Talk:Minicomputer

(Request made by "Deglr6328" on article page, moved to correct position by Fourohfour).

I started editing this article and quickly realized that I disagree with its central premise, which is that the minicomputer range has been (completely?) replaced by the personal computer and PC-derived servers.

Although the low end server space is robust and growing and is indeed based on systems that are essentially racked PCs, there is still a class of systems that are not PCs, and are not mainframes: the high-end SPARC, POWER, and Itanium systems from Sun, IBM and HP. Although these systems are pushing way up into the mainframe space, and are often promoted as "mainframe class," they're still not mainframes. They're also still a significantly different architecture from PC-based servers, although there's a growing similarity. I'd argue that these are the heirs to the minicomputer niche in the server ecosystem.

Before overhauling the article, I thought I'd see if anyone wanted to kick this around for a bit, but if no one else cares, I'll probably come back and make some significant changes in a few days.

In the end, I think the article should focus on the historical mainframes of the '70s and '80s, but there does need to be an explanation of how the industry has evolved since then.--NapoliRoma 20:27, 29 October 2007 (UTC)Reply

Agree also with the previous post, there should be little difficulty differentiating between minicomputer systems and personal computers, at least throughout the 1980's and 1990's. It was mostly an architectural distinction: at first minicomputers were 16 and 32 bit while the PC was 8 or 16 bit. The minicomputer was designed for multi-user, multi-tasking use with a very fast backplane and intelligent driver cards which would support large amount of peripherals, including disk and other types of storage. For the most part, it was the latter 1990's when microcomputer-based servers began to approach this type of power, and even into this decade for the typical desktop computer. Think of the minicomputer as the rack mounted PC server with multiple CPUs and you get an idea of the relative power and capability. It's also worth noting that many large companies are still quite happily running their minicomputers, for example look at how many very large OpenVMS or IBM System1 (AS/400) systems are in use.

Also, the section titled "Mid-1980s, 1990s: The minis give way to the micros" correctly speaks of the decline of the minicomputer but as I worked in this industry from the early 1980's onward I would disagree with the timeline. My perspective will be skewed to large corporate environments, small offices were more likely to move quickly to a PC-based solution because they didn't have a mainframe or even a minicomputer.

I cannot think of a lot of cases where minicomputers were replaced with networked workstations/servers/PCs in the 80's. Throughout the 1980's the PC was really not a business class machine but it did start appearing in corporate environments as a custom 'one-of' solution where a local application or database might needed that had no connection to a corporate network, to control specific technical systems (instrumentation, manufacturing systems, etc). By mid-1980's our ratio of PCs to mainframe/minicomputer terminals was about 5 PCs in 1000 desktops or less than 1%.

Even at the start of the 90's the wide use of terminal emulation (which was finally mature enough to be useful) allowed more dumb terminals to be removed but serious computing was still relegated to the minicomputer - often as a departmental- or site-based resource that still connected to a mainframe with corporate data. By that point a local workgroup might in fact have their own server, for example a group doing computer assisted design (CAD) might have a CAD application server and 'power' desktop systems, but in most corporate environments the data was very often served from a minicomputer (or even through a version control system on a mainframe).

So, the issue of desktop versus server (similar to mainframe) is one that could use some explanation and I would also agree that the newer content could be moved elsewhere, a chart or table showing the progression might be good.

Suggestions about how much of this should be added to balance the PC-centric side of this topic are also welcome.

CRM384 (talk) 22:46, 10 December 2007 (UTC)Reply

I moved this paragraph from the article to here for consideration (edit comments don't belong in the article itself). I moved it from the bottom of the "The minicomputer’s industrial impact and heritage" section:

<!-- the following pgph is redundant wrt the above; could perhaps be 'merged'(?) --wernher -->

<!-- --<

Today at the turn of the millennium few minicomputers are still in use, having been overtaken by Fourth Generation computers built using a more robust version of the microprocessor technology that is used in personal computers. These are referred to as “servers,” taking the name from the server software that they run (typically file server and back-end database software, including email and web server software).

Do with it what you may... — Frecklefσσt | Talk 14:51, 4 January 2008 (UTC)Reply

Would I be correct in assuming minicomputers (1950s-1960s-early 1970s) were misnomers for machines that executed a singular instruction in a matter of milliseconds?

Would I be correct in assuming that microcomputers (late 1970s-1980s-1990s) are misnomers for personal computers that executed a singular instruction in a matter of microseconds?Anwar (talk) 20:39, 7 May 2008 (UTC)Reply

No on both counts :) Lynbarn (talk) 13:44, 8 May 2008 (UTC)Reply

For example, a 1970-vintage PDP-11/20, according to the 1970 Second Edition of the PDP-11 Handbook, takes 2.3 microseconds to do a register-to-register add, 10 microseconds to do a memory-to-memory add in which both operands are pointed to by in-memory pointers, and some time between 2.3 and 10 microseconds for other types of operands. That's at most .01 milliseconds for a single instruction.

