PDP-4 250
Minis table icon

Part II - Making the Cuts, And the Winners Are...

Header graphic3 PII
  • The PDP-4 at left took up two racks and that makes it doubtful that it should be classified as a mini. However, since it came later than the 160 and is also larger, it couldn’t win as the first mini anyway.

    Arnold Del Carlo, Photographer. Sourisseau Academy for State and Local HIstory, San Jose State University

    IBM-1620 250
    Thus we don’t really need to decide for sure whether a size of two racks is too big.
  • How about the IBM-1620 “Cadet” at right? It cost roughly the same as the 160 and is dated the same year. But it’s almost twice the volume and far more power. Weighs 50% more. Perhaps we should do more research to confirm the ship dates, but if they hold, the 160 certainly has a stronger claim to the crown as the first mini, so the 1620 drops out of the competition.

    Once again though, IBM does deserve some recognition for building a very early small computer. If it were shown to have shipped before the CDC-160, that would make the 1620, the largest remaining machine on the list. In fact, it occupies more volume than the PDP-4. There might be a question of whether it’s too large to be a mini.

  • UM1-NKh computer2 250RCould the Russian UM1-NKh at right be the first of the small minis? The problem here is that we must recall our very first ground rule for the mini: It must have been a general purpose computer. The UM1-NKh shipped initially with just 256-words of RAM. There is mention of what appears to be read only memory for programs. The applications seem to be solely in industrial process control.

    With total power consumption of just 100-200W, (sources vary) it was an extraordinarily slow machine, an estimated 60-times slower than the PDP-8. It seems doubtful that was used to run a high level language. In our definition of the general purpose mini, we specifically avoided including such low-level controllers. Unfortunately, the UM1-NKh is admittedly and clearly just such a controller. We’re sorry to take it off the list, because we wanted to avoid any hint of bias against the folks in the East. Indeed, it is a remarkable and early small computer. Also, a caveat: the information we currently have on this machine is sketchy and inconsistent. It’s possible that new information could elevate its status. We intend to update this article as needed in the future.

So here is the list after the first cut:

Maker

Model

Price

Date

Volume

Weight

Package

AC Power

Word Leng

RAM

Mem

Mem Cyc

CDC

160

$60K

Jun-60

~39 cuft

~800 lb

desk

1380 VA

12-bits

4K wds

core

6.4 us

MIT/DEC

LINC

$44K

Mar-62

~49

~1070

rack+console

1000

12

1-2K

core

8

SDS

910

$41K

Aug-62

28

900

1-rack

1870

24

2-16K

core

8

DEC

PDP-5

$27K

Sep-63

24

540

1-rack

780

12

4-32K

core

6

DEC

PDP-8

$18K

Apr-65

8

250

desktop

780

12

4-32K

core

1.6


Comments on the First Cut Table

  • What we have in these finalists is a generally decreasing cost, volume, weight and power with respect to time.
  • If we were to require that a mini have less of any one of these four parameters, it could override the time priority, to favor a later winner.
  • One thing we can say for sure: The PDP-8 was certainly a mini. So at least by that point, we know we have a winner; there’s no need to look later for the first mini.

How to Make the Final Cut?

Should Low Price Be a Requirement?

Suppose price is a strong consideration. It looks like we have nodes at roughly $20K, 30K, 40K and 60K. Bear in mind that these are generally minimal system prices and actual systems might have more peripherals or memory. So actual systems prices might vary a great deal. It seems to me that the prices are pretty thickly populated, meaning that there isn’t a clear dividing line here in our list. It would just depend on the market. The only thing I know to say about the market is that they clearly stood up and cheered, when DEC achieved the $18K figure. So they were definitely happy with something cheap. On the other hand, lots of small to medium size organizations could afford the higher figures and there was a strong pay-back argument to be made. Most of these machines were successful products. If a machine is every bit a mini except for price, does that really make it not-a-mini? Later, high end minis like the 11/70 would surface at prices in the $70K range for the basic processor and memory. Were they not minis? I find it hard to draw a red line based on price alone, so let’s look at other factors.

