Jump to content

CDC 3000 series

From Alfawiki
Control Data Corporation
IndustryComputer hardware
Founded1957; 69 years ago (1957)
FounderWilliam Norris
FateDefunct (merged/acquired)
Headquarters
Minneapolis
,
United States
ProductsCDC 3000 series mainframe computers

The CDC 3000 series (often pronounced "thirty-six hundred" or "thirty-one hundred") represents a collection of mainframe computers developed by Control Data Corporation (CDC). The initial release, the CDC 3600, was a 48-bit architecture introduced in 1963. This model paved the way for the more compact CDC 3400 in 1964, followed by the 24-bit models: CDC 3300, 3200, and 3100, which debuted between 1964 and 1965. These systems succeeded the earlier CDC 1604 and CDC 924 computers.

Achieving significant commercial success, this series became CDC's primary revenue generator throughout the 1960s. Despite the faster performance of the pricier CDC 6000 series, the CDC 3000 line maintained higher sales volumes, although its comparative speed diminished over time. Production ceased in the early 1970s as CDC redirected efforts toward the more sophisticated 6000 series and later CDC Cyber models.

Specifications

Model Word size Release date Price Frequency Memory max MIPS Picture
CDC 3600 48-bit June 1963 $1,200,000 714 kHz (1.4μs cycle) 1.536 megabyte (262144 x 48 bits) 0.700 MIPS
CDC 3400 48-bit November 1964 $680,000 667 kHz (1.5μs cycle) 192 kilobytes (32767 x 48 bits) 0.700 MIPS
CDC 3300 24-bit December 1965 $220,000 800 kHz (1.25μs cycle) 768 kilobytes (262144 x 24 bits) 0.800 MIPS
CDC 3200 24-bit May 1964 $200,000 800 kHz (1.25μs cycle) 96 kilobytes (32767 x 24 bits) 0.800 MIPS
CDC 3100 24-bit February 1965 $120,000 571 kHz (1.75μs cycle) 96 kilobytes (32767 x 24 bits) 0.500 MIPS

Upper 3000 series

The upper tier of the 3000 series employed a word length of 48 bits. The earliest unit, the CDC 3600, was introduced in June of 1963, with subsequent models like the CDC 3400 and CDC 3800 arriving by December 1965. Targeted primarily at scientific computation tasks, these machines acted as successors to the CDC 1604 but were soon overshadowed by the more powerful and complex CDC 6000 series, notably the CDC 6600 launched in late 1964. Several advanced research facilities utilized upper-3000 models temporarily while awaiting delivery of their CDC 6600 systems, benefiting from their compatible assembly language environments.

Lower 3000 series

Utilizing a word size of 24 bits, the lower segment of the series was derived from the earlier CDC 924, which itself was a scaled-down, 24-bit variant of the original CDC 1604. The CDC 3200 debuted in May of 1964, followed by smaller versions: the CDC 3100 in February of 1965 and the CDC 3300 in December of that year. The last model in this range, the CDC 3500, introduced in March of 1967, featured integrated circuits replacing earlier discrete components. Both the CDC 3300 and CDC 3500 offered optional features such as relocation capabilities, floating-point arithmetic, and BDP (Business + Data Processing) instruction sets aimed at commercial and business computing sectors.

The 3150

The CDC 3150 was positioned as a batch processing system equipped with FORTRAN and COBOL compilers to serve business-oriented applications. Its console design emphasized functionality tailored to operational efficiency.

Instruction sets

The upper segment of the 3000 series predominantly utilized 24-bit instructions, typically stored two per word, alongside certain 48-bit instructions. In contrast, the lower 3000 series operated with a 24-bit subset derived from this instruction set, facilitating compatibility across all 3000 models. These architectures were influenced by earlier CDC machines such as the 1604 and 924, preserving some backward compatibility while differing in features like relocation and BDP instruction capabilities.

Memory

Every computer within the 3000 series incorporated magnetic-core memory. Specifically, the CDC 3500 shared its core memory modules with those used in the CDC 6000 and Cyber 70 series.

Architecture

The CPU architecture of the lower 3000 series is based on a 24-bit word size: both instructions and two operand registers—A and Q—are 24 bits wide. It includes four index registers labeled B0 through B3, each 15 bits in length, with B0 permanently set to zero. There is no dedicated status register. The system can directly address up to 32,768 words of core memory, each being 24 bits, commonly organized into two or three memory banks.