For another example, a computer whose central processor was a Pentium II, which was introduced in May 1997, doesn't even execute single instructions in a straightforward way; the microarchitecture of that chip, the P6 microarchitecture, is an out-of-order superscalar microarchitecture, meaning several instrucitons can be in various stages of execution at the same time. Guy Harris (talk) 08:38, 1 November 2023 (UTC)Reply

Recently, it seems that at least some of the IT press has reinvented the term "minicomputer", this time meaning a small microcomputer. Whereas the typical modern desktop (not laptop) microcomputer is about half a metre tall and weighs several kilogram, I figure a new "mini-microcomputer", what the press is already calling a "mini", would be only a few decimetres tall and weigh less than a kilogram. JIP | Talk 11:46, 10 May 2008 (UTC)Reply

My quick Google search suggests that the press is using the terms "mini-computer" or "mini computer", not "minicomputer". I do not think that "minicomputer" is being redefined, just that the new terms are similair. Rilak (talk) 12:44, 10 May 2008 (UTC)Reply

If it's just a matter of punctuation, I think the difference is fairly trivial, and so the terms could be confused. JIP | Talk 12:48, 10 May 2008 (UTC)Reply

Yes, it could be confused. However, "minicomputer" has a "formal" definition, whereas "mini computer" and variants of do not, AFAIK. I suppose you were thinking of adding something for clarification? I say go for it if the potential for confusion is that great. Rilak (talk) 06:23, 8 November 2008 (UTC)Reply

because some low iq people think mini is smaller than micro 101.119.103.213 (talk) 16:05, 24 August 2024 (UTC)Reply

I find recent edits to be of concern. The source provided does not explicitly describe the IBM 1130 as a "minicomputer", so why the claim that it was the first successful one? Rilak (talk) 08:49, 18 November 2008 (UTC)Reply

Fixed up a formatting error (no idea why it happened, looked completely wrong on my screen), broke up that part into a bit more readable chunk and fixed a simple typo... - Aereus Pestifer (talk) 02:46, 15 September 2009 (UTC)Reply

The article at http://en.wikipedia.org/wiki/PDP-8 says the PDP-8 was introduced on 22 March 1965. Since the article does not give an exact date then I think the 1964 should be changed to 22 March 1965.

Eddyq (talk) 17:40, 27 August 2010 (UTC)eddyqReply

I don't think the term "multiuser" belongs in the definition of minicomputer. The early minicomputers did not have this capability when introduced and it is primarily a matter of software, not hardware. Standard sources (OED, Britannica) do not mention multiuser. Also the term minicomputer as defined here is obsolete. A Google search shows no current usage. If anything the term "mini" is coming to mean a smaller PC or laptop.--agr (talk) 15:34, 13 February 2012 (UTC)Reply

Having been there in the late 60s, 70s and early 80s, and programing on all three platforms. The minis like the main frames had propriety operating system. And when CP/M and MS-DOS came on the scene, this gave the hardware the opportunity of sharing software written for the operating system to be shared.

The shared operating systems of the microcomputer at the end of the 70s and early 80s, gave the "Software" companies the rise above the "Hardware" companies, i.e. Microsoft Corporation, Lotus Corporation, Ashton-Tate to name a few in this era.

I also do not see there is any mention of the minicomputers created for many of the corporations for their own use, then put on the market for other businesses. i.e. General Electric, Lockheed, and Singer to name a few.

Zarcom (talk) 09:32, 13 November 2012 (UTC)Reply

In my humble opinion, the description of "mini-computer" missed the boat entirely. In the technological progression of the time (late 1950s - early 60s) transistors were in general use. By the beginning of the 60s logic design had entered the "discrete component" phase. Vacuum tubes had been replaced by transistors, but there were much greater advantages to be gained than simple substitution of transistors for tubes. Besides the tremendous difference in reliability over tubes there was an enormous savings in production costs; a transistor had 3 solder able leads, a vacuum tube base many more - 5 to 9 were pretty common. But it was quickly recognized that the circuit design goal was not to deal with components, but functions. So a whole new industry developed to supply "logic/function cards" populated with logical "gates, invertors, and amplifiers" to the manufacturers of computers and control devices (who generally held patents on the foundation designs). Companies like TI, DEC, SDS became leaders in producing a wide range of standardized logic cards. A lot (most?) digital design of the era was done writing Boolean equations representing the input and output signals, then searching the suppliers handbooks for the best solution translated into the appropriate collection of logic cards. After a few development iterations of the logic cards someone recognized that a computer could be designed around the existing logic card offerings. The architecture was different than mainframes (which were mostly based on the Erkert/Mauchly patents) and focused on a bus system of control and data flow within the computer. This distinction apparently opened the doors for marketing them without conflict with the patents that restricted "mainframe" competition. Eventually the function of the 'logic cards" was swept into the design of integrated circuits and the rest is history: enter the PC. — Preceding unsigned comment added by 71.191.131.193 (talk) 15:54, 30 January 2013 (UTC)Reply