Should We Exclude Based on Weight?

Our remaining contenders drop from the 1000lb range, down to a couple hundred pounds. One might very well want to require that a mini be something that a couple guys could lift. The 160, LINC and 910 are in the 1000lb region and the PDP-5 is half of that. The PDP-8 is clearly passable. So we have the heavyweights, the lightweight and one in the middle. Leaving that one (the PDP-5) aside for the moment, we simply have to decide whether a mini can be a 1000lb-ish box or not. Considering that actual minicomputer systems often had large disk drives, tape drives and other peripherals taking up multiple racks, I don’t really see 1000lb as being a problem if the size was reasonable.

How About Volume?

PDP-5 computer 250PDP-8 in a rack 200This is actually a scalar way of getting at size. Like weight, volume is generally decreasing with time. We might decide that a mini simply has to fit on a desktop; that racks are just out. On the other hand, the PDP-5 at right was very close to being a PDP-8, just implemented with general purpose logic cards instead of more compact packaging. If it’s practically the same as the iconic PDP-8, costs not much more and burns the same power, is it not the same as a PDP-8? The PDP-8 packaged in rack form is shown at left, taking up most of a standard bay. If the PDP-8 is SO much of a mini icon, is the PDP-5 to be excluded only by its skin?

One other point about volume: I notice that desk models like the CDC-160 seem to look much less obtrusive for a given volume than rack models. I attribute this to the fact that a desk form factor is closer to being a space-efficient cube, than is a vertical rack. Also, office workers need a desk anyway, so if the computer doubles as a desk, it has theoretically saved that space. In any case, the 39-cuft CDC-160 desk seems no more imposing than the 24-cuft PDP-5 rack. So perhaps volume and packaging need to be given more consideration than just being a number. In any case, I can’t find any compelling reason to reject any of the remaining machines, based on volume. To me, a mini could easily occupy a rack or a desk package.

Conclusions

It looks like we have found two categories in our remaining list of early minis: The earlier “large minis” typically weighed ~1000lbs, cost $40-60K and occupied a single rack or desk package. (The LINC also had a desktop console but that could trade against the teletypes used by other minis.) The “small minis,” heralded by the PDP-8, were typically desktop, weighed 250lbs or less and cost less than $20K. The PDP-5 was a transitional model, with much lower cost, half the weight and lower power than the other large minis but still packaged in a rack.

CDC-160 computer 500The CDC-160 at right was the first of the large minis in June, 1960, making it the first, general purpose, production minicomputer. That gives Seymour Cray the distinction of having created the first minicomputer, as well as (arguably) the first supercomputer! While its $60K price is higher than the other two left in our list, it hardly seems out of place. Considering that recently eliminated large mini candidates were coming in at $120K, $64K, $90K and $65K, it seems reasonable. In view of its early place in the list, the price isn’t considered an impediment in the large mini category. Hence, the CDC-160 is the winner of the large mini category.

CDC-160 control panelBy the way, in the detail of the control panel at right, you can see that the digits displayed the model numbers of three CDC computers: The (0)160 itself, the 1604 mainframe for which the 160 was an I/O processor and the 6600 supercomputer. The funny thing is, the 6600 was not publicly announced until two years after this photo was published!41

PDP-8 small mini winnerSo that leaves us with just the PDP-8 at left, as the first small mini (general purpose, production minicomputer) in April, 1965. That explains why it was such a phe­nom­enon. While it was only one of a parade of machines with rapidly shrinking sizes and prices, it did represent a major step in reduction of weight, size and price. All of those taken together made it a market winner and the winner of the small mini category. The PDP-8 founded a product line which would consist of more than twelve models, last 25-years and produce some 50,000 units. (Ref#15, p.181) By any measure, it was an epic success. One reservation: As noted above, the Soviet computer, UM1-NKh, was rejected based on the crude information we have available. Should it prove to be something other than what we surmised, the situation could change.