Instruction format consists of a 6-bit operation code, a single bit for indirect addressing, two bits selecting which index register to use, and a 15-bit address field. Arithmetic operations employ ones' complement representation, supporting both positive and negative zero values. The A and Q registers can be combined to form a 48-bit register for certain calculations; additionally, there is an E register also sized at 48 bits.

The CDC 3600 CPU reached speeds near one million instructions per second (1 MIPS), earning it classification as a supercomputer in the year 1965.

Seymour Cray was responsible for much of the architectural design of the 3000 series before transitioning his focus to the CDC 6000 line. Several key innovations foundational to the CDC 6600 supercomputer were initially introduced within the 3000 series.

Software

The upper-tier 3000 models operated using the SCOPE (Supervisory Control Of Program Execution) operating system. A variant known as Tape SCOPE functioned as a serial batch OS without buffering capabilities; Drum SCOPE enhanced throughput by adding print spooling functionality. CDC also developed SUMMIT (Simultaneous Usage of Multiprogramming, Multiprocessing with Interactive Timesharing) for the CDC 3800 to exploit its advanced hardware features; however, this OS was never officially released due to shifting priorities toward the CDC 6600 project.

Initially, lower-end 3000 systems ran RTS OS but quickly transitioned to MSOS (Mass Storage OS). The main operating system for the CDC 3300 and CDC 3500 was MASTER (Multi Access, Shared Time Executive Routine), which supported multitasking and was disk-based. MASTER was optimized for batch processing by incorporating card caching and print spooling and took advantage of page-mapping hardware to manage memory efficiently.

A real-time version of SCOPE was available for both lower and upper segments of the 3000 series, with a disk-based variant later developed specifically for upper-tier machines.

Supported programming languages encompassed FORTRAN, COBOL, ALGOL, and assembly language COMPASS. An APL environment was created for the upper 3000 series at the University of Massachusetts Amherst and subsequently distributed among users.

Character set

Operating systems within the CDC 3xxx family generally utilized six-bit character encoding, resulting in a 64-character set that excluded lowercase letters. This set consisted of (in display code sequence):

A-Z 0-9 + - * / ( ) $ = (space), . # [ ] % " _ ! & ' ? < > @ \ ^ ;

This six-bit character encoding extended four-bit BCD codes and is referred to as BCDIC (BCD interchange code). IBM later expanded this scheme into an eight-bit format known as "extended BCDIC," or EBCDIC.

Character representation varied depending on output devices like line printers and certain terminals:

Character Occasionally
rendered as
# ≡
" ≠
_ → or {
! ∨
& ∧
' ↑
? ↓ or }
@ ≤
\ ≥
^ ¬

It is noteworthy that control characters such as carriage return and line feed were not included in this character set; instead, these functions were managed through record structure encoding.

United States

From the mid-1960s until roughly 1980, Oregon State University operated a CDC 3300 running a proprietary operating system known as OS3 (Open Shop Operating System).

The Internal Revenue Service of the United States made extensive use of CDC 3000 series computers.

During the early 1970s, Northeastern University in Boston employed a CDC 3300 system.

Prior to its acquisition by United Computing Systems, International Timesharing Corporation (ITS) based in Chaska, Minnesota, provided timesharing services utilizing CDC 3300 machines.

California State University at Northridge implemented a dual CDC 3170 setup to support timesharing throughout the California State University network. Numerous campuses also operated CDC 3150 models primarily for batch processing tasks. By 1970, most California State College locations were equipped with CDC 3150 units; San Jose State and Los Angeles State functioned as regional data centers housing CDC 3300 systems. Typical configurations featured 24K words of 24-bit core memory, multiple tape drives, drum printers, card readers/punches, and removable-pack disk drives. Humboldt State College replaced IBM 1620 systems with these CDC models for both administrative and educational functions.

The US Air Force deployed several CDC 3800 series computers at the Air Force Satellite Control Facility (now known as Onizuka Air Force Station) to handle orbital planning and satellite defense system maintenance until IBM mainframes took over in the mid-1990s. These machines utilized the JOVIAL programming language to maintain computational precision.

From approximately 1968 to 1975, the Center for Naval Analyses operated a CDC 3800 to support scientific computing related to Navy operations research.