Someone above mentioned there is a formal definition of a minicomputer. This definition should probably be listed. I disagree with this bit: "the contemporary term for this class of system is "midrange computer", such as the higher-end SPARC, Power Architecture and Itanium-based systems from Oracle, IBM and Hewlett-Packard." Midrange, for me, is the IBM word for a minicomputer. It is still in use for POWER systems running i or AIX. You could almost argue that the contemporary word for minicomputer is server, but not quite as not all minicomputers were used like servers. Any modern use, like small form factor microcomputers is not relevant for this article. — Preceding unsigned comment added by 2A02:908:180:F6C0:35F3:BEFE:542D:DF60 (talk) 07:03, 15 July 2017 (UTC)Reply

The basic definition of minicomputer I think is very much of its time. The definition in Britannica has it as " smaller, less expensive, and less powerful than a mainframe or supercomputer but more expensive and more powerful than a personal computer " . In the mid 1970s when I finished my degree in computing and started work using computer systems, the contemporary definition of minicomputer was something akin (in my experience) to the DEC PDP-9 and 15 and Raytheon RDS systems. There was no such thing as a 'personal computer' at the time, and minicomputers were seen as multi cabinet systems with various peripherals which didn't need the traditional environmental requirements of 'mainframes' - ie raised floors, air conditioning, purpose built computer rooms. Having remained in IT up to 2019 and progressed into operating in large data centres of thousands of rack mount servers and blades, I can't recall when I last heard the term 'minicomputer' used to describe any systems - probably late 1970s, but by then we were using DEC VAX systems. J R Spigot (talk) 14:35, 5 August 2023 (UTC)Reply

Hackerculture1986 a new editor added a section "Today's hobbyist constructions and interest revival" for which after searching I could not find a reliable source. Creating a section based upon two examples cited (one referenced) gives them undue weight. Absent a reliable source this sections amounts to Hackerculter's point of view and is not appropriate for Wikipedia. Accordingly I am reverting the section until a reliable source is found. Tom94022 (talk) 21:06, 3 August 2018 (UTC)Reply

Un minicalcolatore è una specie di mainframe, ma molto più piccolo. viene usato per svolgere operazioni amministrative.5.169.83.237 (talk) 15:14, 19 October 2021 (UTC)Cite error: There are <ref> tags on this page without content in them (see the help page).Reply

(In English, this being en.wikipedia.org, not it.wikipedia.org: "A mini-computer is a kind of mainframe, but much smaller. it is used to carry out administrative operations.") OK, so what type of computer isn't a kind of mainframe? Guy Harris (talk) 18:30, 19 October 2021 (UTC)Reply

The term miniature computer existed prior to the term minicomputer.

• 1954 ARMA miniature computer
• 1957 ad for Transarc Miniature Computer.
• 1957 miniature electronic computer
• 1958 children's book "miniature automatic computer"

Jamplevia (talk) 23:02, 11 February 2022 (UTC)Reply

The first three of those are analog(?) computers used for military aircraft control, and much smaller than most minicomputers; the fourth was a fictional device. Is there any connection between the term "miniature computer" and the term "minicomputer" (other than syllabic) worthy of note in the article? Guy Harris (talk) 19:17, 12 February 2022 (UTC)Reply

I would say there is no connection worthy of inclusion in this article or given WP:UNDUE in Wikipedia for that matter. Tom94022 (talk) 19:30, 12 February 2022 (UTC)Reply

The article currently claims that some midrange systems of the 60's were programmed in a form of BASIC. This seems extremely unlikely. BASIC didn't really escape Dartmouth until Version The Fourth in 1968, and that year saw a number of versions come out including HP's T-S BASIC which was the most popular of the era. I'm not aware of any versions on previous midrange machines, and certainly none were it was a primary language. I am aware of many machines after this period that did, like the HP's and Wang 2200s, but those are "true" minis. Can the author of this bit expand? I'm inclined to remove it. Maury Markowitz (talk) 18:27, 1 August 2023 (UTC)Reply