Notice that we have been able to nail down the winners without having to resort to any arbitrary choice of threshold for any of the major parameters (price, weight, volume or power), in the qualified competition. Rather, it was done by recognizing general categories of these parameters and judging individual cases, ordered by time, to find the winners. [Albeit, parameter standards were loosely imposed in our definition of the mini.]

Honorable Mentions

We should have a moment of silence for the various drum computers, which foreshadowed the minis. Machines like the Librascope LGP-30 and Bendix G-15 (both shown on previous page) were quite successful in their day. Over 500 LGP-30s were built.23 At the time, there was simply no low-cost way to build main memory electronically, so the drum was pressed into service. Its foibles of required programming tricks, slow access, noise, vibration, power, weight and size combined to put these machines in a different class from the minis. However, they did deliver low-cost, stored-program computers to small organizations who sorely needed them. They also introduced a new generation to the advantages of small computers, paving the way for the minis.
IBM 1401 computer 350

The IBM 1401 at right, first announced in 1959, was truly a workhorse for small organizations around the World. While it was certainly smaller than most mainframes at the time, it just didn’t fit the category of a mini or even a large mini: 2170-lbs, 62cu ft, 3.7kVA, 3-phase power.42 So it was out on power and weight, and size was questionable.

IBM model-A 150To complicate matters, I have to admit that IBM did sell a rarely-seen, half size configuration of the 1401 as seen at left. Called the Model-A, it had just 1.4KB of memory. However, it would not meet our standard of being able to run a high level language. At 1013lbs, 31cuft and 3.2kVA 3-phase power, it was closer to a mini but still out on power as well.42

With over 10,000 units built, The 1401 was clearly a superstar of small computers. It was also the first computer I ever touched.

In the mini competition, IBM’s 1620 (shown above) was edged out by a matter of months for the honor of being the first large mini. Being the largest on the list after the PDP-1, some might question whether it even fits our large mini category but fortunately, we didn’t have to address that here. IBM’s smaller (but still way out of our small category) 1130 was preceded by only months by the PDP-8. Considering that IBM’s machines sold in large numbers and that they had a huge production force to prepare, one could argue that the bulk of Big Blue’s work on small computers was ahead of the nimble, smaller companies. Clearly, these early minis deserve a place of honor.

Afterword

What we’ve really seen here is that it’s not easy to define exactly what we mean by something like “the first minicomputer.” It’s complicated by the fact that there is an incredible amount of diversity in the free marketplace and that there is a constant forward march, reducing cost and other limiting factors. Perhaps the real value of this quest was not just to answer the question but to understand better, the progression of clever machines which contributed to the emergence of the mini. Truly a remarkable journey.

Acknowledgments

We would like to thank the following sources for the photos used in this article. All copyrights of these images are the property of their respective owners:

Notes on the Tabular Data

For simplicity, we state all power figures as VA (except in a couple direct quotes), whereas some sources state watts and others VA. Strictly speaking, the watts figure would be lower than the VA value but the difference is typically modest. While power is important for electricity costs, physical installations are usually concerned about maximum amperage and therefore, VA.

