AN/FSQ-32
The AN/FSQ-32 was a computer made by IBM (International Business Machines) in 1960 and 1961 for the United States Air Force Strategic Air Command (SAC). IBM called it the 4020 Military Computer, but it was more commonly known as the Q-32. Only one unit was ever built.
Contents |
[edit] History
The Q-32 was installed at System Development Corporation (SDC) headquarters in Santa Monica, California. It was used as a development machine for the compiler and operational software for the AN/FSQ-31V, which was the Data Processing Element of the SAC Automated Command and Control System.
[edit] Near-death of the Q-32
The software for the Q-32 was developed by System Development Corporation (SDC) of Santa Monica, California. The original contract called for the Q-32 to be programmed as a solid-state replacement for the vacuum-tube AN/FSQ-7 computers in the USAF's Semi-Automatic Ground Environment (SAGE) system. However, the Q-32 was never used for SAGE, making the Q-32 "surplus." Moving the huge monolithic Q-32 was deemed economically infeasible, and the giant computer was declared to be a surplus item and was nearly junked.
[edit] New life for the Q-32
At that point in the history of computers, available technology permitted even the largest of computers to handle just one job at a time, known as batch processing. The entire mainframe, occupying perhaps an entire floor of a large office building and requiring massive refrigeration units to keep its electronics cooled, was consumed with each single job. No matter how large or how fast the computer, only one job could run at a time. That inefficiency resulted in the Q-32 and all other large mainframes having huge amounts of wasted idle time. Further, the single processor nature of the Q-32 could not assign priorities based on military variable urgency requirements.
[edit] Q-32 becomes research machine
In the early 1960s, the Q-32 was taken over by the Department of Defense's Advanced Research Projects Agency (ARPA) for further research purposes. The proposal became reality, and the Q-32 remained in Santa Monica for several years to come. System Development Corporation (SDC) was retained by ARPA to spearhead a large research project whose objectives included ways to permit the Q-32 to handle multiple batch tasks simultaneously. The methodology was dubbed "time-sharing," although it is a matter of dispute as to who coined the term, and when it was first used. Another research objective was to design and implement ways to allow simultaneous online processing access to the Q-32 from multiple geographically-separated computer users.[1]
On April 23, 1963, J.C.R. Licklider, PhD, a scientist who had recently moved from MIT to ARPA as Director of Behavioral Sciences Command & Control Research, wrote a memo which he titled "Memorandum For Members and Affiliates of the Intergalactic Computer Network."[2] In it he explores the early challenges presented in trying to establish a time-sharing network of computers with the software of the era. An early remote user of the Q-32 was the Augmentation Research Center at the Stanford Research Institute.[3] Ultimately, this work led to ARPANet, the precursor of today's Internet. However, the lack of memory management in the Q-32 was a limitation recognized as early as 1963.[2]
[edit] Architecture
The system was divided into functional sections:
- Central Processing Unit
- Memory
- High-Speed Input/Output
- Low-Speed Input/Output
- Operations Console
[edit] Central Processing Unit
Memory was addressed by words, which were 48 bits long. Each word was divided into eight 6-bit bytes. A 6-bit byte, as opposed to the 8-bit byte in common use today, was common in IBM and other scientific computers of the time. The address space provided a maximum of 256K words.
The Instruction set was rather complicated for its time. The instructions were a fixed length of one word providing 24 bits for the operation and 24 bits for the address. The address consisted of 18 bits (3 bytes) for the memory address, with other bits used for the specification of index registers and indirect addressing.
The operation field provided the operation code and a variety of modifiers. Some modifiers allowed instructions to operate only on specific bytes of a word or on specific bits of a byte without separate masking operations. Other modifiers allowed the single 48-bit ALU to operate on a pair of 24-bit operands to facilitate vector operations.
Other parts of the CPU were some sense switches, which could be used to control various software functions, the run/halt switch, and a switch, amplifier, and speaker assembly, which could be used to provide audio feedback or even play music, by connecting one of four bits in the main accumulator which could then be toggled under software control at an appropriate rate to produce whatever tones one wanted.