I've added {{Citation needed}} to that claim - I'm not sure what the author of that claim is referring to, either. Guy Harris (talk) 19:18, 1 August 2023 (UTC)Reply

Why tag the statement, it seems HP Time-Shared BASIC on HP 2000 from November 1968 supports it? Tom94022 (talk) 00:10, 2 August 2023 (UTC)Reply

Maury Markowitz already addressed that in his comment - he views the HPs as minicomputers rather than as a midrange system of the sort to which they were referring. If you want to argue otherwise, go ahead. Guy Harris (talk) 06:37, 2 August 2023 (UTC)Reply

OK who am I to disagree, so how about the GE225 and GE235 and timesharing systems referenced in the BASIC article? They too would not require a ref, just a link. FWIW the 1968 GE 430/440 came with BASIC standard but that would require a link. Tom94022 (talk) 06:08, 4 August 2023 (UTC)Reply

The statement in question is in Minicomputer § Definition and is

Another common difference was that most earlier small machines were not "general purpose", in that they were designed for a specific role like process control or accounting. On these machines, programming was generally carried out in their custom machine language, or even hard-coded into a plugboard, although some used a form of BASIC.^{[citation needed]}

I'm not sure why "programming was generally carried out in their custom machine language" is relevant to the "special-purpose" claim (that sounds like "generally programmed in assembler language"). However:

• for the GE-200 series, the GE-235 System Manual claims on page I-2 that "Programming aids available for the GE-235 include ... FORTRAN, a scientific compiler", along with some GE proprietary languages and two assemblers (macro and plain), so it's not limited to its "custom machine language" or to BASIC - yes, there was a time-sharing system that supported BASIC, but that wasn't the sole software for it;
• for the HP 2100 series, the HP press release for the HP 2116B notes that it - and its 2116A predecessor - had FORTRAN and ALGOL compilers - yes, there was a time-sharing system that supported BASIC, but that wasn't the sole software for it;
• for the PDP-5, DEC had a FORTRAN system that was developed for the PDP-8 but that "can also be used without change on Digital's Programmed Data Processor-5".

So I'm not sure what those two sentences are talking about. At this point, my inclination is to remove them; if somebody wants to talk about a truly specialized machine that's less limited than those three, and they have references for it, they can add it back. Guy Harris (talk) 01:30, 6 February 2024 (UTC)Reply

I added {{Example needed}} to get somebody to indicate what machines they're thinking of. Guy Harris (talk) 03:16, 6 February 2024 (UTC)Reply

I have a degree in computer science from Edinburgh University in 1974. At school in 1970-71 we wrote programs on paper tape and took these to a 'computer system' at Dundee University which was the archetypal concept of a mainframe environment. Raised floor, glass partitioned, air conditioned, operators in white coats. At Edinburgh University we had PDP-8, PDP-9 and I think PDP-15 (?). No special computer room environment, these were 'minicomputers'with tape drives, plotters and printers.

At work in oil exploration in the mid 1970s boom, when there were many many companies involved, everyone had some form of onsite minicomputer system configured with tape drives, plotters, disk storage and sometimes array transform processors. Data was processed very quickly behind the seismic acquisition crews allowing rapid decision making and this was only possible because you could operate these in an office environment, portakabin, villa in a compound etc

Everyone knew these as minicomputers which did everything a mainframe could do without all the infrastructure and formality of a mainframe setup. Most of the oil companies I knew had minicomputers in their offices, they may have had an established computer department and mainframe but there was a definite demarcation in all senses between the mini and mainframe users.

Living and working through that time, while everyone now wants to establish a physical/rational definition of minicomputer, to many of us the definition is more of a concept, a mindset.

Few people had cameras, the oil exploration industry has been boom and bust, boom and bust and while there were scores of exploration and oil companies, now effectively there are only a handful and so much information and knowledge has been lost.

I posted two photos of myself working on what were known as minicomputers at the time, 1974/75 in Mogadishu in a room in a villa in a compound, and 1978 in an office in Benghazi in a compound. I worked in and visited many countries in Africa, Middle East, Far East and North America in small offices, portakabins, apartment buildings, all with minicomputer systems.

So personally I think the term minicomputer as very much of its time and represents the step change in computer usage, much like 'personal computer' - who would use that term today? J R Spigot (talk) 23:28, 24 January 2024 (UTC)Reply

Thanks for sharing your history but u know an editor's own personal experience are prohibited as a source in Wikipedia. There is a definition section in this article which is where such material from a reliable source (RS) belongs. I happen to agree with you that no requirement for a machine room is one important definitional aspect of minicomputers but to add it needs an RS. If I have the time I might look for one. Tom94022 (talk) 08:22, 25 January 2024 (UTC)Reply