Second Generation Computers (1955 - 1964)

Third Generation Computers (1964 - 1971)

The third generation of computer hardware technology began to be implemented in the mid 1960's93, with many improvements being driven by IBM research. As the technology developed, record processing and record storage came to depend on two physically quite different technologies.

Processing and short-term memory for "work in progress" used by the computer came to be purely electronic. Because electronic circuits used no mechanical components, data could be very rapidly moved and manipulated. The downside was that the information vanished as soon as the power was lost.

Magnetic media rather than electronics came to be used for long term storage, because once the media was magnetized in the appropriate pattern to store the data, the pattern persisted, and could be stored off-line and read back into the computer years later. The downside was that reading and writing required the magnetic media to be physically transported past the read/write heads and/or the heads to be physically positioned. By comparison to moving electrons in an electronic circuit the magnetic storage process was slow and ponderous, and could only record and access information in linear sequences - meaning that reading or writing would often require major mechanical movements of the magnetic media and or reading apparatus to locate a particular item of data required. As will be seen, beginning with the third generation computers, both technologies began incredible shrinking acts, allowing increasingly more data to be read, written and processed faster, and to be packed into ever smaller and less expensive physical devices

In 1964 IBM introduced its System/360 model computers with transistorised "Solid Logic Technology"94 processors, but still using ferrite core memory.

Printed circuit boards and transistor technology began their incredible shrinking act, memorialised in Moore's Law95 - with Intel launching its first large-scale integrated circuit processor, the 4 bit 4004, in 197196. This was followed in 1972 by the 8 bit 8008 processor which powered the first personal microcomputers, and which provided Paul Allen and Bill Gates with their first experience programming microprocessors (Delany 1997) (Figure 10). Large scale integration also applied to the processor's random access memory, with single chips able to store prodigious amounts of data.

In 1971, IBM also introduced its Magnetic Tape Selectric Typewriter, a product that pioneered the application of magnetic recording devices and transistorised electronics to typewriting, and gave rise to the concept known today as word processing. Referred to as "power typing," the feature of revising stored text improved office efficiency by allowing typists to type at "rough draft" speed without the pressure of worrying about mistakes." This was based on an earlier generation of electromechanical typewriters (often used as input/output devices for early generation computers) using paper tape as a recording mechanism (Eisenberg 1992)97. As will be seen in Episode 3, below, interesting things started to happen when electric typewriters connected to the computers started to be used for authoring human readable texts.

Figure 10. The Intel 4004 microprocessor98 (left) from 1971 - the first commercial large-scale-integrated microprocessor on a single chip containing 2,300 transistors, 4-bit logic and able to address 640 bytes of memory. This was followed in 1972 with the 8008 microprocessor99 (right) with 3,500 transistors and 8 bit logic and able to address 16 KB of memory. Compare these with the three generations of component parts that came before the microprocessor: a vacuum tube module from around 1950, single transistors on a printed circuit board from 1956, and small-scale-integrated circuits from 1964 illustrated by da Cruz (2001). The 4004's word length was 4 bits with a clock speed of approximately 100,000 Hz. It contains no internal RAM. It is possible to discern Individual circuits and transistor elements in this photomicrograph of the whole chip.

Where magnetic storage is concerned, IBM introduced its first fixed disk system - the RAMAC - in 1957. Based on IBM's rental practices at the time and price adjusted to current dollars, it would then cost $100,000,000 (one hundred million dollars) to purchase RAMAC storage for a gigabyte of data. In 1998, disk storage cost around $50 per gigabyte (Gilheany 1998). IBM introduced removable magnetic disk storage in 1962, with the IBM 1311 (Figure 11). This "hastened the end of the punched card era". A single disk pack could hold as much information as 25,000 punched cards (2 MB)100. A Disk/Trend table shows the increasing storage capacity and reduced size of the media from 1950 through 2000101. What it does not show is that these units also became less and less expensive. Gilheany (1998) also tabulates the annual decline in cost of storage for each year since 1992 and estimates that the annual decline in price per unit of storage is around 37.5 percent.

The photos show the operator loading an IBM 1311 memory disk pack on the IBM 1440 computer. In the picture sequence, the operator carries a disk pack containing six 14-inch memory disks to the IBM 1440. She places the disks on the disk storage drive spindle, removes the pack cover, and the unit is ready to operate. The time required is less than one minute. Disk packs can be removed in the same manner and stored like books on a library shelf until they are needed again. This data storage technology makes it possible for a user to maintain a separate disk pack for each of his major data processing jobs. The disk packs, which weigh less than 10 pounds, have a storage capacity of nearly 3,000,000 alphanumeric characters. [Photos and text from Weik (1964.]

Figure 11. IBM 1311 disk drive with removable storage. A unit cost $90,000 upwards to purchase in 1964102.

The Fourth Generation and Beyond