Jump to content

IMLAC

From Alfawiki
Adapted from the English Wikipedia article IMLAC, available under CC BY-SA 4.0Updated 25 July 2025
IMLAC Corporation
IndustryElectronics
Founded1968; 58 years ago (1968) in Needham, Massachusetts
DefunctTemplate:End date and age
FateAcquired by Hazeltine Corporation
Products

IMLAC Corporation was a U.S.-based electronics firm located in Needham, Massachusetts, recognized for its graphical display technologies, particularly the PDS-1 and PDS-4 models, which were prominent during the late 1960s and throughout the 1970s.

The PDS-1 debuted at the Fall Joint Computer Conference in late 1969 as the first cost-effective commercial interactive graphics terminal capable of motion display. Initially priced at $9,450 per unit, bulk purchases reduced the cost to $6,545 each. Offering similar functionality to the significantly larger and more expensive IBM 2250, the PDS-1 represented a major step forward toward contemporary computer workstation designs.

This system comprised a CRT display, keyboard, light pen, and a control panel with all electronics integrated into a desk pedestal. It housed a straightforward 16-bit minicomputer, featured 8 to 16 KB of magnetic-core memory, and included a dedicated display processor to manage CRT beam positioning. By 1971, an optional mouse device was introduced.

The company’s name, IMLAC, is derived from a character in Samuel Johnson’s novel The History of Rasselas, Prince of Abissinia.

Timeline of products

  • 1968: Imlac was established with an initial focus on interactive graphics terminals aimed at stock traders; this market strategy did not come to fruition.
  • 1969: The PDS-1 was launched targeting the broader graphics market; initial deliveries occurred late that year.
  • 1970: Introduction of the PDS-2, which integrated the PDS-1’s circuitry directly into the desktop display unit.
  • 1972: Release of the PDS-1D featuring enhanced circuitry and an improved backplane.
  • 1973: Launch of the PDS-4 model, which doubled processing speed and display capacity while offering flicker-free graphics; it also provided an optional floating-point arithmetic module.
  • 1974: The PDS-1G model was introduced.
  • 1977: Approximately 700 units of the PDS-4 were sold within the United States; these systems were custom-built rather than mass-produced.
  • 1978: Debut of Dynagraphic models 3205, 6210, and 6220 designed primarily for use with proprietary Fortran-based graphics libraries on larger host computers without requiring user programming within terminals.
  • 1979: Imlac was acquired by Hazeltine Corporation, known for manufacturing text-only terminals.
  • 1981: Hazeltine released the Imlac Dynagraphic Series II, compliant with SIGGRAPH’s CORE 1979 standard for 3D graphics libraries. Priced at $9,000 OEM, it featured a resolution of 2K x 2K pixels, included 192 KB of RAM, and incorporated an Intel 8086 microprocessor embedded within the monitor enclosure.

The DEC GT40 terminal offered similar design principles and pricing comparable to the PDS-1D but utilized a compact PDP 11/05 board set as its local minicomputer. This provided access to extensive software development tools while typically operating alongside larger PDP host systems.

Refreshed vector display

The PDS-1 utilized a 14-inch monochrome vector display refreshed from local memory with an addressable resolution of 1024 by 1024 points; this could be extended to a maximum of 2K x 2K in modes using small fonts. The CRT’s electron beam was programmatically directed along both X and Y axes to render individual sloped lines and characters akin to pen plotters. This method allowed skipping blank screen areas and enabled drawing in any sequence.

Vector displays—now considered obsolete—differ fundamentally from raster scan displays by rendering only visible lines and curves instead of scanning the entire screen as a pixel grid. This technique required substantially less memory; for instance, black-and-white raster graphics at XGA-level resolution (1024x768) demanded approximately 96 KB of video RAM—about twelve times more than needed by the PDS-1—making vector displays economically advantageous given the high cost of memory at that time.

These displays were well-suited for applications such as data visualization charts, computer-aided design (CAD) diagrams, three-dimensional wireframe graphics, text editing, page layout tasks, and simple gaming. However, they lacked capabilities like color rendering, filled shapes, or WYSIWYG font accuracy necessary for professional publishing environments.

To prevent flicker, the PDS-1’s screen refreshed at a rate of about forty times per second. However, because beam deflection relied on electromagnetic coils rather than electronic scanning techniques used in raster displays, refresh rates were limited. Complex images containing over eight hundred inches of lines or more than twelve hundred characters would produce noticeable flicker due to retrace times exceeding refresh intervals.

Competing terminals such as Tektronix’s lower-cost Model 4010 employed storage tube CRTs that retained images without continuous refreshing or local memory. While this approach eliminated flicker and reduced memory requirements, it lacked interactivity and animation capabilities since complete screen redraws were necessary after any erasure.

Unlike fixed-character displays common at that time (for example, CDC 6600 consoles utilizing Charactron CRTs with stencil masks), the PDS-1 generated all character shapes, sizes, and spacing purely through software control. Each glyph was drawn by executing a dedicated vector stroke subroutine allowing arbitrary font designs, expanded character sets, and support for complex scripts including Arabic. Smaller fonts sacrificed visual elegance for faster rendering by approximating shapes with diamond-like strokes; this system also accommodated electronic schematic symbols.

The monitor could be configured in either portrait or landscape orientation with a rectangular screen face. The nominal grid of one thousand by one thousand points was stretched by approximately one-third along its longer axis to maximize screen utilization for text and graphical content. This non-square pixel arrangement required software compensation. For applications focused on graphics accuracy rather than text layout, an unstretched grid option was available but resulted in unused screen margins.