  • CDC-160 — Ref#8: Model 160 computer [alone] $60K, dated June 1, 1960. Ref#9: June, 1960 build date. Ref#10: dated 1960, shows 12-bit word. Ref#11 shows first shipment 1959. Ref#12: 1380VA, 6.4us memory, 4K core. Ref#13: The actual 160A model dimensions are given and show 38.5cuft volume, 810lbs weight. Since the 160 and 160A models are very similar physically, we take the volume and weight as being decent approximations for the 160. Ref#14 shows only 700VA. Ref#16 suggests that by July, 1960, CDC had been selling the 160 into the scientific market very successfully, helping to support the June, 1960 date.
  • PDP-1 — Ref#15 p.121: First delivered November, 1960. After the third shipment, the console was integrated as in the final PDP-1/C. “PDP-1/C used four cabinets instead of the three cabinets of the earlier versions.” pp.165,166: $120K with paper tape I/O and typewriter, 2160VA, 1350lbs, 94cuft, Memory “1,4...165K” but PDP-4,7,9 of the series were limited to 32K. On p.128, Bell states, “...4Kword memory which was later expanded to 64Kwords.” The next reference has the PDP-1 at 4K max. Therefore, we surmise that PDP-1 was not built beyond 64K. Ref#14 p.765: $110K, 70cuft+console table, 1600lbs (presumably includes the then-separate console station), 1K-4K 5us core, 18-bits, 800VA.
  • IBM-1620 — Ref#18 p.680: Announced October 21, 1959. No ship date found. Comment from Mike Radow on Columbia Univ site (Ref#17): “By 1959 – or 1960, at the latest – ‘Watson North’ had a small 1620...” From this we conclude that the computer was shipping no later than Dec-1960. Ref#14, dated Mar-1961 lists two installations and that may not be exhaustive, lending some support to the Radow comment. Ref#40 gives deliveries beginning in 1962 but checking its reference (which is our Ref#18 pp.508-513), we do not find support for that. Bashe does say, “Well over a thousand systems had been delivered by the end of 1963.” He also says that “around the end of 1959,” the 1620 was chosen to be IBM’s solution for process control in industrial plants. The system for the first of three field trials of those was completed in Oct-1960. Ref#14: $64K, 1210lbs, 2000VA, area is 22sqft. Estimating height as 30in for desk and 13in for the console yields 43in and hence 79cuft, 20K-60K digits of memory. Ref#19: Decimal digits are represented by 6-bit codes which include an end-flag for variable length data. Memory is addressed by individual 6-bit digits but 12-bits are read from memory at each 20us access. Data fields are variable length, from 12-bits to all of remaining memory. Instructions are 72-bits. Text characters are represented by two digits (12-bits). Memory size is 10-30K of 12-bit words. Considering how to specify word length for the table, 12-bits would represent the width of memory well but with instructions being 72-bits, we felt the most useful expression would be 12-72-bits. The only thing left out is the variable length data fields. However, that is more like a microcoded feature than a measure of the width of a hardware bus, so we settled on the 12-72-bit word, as the best short answer.
