a large-scale programming environment, with over a half-million lines of system code accessible to the user. Object-oriented programming techniques are used throughout the Symbolics-Lisp system to provide a reliable and extensible integrated environment without the usual division between an operating system and programming languages. All of the system software is written in Symbolics-Lisp. A high-performance implementation of Symbolics Prolog is also available, as are compilers for more conventional languages.

Outstanding documentation is a hallmark of Symbolics systems. The award-winning Document Examiner provides full access to Symbolics documentation in formatted form on the high-resolution screen of a 3600-family console. Within the Document Examiner, users can use the mouse to point at topic and object names. When they press a mouse button, the Document Examiner automatically calls up the documentation for those items.

Typical applications of Symbolics computers include the following areas:

The economics of computer hardware and the computational demands made by modern software have converged to make personal, networked computers more attractive than timeshared systems. Foreseeing this changing situation, researchers at the M.I.T. Artificial Intelligence Laboratory initiated the Lisp Machine project in 1974. The project was aimed at developing a state-of-the-art personal computer that would support programmers developing large and complex symbolic programs. An important decision was made early in the design process: for consistency throughout the software environment, all of the system code would be written in a single language - Lisp.

The Lisp Machine concepts rests on the following tenets:

• Dedicated personal computer and console
• Fast Lisp execution
• Tagged architecture (run-time data-type checking and generic instructions)
• Virtual memory
• Integrated local area network
• Interactive, high-resolution, bit-mapped graphics

In 1980, Symbolics, Inc., was formed with the purpose of combining past experience with the latest technology to develop a line of symbol processing computer systems and related products. Symbolics introduced the Symbolics LM-2 in 1981. The LM-2 was basically an M.I.T. CADR, repackaged for higher reliability and easier servicing. Symbolics made numerous improvements to the system software and offered options such as Fortran-77, color graphics hardware, and the LGP-1 laser graphics printer.

From 1979 to 1982, research continued on a much more powerful and cost-effective computer architecture. This system, known as the 3600, is based on a completely new design. With the 3600 family, Symbolics has replaced the microcode emulator architecture with special hardware to support high-speed symbolic and numeric computations and to expand virtual memory to 1 Gbyte. Ethernet hardware and software support is standard on all 3600-family computers. The console has been completely redesigned to offer a larger viewing screen with built-in 16-bit digital audio operation. To correspond to the new hardware, the software development team made major enhancements to the software system. In 1982 and 1983 the menu of 3600-family options expanded to include the Symbolics Pascal Tool Kit, the MACSYMA symbolic mathematics system, and a line of high-resolution color graphics products offered by the newly-formed Symbolics Graphics Division.

The Symbolics 3670 system was introduced in early 1984, with the 3640 following later in that same year. The 3670 is the most flexible member of the 3600 family, with a wide range of memory, disk, and add-on options. 3670s are designed as high-performance program development workstations or as networked file, database, or knowledge servers. The 3640 is a compact, high-performance implementation of the 3600 architecture. Also introduced in 1984, the Floating-Point Accelerator option increases the speed of IEEE-compatible floating-point multiplications by a factor of 500%, while speeding up integer divisions by 50%. Release 6 of the Symbolics-Lisp system software included the Ephemeral-Object Garbage Collector. This proprietary system reduces paging, and is the most efficient storage management system in symbolic computing. The Document Examiner included with Release 6 won an award from the Society of Technical Documentation. The Symbolics Graphics Division introduced S-Paint, S-Geometry, S-Render, and S-Dynamics as part of an integrated video-compatible image-making and image-processing facility.

All system software for Symbolics computers is written in a dialect of the Lisp language called Symbolics-Lisp. Lisp is a computer programming language that originated as a tool for Artificial Intelligence (AI) research. AI is a branch of computer science that seeks to understand and model intelligent behavior with the aid of computers. Lisp is designed to symbolically represent objects in the world and the relationships that exist among them.

"Lisp" stands for "List Processing Language" as it was dubbed by Professor John McCarthy of M.I.T. (now at Stanford University). At the lowest level all objects in Lisp are represented as lists, even the expressions in the Lisp language. This uniformity of representation has proven to be a tremendous advantage of Lisp over other languages. It makes Lisp easy to extend, which accounts for the longevity of the language and the tremendous range of applications written with it.

At this time, most major artificial intelligence systems are written in Lisp, including programs for expert problem-solving, common-sense reasoning, learning, natural language processing, education, speech, intelligent signal processing, and vision. Lisp has many features that make it useful for symbol processing.

• Lisp is easy to learn; the parenthesis notation makes Lisp syntax uniform.
• Lisp is interactive.
• Lisp functions and data have the same form; programs can generate other programs and then pass control to them.
• Lisp excels in representing arbitrary-sized objects in which the number of details or properties cannot be predicted (declared) in advance.
• The Lisp environment provides powerful editing and debugging tools.
• Lisp is extensible.

Symbolics-Lisp is a Lisp system developed specifically for Symbolics computers. All of the system code is processed by the Symbolics-Lisp compiler. Applications in symbolic mathematics (MACSYMA), document processing, and computer aided design have also been developed exclusively in Symbolics-Lisp.

Symbolics-Lisp evolved originally from the Maclisp dialect developed in the 1970s. It is somewhat compatible with Maclisp, while introducing many new features and improvements. These include:

• A full range of data types, including many numerical types, lists, strings, arrays, planes, and user-defined structures.
• Modern control constructs, including a very general loop iteration facility, asynchronous nonlocal exists, coroutines, and processes.
• Flexible function calling and multiple-value returns.
• Stream-oriented input and output.
• The Flavor system for object-oriented programming with message passing.
• Macros for extending the Symbolics-Lisp syntax.
• Predefined functions that support such operations as sorting, hash tables, linear equations, and matrix operations.
• Multiple namespaces (packages).

The Flavor System offers the following advantages:

• Object-oriented programming permits more modularity in programs by encapsulating procedures and data into a flavor object
• New flavor types can be cloned by combining existing flavor types
• Flavor inheritance is non-hierarchical

In the past two years, the Common Lisp standard has unified the Lisp community around a compatible dialect. Symbolics has been a prime supporter and an active participant in the Common Lisp movement. The committee to define Common Lisp included nine members of Symbolics' technical staff.

In keeping with the trend toward Common Lisp, the Zetalisp dialect of Lisp, in which much of Symbolics system software is written, has been modified to make it more compatible