Dual processors

The PDS-1 featured a display processor and a minicomputer that operated concurrently, sharing the same memory space. The display processor handled 1-byte commands for characters and curves, 6-byte instructions for vectors, and 2-byte unconditional jump commands. It did not include traditional ALU operations and never altered memory contents directly. Jump instructions facilitated subroutine calls for recurring symbols and organized objects into linked lists to streamline editing tasks. Positioning was restricted to integer coordinates without support for real-time rotation or scaling transformations. When symbols extended beyond screen boundaries, smearing occurred, necessitating higher-level clipping managed by separate data structures—a shortcoming resolved in subsequent models. Assembly language was used to program letter font subroutines, while line drawings and layouts were dynamically produced by programs running either on the local minicomputer or on remote systems.

The integrated minicomputer played a crucial role in ensuring responsive user interaction without depending on remote time-sharing systems. Its main function was constructing and refreshing the display list each cycle. To reduce costs, Imlac designed a custom minicomputer with a single-accumulator architecture reminiscent of the DEC PDP-8 but featuring 16-bit instructions and data width. It lacked hardware multiply/divide operations, floating-point support, microprogramming capabilities, virtual memory addressing, cache memory, and stack functionality. Addressing modes were confined to indirect pointers with auto-increment capability. Programming was primarily done in assembly language; eventually, a self-hosted Fortran compiler was introduced despite slow compilation times due to limited memory resources. Certain models included optional IBM 2310 cartridge or 8-inch floppy disk drives with basic disk operating system support; however, these options were removed in later revisions.

The PDS-1’s electronic components utilized low-density 7400 series TTL integrated circuits mounted on small printed circuit boards containing up to twelve chips each. The system architecture comprised three racks housing 25 cards connected via a wire-wrap backplane without a standardized bus structure. Comprehensive customer documentation provided gate-level schematics, allowing users to design custom interfaces and gain full system insight. Core memory cycle times were 2.0 microseconds (reduced to 1.8 microseconds in the PDS-1D), while TTL logic operated at speeds approximately ten times faster.

The base PDS-1 did not include hardware dedicated to rendering long vectors; instead, the minicomputer generated extended sequences of short-stroke instructions calculated using Bresenham's line algorithm, efficiently determining points without multiplication or division operations. The optional long vector hardware required only add/subtract circuits. Running long vector programs on systems lacking this hardware could produce display processor errors that might damage the monitor phosphor or deflection amplifiers.

Applications

Limited quantities of the PDS-1 and PDS-4 were purchased by research institutions and universities, serving as pioneering platforms for computer applications and training future graphics system developers. The FRESS hypertext system enhanced user experience on the PDS-1 by enabling hyperlink creation via light pen input alongside multi-window editing capabilities. Additionally, PDS-1 units contributed to the development of Arpanet’s network graphics protocol.

In 1973, the first overlapping multi-window environment—the DNLS (Display NLS)—was fully realized on the PDS-1 by SRI. DNLS introduced "Display Areas" (DAs), which could be interactively moved, resized, and overlapped; however, these areas only stored text strings. DNLS operated across multiple PDS-1 machines connected through ARPANET sites and served as an inspiration for the overlapping window system later implemented in the Xerox Alto.

Imlac display systems found integration within commercial products focused on visual design and specialized software solutions. Imlac marketed CES, a newspaper layout and typesetting application running on the PDS-1 platform. MCS's Anvil mechanical CAD system utilized later Imlac workstations for interactive design of mechanical components that were subsequently machined automatically from metal stock.

Certain applications—such as text editors—were developed entirely in Imlac assembler and operated standalone. Notably, Hofstadter authored Gödel, Escher, Bach using an Imlac-based editor. Nonetheless, most graphics programs demanded robust floating-point capabilities, compilers, and file management systems; these typically ran on costly timeshared computers that transmitted rendered image data to Imlac terminals emulating generic graphics displays. Common uses included producing architectural renderings and offline-generated animated walkthroughs. The absence of standardized program libraries for animation or interactive drawing limited widespread adoption of the PDS-1 for several years.

Despite these constraints, students frequently engaged in extensive assembler programming during their free time as a hobby. Among the most prominent applications developed on the PDS-1 were early interactive games. The two-player Spacewar! was ported from a PDP-1 demonstration version. Freeway Crossing, an early antecedent of Frogger, originated on a PDS-1 during a 1971 psychology study. Mazewar, recognized as the first online multiplayer computer game, ran on pairs of PDS-1 units; subsequent versions supported up to eight players networked through an MIT-hosted PDP-10 that managed game AI. Mazewar matches between MIT and Stanford generated significant traffic over the early Arpanet.

Pixels supplant vector screens

The continuous exponential advancement in computer memory—encompassing density, capacity, and affordability—commonly referred to as Moore's Law, led to the obsolescence of vector displays as raster graphic screens became financially accessible. While vector displays initially prevailed due to their lower expense, the significant reduction in memory costs eventually caused raster displays to take precedence.

At Xerox PARC, the Imlac PDS-1 was notable for its interactive graphics capabilities; however, its subpar text rendering motivated Chuck Thacker to create the bitmapped Xerox Alto in 1973, which became a foundational innovation in graphical user interfaces well before cost-effective memory for personal computers was available.

During the 1980s, vector terminals such as the PDS-1 were supplanted by non-programmable raster terminals like the AED767 and programmable raster workstations including the Terak 8510/a and PERQ Unix platforms. These systems eventually gave way to microprocessor-driven mass-market devices such as the Macintosh, Windows PCs, video game consoles, and ultimately integrated circuits within smartphones.

Emulation

In 2013, an emulator for the Imlac system called sImlac was created. A modern iteration can be found on the GitHub page of Seattle’s Living Computers: Museum + Labs, ensuring preservation of this groundbreaking technology’s heritage.

See also

External links