  • CDC-160A — Ref#13: 38.5cuft, 810lbs, 1840VA. Ref#21: Memory 6.4us, 8-32K core. Ref#22: $90K includes computer with 8K and paper tape reader/punch. Ref#9: July, 1961 build date. Ref#24 gives release in 1963 but contains other errors, as noted. Ref#25 and #26 help support Ref#9. Due to the authoritativeness of 25/26, we will go with Ref#9 July, 1961. This could use better documentation though.
  • LINC — Ref#15 pp.176-178: First shipment 3/62, $43600, 1000VA, 12-bit word, 1-2K core, 69 x 32 x 30in (main rack) plus separate tape, keyboard, console and interconnect boxes. Rack is 38cuft. From pictures, we estimate the desktop units to be 40 x 24 x 19in or 11cuft, giving 49cuft total. To estimate the weight, we note that the PDP-4 is about the same volume. Scaling from the volume ratio, we get 1030 x (49/47) = 1074lbs.
  • PDP-4 — Ref#15 pp.165,166. Volume given as 2-bays of 69 x 21 x 28in yielding 47cuft. Price is with 4K, paper tape reader/punch and typewriter.
  • SDS-910 — Ref#22 $41K is for 2K x 24-bit memory. Price with 4K is $56K. Dimensions 75 x 24 x 27in yields 28cuft. 900lbs, 2-16K 8us memory. Power is given as 800watts but Ref#28 shows 17amps at “110V” which implies 1870VA--Going with the more pessimistic value. Ref#27 gives first delivery date of August, 1962.
  • PDP-5 — Ref#15 pp.198,199. Memory 4-32K, p.179.
  • UM1-NKh — From Ref#1: 32 x 20 x 13in (4.8cuft), 150-watts, 132lbs. Ref#2 mentioned ROM for instructions. Ref#3 gives 33 x 21 x 13in without power supply. Ref#4 gives “Bit: number - 15 bits (14 + 1 sign) commands - 20 bits, 5000 additions per second, [RAM] - 256 words, Permanent storage device commands - 2048 words, [35 x 21 x 13in (5.5cuft)], weight power supply - [176lb], power consumption - 200 watts.” Note that this doc refers to it as UM1-HX but from other refs we believe it is the same or similar processor. From the 200us add time, we estimate a memory cycle of roughly 100us, based on similar computers. Ref#5 has “By order of the Central Committee of the CPSU and the USSR in 1963 began development and mass production of [UM1-NKh]at the Leningrad Electromechanical Plant.” Not having the month of shipment, we can only say that is was by Dec-63. Ref#6 confirms that production started in 1963, states that 200 units were built and had a cost of 50000 rubles. The ruble of 1961 was equal to 0.987g of gold. Value of gold in 1963: $35.09/troy-oz. So the cost of the UM1-NKh was $55676. However, Note-35 on p.191 gives a weight of 55lb, 100-watts power and volume of 2.5cuft, which differs from above. Ref#7 states that the UM-1 (the predecessor of UM1-NKh) had a 13-bit word but doesn’t indicate whether or not that changed in UM1-NKh.
  • PDP-8 — Ref#15 pp.198,199.
  • IBM-1130 — Ref#29 p.497 shipped in Dec-65. Ref#30 p.10 760lb, 208/230V 1.1kVA, Dimensions 58.25 x 29 x 45.5in yields 44cuft. Ref#31 p.3 4-32K 16-bit memory, 3.6us for Model-1 and Model-2, 2.2us for Model-3. Ref#32 Price was $32,280 in March, 1965. See also http://ibm1130.org/.