[edit] Memory
The Q-32 was equipped with 128k words (48 bits plus two parity bits) of memory that was oil and water cooled. Also considered as part of the memory subsystem in that they were addressed via fixed reserved memory addresses, were 4 48 position switch banks, in which a short program could be inserted, and a plug panel, similar to the one used in IBM Unit-Record equipment, that had the capacity of 32 words, so longer bootstrap or diagnostic programs could be installed in plug panels which could then be inserted into the receptacle and used. This served as a primitive ROM. The memory had a cycle time of 2.5 microseconds.[4]
[edit] High-Speed Input/Output
The High-Speed I/O section provided interfaces to the Drum Memory system, which consisted of a control system, and two vertical drum memory devices. Each drum read and wrote 50 bits at a time in parallel so transferring data could be done quickly. The drums were organized as 17 fields with 8192 words per field for a total capacity of 139264 words. The motors that rotated the drums required 208 VAC at 45 Hz so a motor generator unit was required to change the frequency from 60 Hz. This added to the noise level in the computer room. The other connection to/from the HSIO was to the SACCS EDTCC, which then interfaced to the rest of the SACCS.
[edit] Low-Speed Input/Output
The Low-Speed I/O section interfaced to several different devices:
- Communications Multiplexor
- Tape Controllers 1 and 2, connected to 16 IBM 729-V Tape Drives
- Disk File Controller, which was a modified Tape Controller, connected to
- Bryant Disk File, which had 25 disks that were 39" in diameter, 125 read/write heads that were hydraulically actuated, and had a total capacity of 26 megabytes
- IBM 1401, which controlled data transfers from unit-record equipment:
- 2 IBM Selectric Typewriters, (I/O Typewriters) one of which was used for operational messages and the other for diagnostic messages and maintenance activities.
SDC developed the system software using JOVIAL (Jules Own Version of the International Algebraic Language), one of the first high-level complied programming languages.
[edit] Software
By June 1963 the Time-Sharing System (TSS) Model Zero was demonstrated. It was influenced by early experiments at Bolt, Beranek, and Newman, and the CTSS project at MIT.[5] Terminals included several Teletype Model 33 ASRs. It was used until about 1968.[6]
[edit] See also
[edit] References
- ^ Oral History interview with Jules I. Schwartz, Charles Babbage Institute, University of Minnesota. Schwartz worked for the Rand Corporation on timesharing for the AN/FSQ-32 computer.
- ^ a b J. C. R. Licklider (April 23, 1963). "Memorandum For: Members and Affiliates of the Intergalactic Computer Network; Topics for Discussion at the Forthcoming Meeting". Washington, D.C.: Advanced Research Projects Agency. http://www.kurzweilai.net/articles/art0366.html?printable=1. Retrieved April 21, 2011.
- ^ Douglas C. Englebart (June 1986). "The Augmented Knowledge Workshop". Proceedings of the ACM Conference on The history of personal workstations (Palo Alto, California: ACM). doi:10.1145/12178.12184. ISBN 0-89791-176-8. http://www.dougengelbart.org/pubs/augment-101931.html. Retrieved April 20, 2011.
- ^ The IBM 4020 Military Computer. IBM General Systems Division. October 1959. http://bitsavers.org/pdf/ibm/4020/4020_Military_Computer_General_Info_Oct59.pdf.
- ^ Jules I. Schwartz; Edward G. Coffman; Clark Weissman (1964). "A general-purpose time-sharing system". Proceedings of the AFIPS April 21-23, 1964, spring joint computer conference. doi:10.1145/1464122.1464163.
- ^ Charles P. Bourne; Trudi Bellardo Hahn (2003). A history of online information services, 1963-1976. MIT Press. pp. 17–18. ISBN 9780262025386. http://books.google.com/books?id=LTTvmUU8rskC&pg=PA18.
[edit] Further reading
- Jules I. Schwartz; Clark Weissman (1967). "The SDC timesharing system revisited". Proceedings of the 22nd National Conference of the Association for Computing Machinery. doi:10.1145/800196.805996.
- Jules I. Schwartz (1966). "Online programming". Communications of the ACM. doi:10.1145/365230.365266.
- Jules I. Schwartz (1968). "Interactive systems: promises, present and future". Proceedings of the December 9-11, 1968, fall joint computer conference. doi:10.1145/1476589.1476606.