Michigan State University's Computer Science and Engineering Department maintained a CDC 3600 system.

CDC's South West Region clientele during the late 1960s and early 1970s encompassed:

Long Beach Memorial Hospital (for patient management and billing)
Engineering departments within California state universities (Fullerton, Los Angeles)
California State College Los Angeles serving as the southern regional data center
California State College San Jose functioning as the northern regional data center with a CDC 3300; Los Angeles State also hosting a southern regional CDC 3300 data center
Jet Propulsion Laboratory (for outer space module testing)
University of Texas at El Paso
Vandenberg Air Force Base missile launch control operations (using two CDC 3600s)
Thompson-Ramo-Woolridge in Redondo Beach
Engineering faculties at Cal Poly Pomona and Cal Poly Northridge
Point Mugu Naval Air Station (tracking weather balloons and analyzing data)
El Centro Naval Air Station
Camp Pendleton US Marine Corps – served as a development and test site for the Marine Tactical Aircraft Command and Control System (MTACCS), originally on a CDC 3300, later upgraded to a CDC 3500 around 1973-74
Beverly Hills Data Center – part of the CDC Cybernet network, notable for comprehensive banking computations for approximately twelve local banks; connected to the Los Angeles Data Center (CDC 6600). It replaced an IBM platform and included multiple CDC 854 disk drives alongside a Burroughs B-3000 magnetic ink character reader (MICR).

Europe

CERN operated a CDC 3800 from 1966 until January 1969 before transitioning to a CDC 6600. Subsequently, the CDC 3800 was acquired by the State of Geneva and installed at the University of Geneva.

Numerous CDC 3000 series machines were utilized throughout France in universities, software companies, and industries such as aerospace, nuclear energy, and petroleum.

The University of Oslo ran a CDC 3300 between 1967 and 1976.

In England, Computation Research & Development Ltd—a London branch of Freeman Fox & Partners—employed a CDC 3300 for engineering and commercial computing tasks until it was retired in 1974.

The German Meteorological Service (Deutscher Wetterdienst) used CDC models including the 3800 and 3400 during the late 1960s for numerical weather forecasting purposes.

Oceania

Australia was home to a significant number of CDC 3000 series computers. In 1964, the Commonwealth Scientific and Industrial Research Organisation (CSIRO) installed a CDC 3600 in Canberra, supported by satellite CDC 3200 units located in Melbourne, Sydney, and probably Brisbane and Adelaide. The Bureau of Census and Statistics operated several machines including the 3600, 3300, and 3500 models in Canberra, while CDC 3200s in various state capitals were upgraded to 3300s around 1970. Monash University acquired a CDC 3200 in 1964. Additionally, the Defence Signals Directorate maintained at least one CDC system. BHP utilized two CDC 3300 computers each at its Newcastle and Port Kembla data centers; these were upgraded to CDC 3500 models in 1977 and remained operational until the mid-1980s.

Communist bloc

Between 1969 and 1971, Hungary obtained a CDC 3300, which was used by the National Academy of Sciences throughout the 1970s for scientific calculations.

Other installations of the CDC 3300 within former Eastern Bloc countries included:

  • Computer Research Centre Bratislava, supported by the OSN Development program
  • Central Statistical Office in Prague
  • Romanian Aircraft factory in Bucharest, which operated a CDC 3500 system

Standard peripherals

  • 405 – High-speed punched card reader for 80-column cards
  • 415 – Punch device for 80-column cards
  • 501 – Rotating drum line printer with a width of 136 characters (limited to uppercase letters)
  • 505 – Line printer model
  • 512 – Chain-type line printer
  • 601, 604, 607, 609 – Magnetic tape drives (with the 609 model supporting nine-track tapes)
  • 700 series – Optical page readers
  • 800 series – Rotating magnetic memory devices
  • 808 – Rotating disk drive featuring approximately 32-inch disks and holding around 32 to 36 disks per spindle
  • 852 – Disk drive with removable media pack; capable of storing between two to three million seven-bit characters across six disks; data format compatible with IBM 1311
  • 853 – Similar to the 852 but with a capacity of four million characters; media mechanically interchangeable with IBM 1311 though data incompatible
  • 854 – Comparable to the 853 but offering eight million characters of storage capacity
  • 915 – Optical page reader

Many of these peripherals were also initially offered alongside the CDC 6000 series.

Images

See also

External links