References

  1. Firdman, Decision Making, 52-53; Soviet Cybernetics Technology: V. Soviet Process Control Computers, Memorandum RM-4810-PR (Santa Monica, CA: Rand Corporation, Nov. 1965), 21-31.
  2. Malinovsky, B. N., Pioneers of Soviet Computing 2nd Ed., p.146: “the logic part of the [UM1-NKh] and read-only memory (ROM) of constants and instructions used discrete elements”, 2010. http://www.sigcis.org/files/SIGCISMC2010_001.pdf (4MB file)
  3. CIA report, “Computers in Communist Countries: Production, Requirements and Technology”, p.6, 1966.
  4. http://www.computer-museum.ru/books/vt_face/prilogenie_16.htm, as translated from Russian by Google. Note that choosing “English” at the top of the Russian page does not bring up a translation. You need to give the URL to Google’s translation feature. Retrieved January 23, 2014.
  5. http://www.computer-museum.ru/books/vt_face/9_staros_1.htm, as translated from Russian by Google. Note that choosing “English” at the top of the Russian page does not bring up a translation. You need to give the URL to Google’s translation feature. Retrieved January 23, 2014.
  6. Usdin, S. T., Engineering Communism, p.213, Yale University Press, New Haven, 2005.
  7. Firdman, H. E., Maverick for Life, Without Parole, 1st Books Library, 2004.
  8. Control Data Corp., “Price Index on 160 Computer and Associated Peripheral Equipment,” June 1, 1960.
  9. http://www.computer-archiv.de/, Control Data Corp. page shows “built starting 1960.06” [June, 1960], retrieved January 13, 2014.
  10. Control Data Corp., 160 Computer Programming Manual, 1960.
  11. http://www.cray-cyber.org/memory/cdc.php, Time line of CDC shows “1959 CDC 160 first shipment.” Retrieved January 13, 2014.
  12. Control Data Corp., “Control Data 160 Computer,” sales brochure, August, 1961. Specifies 115V 12A.
  13. Control Data Corp., Control Data 160-A Computer Installation Manual, September, 1962.
  14. Weik, M. H. (Ballistic Research Laboratories), A Third Survey Of Domestic Electronic Digital Computing Systems, March, 1961.
  15. Bell, C. G. et al., Computer Engineering, Digital Press, Maynard, 1978.
  16. Gandy, A., The Early Computer Industry: Limitations of Scale and Scope, pp.277,278, “CDC had been selling the 160 computer into the scientific market very successfully. In July, 1960 a team from NCR visited the CDC operation in Bloomington, Minneapolis. It was arranged that NCR would sell the system to the banking and retail sectors. The system was sold as the NCR 310.” Palgrave Macmillan, 2012.
  17. http://www.columbia.edu/cu/computinghistory/1620.html, Columbia University page on the IBM-1620. Retrieved, January 14, 2014.
  18. Bashe, C. J. et al., IBM’s Early Computers, MIT Press, Cambridge, 1986.
  19. IBM Systems Reference Library, IBM 1620 Central Processing Unit, Model 1 [manual], 1965.
  20. Malinovsky, B. N., History of Computer Technology in Persons, in Russian via website, retrieved January 14, 2014.
  21. Control Data Corp., “Control Data 160-A Computer,” sales brochure, no date.
  22. Weik, M. H. (Ballistic Research Laboratories), A Fourth Survey Of Domestic Electronic Digital Computing Systems, January, 1964.
  23. Flamm, K., Creating the Computer: Government, Industry and High Technology, Brookings Institution, Washington, 1988.
  24. http://en.wikipedia.org/wiki/Control_Data_Corporation, CDC Wikipedia article shows “release” of the CDC-160A in 1963 but it also shows the 160 as 1961 (which conflicts with documented dates). Retrieved January 14, 2014.
  25. http://www.cbi.umn.edu/collections/cdc/histtimeline.html, “CDC Historical Timeline,” States that the CDC-160 was “delivered” in 1960 and the CDC-160A was delivered in 1961. This is the website of the Charles Babbage Institute (CBI) at the University of Minnesota. The CBI holds the archives of Control Data Corp.
  26. http://www.cbi.umn.edu/collections/cdc/prodtimeline.html, Similar to Ref#25 but “CDC Product Timeline.” This dates the CDC-160 as December, 1959 and the CDC-160A as April, 1961. However, it doesn’t specify whether these dates are for introduction, first shipment or something else.
  27. Bell, C. G. et al., Computer Structures: Reading and Examples, McGraw Hill, New York, 1971.
  28. Xerox Data Systems, XDS-910 Computer Reference Manual, El Segundo, 1970. Note that Xerox acquired SDS and hence, the SDS-910 was renamed as XDS-910.
  29. Pugh, E. W. et al., IBM’s 360 and Early 370 Systems, MIT Press, Cambridge, 1991. Also covers smaller computers from 1960 on.
  30. IBM Systems Reference Library, IBM 1130 Installation Manual — Physical Planning, 1971.
  31. IBM Systems Reference Library, IBM 1130 Functional Characteristics, 4th ed., no date.
  32. Bell, C. G., “STARS: Rise and Fall of Minicomputers,” IEEE Global History Network, http://www.ieeeghn.org/wiki6/index.php/STARS:Rise_and_Fall_of_Minicomputers, Retrieved January 20, 2014.
  33. Murray, C. J., The Supermen: The Story of Seymour Cray and the Technical Wizards Behind the Supercomputer, John Wiley & Sons, New York, 1997.
  34. Metropolis, N. et al, editors, “A History of Computing in the Twentieth Century,” Academic Press, Orlando, 1980.
  35. May, K. O., “Historiography: A Perspective for Computer Scientists,” in Ref#34 pp.11-12, “...because it is meaningless to speak of the invention of the computer. Things are much more complicated than that: controversies over who invented something are usually unprofitable [sic] — these are complex events in which many people take part...a particular aspect of the computer or computing changes its meaning with time. And so statements about who did what exactly first are not very meaningful.”
  36. http://www.wisegeek.com/how-much-weight-can-a-standard-elevator-hold.htm, “The capacity of a standard passenger elevator varies, but generally falls between 1,000 and 6,000 pounds...Generally speaking, a standard elevator in a low-rise building can hold about 2,000 or 2,500 lbs. The larger the building, the more capacity it the elevator will hold, however.” Retrieved January 22, 2014.
  37. International Code Council, Inc., 2010 Oregon Structural Specialty Code, pp. 374, 375, http://ecodes.biz/ecodes_support/free_resources/Oregon/10_Structural/10_ORStructural_main.html, Shows con­cen­trated load limits for office buildings at 2000lbs. Schools, stores and libraries are 1000lbs. Retrieved January 22, 2014.
  38. In Ref#32, Bell features the PDP-1 in his article on minis; albeit, he does not explicitly call it a mini there.
  39. http://en.wikipedia.org/wiki/Minicomputer, Previously, this Wikipedia article claimed that, “The first minicomputer was created in the USSR in 1958–1962. The computer, designated UM-1NKh...” That seemed questionable, in the light of the findings presented here. No definition of mini was given with the claim and it does not fit the New York Times definition quoted earlier in the article. That’s because it cost much more than $25K. Retrieved January 22, 2014. That section has since been edited by Tom94022.
  40. Reilly, E. D., Milestones in Computer Science and Information Technology, p.130, Greenwood, 2003. “In 1962, IBM began delivery of a remarkable little machine originally called the Cadet but marketed as the IBM 1620.” The date seems doubtful. (See IBM-1620 under “Notes on the Tabular Data” above.)
  41. From Ref#33 p.92, the CDC-6600 was announced in August, 1963. The brochure from which the CDC-160 photo was taken was marked August, 1961. However, Ref#29 p. 379 states, “News...came in July [1962], when trade publications reported that the Atomic Energy Commission had contracted for a CDC-6600 system...” Of course, that isn’t the same as a formal announcement but if taken as such, then the brochure “spilled the beans” about the 6600 only about a year prior. On p.382, Ref#29 also confirms the formal announcement of the 6600 in August, 1963. Development of the 6600 had started around the end of 1960. (Ref#33 p.79: “By early 1962 almost sixteen months had passed, but all Cray and his group had to show for their time and effort was the faulty multiply unit.”)
  42. International Business Machines Corp., Installation Manual — Physical Planning, IBM 1401 Data Processing System, 1962. Note that IBM sold the 1401, bundled with a card reader/punch and printer. Since AC power is routed through the reader/punch, they quote combined power. Model-D was taken as a representative example of the full CPU configuration.

Copyright © 2014-2017 by Stephen H. Lafferty. All rights reserved.

 

Please feel free to add your comments on the previous page.

<<< Return to Part I - Defining Minis and Finding Candidates

 

 

Making the First Cut

Now we can go over the list and find the most obvious candidates to cut. At this point, we’re not going to reject any for date priority, which might otherwise win on the basis of a parameter threshold.

  • IBM-1130 250The PDP-8 beats the IBM-1130 at right, both on ship dates and parameters, so the 1130 is out. However, we note that coming in the same same year as the PDP-8, IBM does deserve some recognition for building the early mini.
  • Since the CDC-160A came later and cost lots more, the fact that the CDC-160 does qualify as a mini makes the 160A superfluous, so it’s out.
  • At twice the price of the 160 and occupying 4-racks, the PDP-1 (shown on previous page) is preempted by the 160. So the PDP-1 is out as the first mini, without having to decide exactly how big a mini can be.