The Early Computer Industry
Computer Memory
Volatile memory
Think of the concept of computer memory like a piece of paper. Memory is a medium where information can be written, read, or erased. At the simplest level of written communication, the combination of different letters or symbols create certain words, conveying information. In morse code, for example, contrasting dots and dashes serve as distinct building blocks that stand for specific words when combined. To "write" information in a language a computer can understand, engineers use binary code, which consists of two basic units of electricity: a negative charge (0) and a positive charge (1).
Reading positive and negative charges is more complex for a computer than it is for a person to understand a written language. Depending on the medium used to write the information (vacuum tubes, cores, chips, etc), some systems applied a negative charge, "0," when accessing stored informational data, leading erasure of the informational data itself. Other systems lost all informational data every time the system powered down and stopped receiving an electrical charge. The types of systems mentioned above used volatile memory, which only temporarily stored informational data. While a memory system with a quick reset cycle is necessary in some situations, continuously destroying all stored information becomes inefficient. Complicating the situation, early memory also required sequential reading.
Sequential memory
As a person, you could choose to read the Great Gatsby and then halfway through choose to read Of Mice and Men. It is not necessary to read one book to understand the other. However, early computer memory required information to be read in a serial order to complete any task. Devices such as hard disk drives still use sequential memory today, but that's by design rather than for a lack of an alternative.
From the mid to late 1900s, computer memory experienced numerous changes as engineers sought to create memory components that offered more flexibility to create tailored systems suited to a person's needs. As the forms of reading, writing, and storing computer memory became more diverse, computers became increasingly capable of performing significantly complex actions at a faster rate and for a cheaper price.
Early Forms of Memory
Vacuum Tubes
When Bill was at Devry Institute, vacuum tubes were in demand. Vacuum tubes provided fast switching speeds that enabled higher calculation rates. However, vacuum tubes were fragile, unreliable, produced too much heat, used too much power, and took up vast amounts of space. Vacuum tubes are analogous to the transistor as they are quite literally the analog form prior to the development of solid state devices. A Bell Laboritories paper co-authored by William Shockley, the co-creator of the transistor, noted:
"The behavior of this device is closely analogous to that of a vacuum tube: the emitter region [of the transistor] corresponds to the cathode, the base to the region around the grid wires, and the collector region corresponds to the plate."#
Magnetic Drum
Drum memory consisted of a long metal cylinder coated in magnetic material that held 40 tracks. Each track in the magnetic drum received a read-write head located on the axis of the drum. In 1951, Engineering Research Associates (ERA) designed and built the first computer to use magnetic drum memory known as the ERA 1101, later renamed the UNIVAC 1101.
The memory element in the ERA 1101 held a capacity of 16,384 words and a drum rotational speed of 3,500 revolutions per minute. It consisted of 38 different operations that included ten arithmetic operations, thirteen insert or transmissive operations, four jump or transfer operations, four manipulative aids operations, two shifting operations, two output operations, and three stops. Unlike modern computers, the ERA built the ERA 1101 only for integer math.
Cathode-Ray Tube (CRT)
The Braun Tube became the earliest version of the cathode-ray tube memory invented by German physicist Ferdinand Braun in 1897. However, the Williams-Kilburn Tube developed by Manchester University in 1947 became the first high-speed random-access electronic memory using a CRT. It is the Williams-Kilburn Tube that Jay Forrester's Project Whirlwind team initially used to develop the first digital computer at MIT and ultimately leading to the creation of magnetic core memory
While innovative for its time, CRT random-access memory suffered numerous drawbacks including limited storage capacity, high production costs, prone to failure, high rejection rate during production, and data loss. The data stored in a CRT lasted a fraction of a second before fading, necessitating the need to frequently refresh the data rendering permanent long-term computer memory impossible. However, if the data on the tube was read, the tube erased the information, requiring rewrites to preserve the data. For Forrester, cost and his increasing frustration with the complexity of CRT random-access memory forced him to pivot his focus away from analog memory to digital memory. In 1953, Forrester introduced magnetic core memory in MIT's Whirlwind Computer.
Core Memory
Despite the fact that the invention of transistors precede it by a few years, the invention of core memory by Jay Forrester in 1949 completely overshadowed transistors for a time. Core memory posed a more reasonable option in computers when combined with the technology available at the time. From 1955 to 1975, magnetic-core memory was the dominant form of computer memory. The core memory board consisted of magnetic wires that looped into tiny donuts. Each donut, or core, was made from semihard ferrite material and represented one bit.
Wires fed through the core were designated as horizontal x-address lines, vertical y-address lines, and sense lines. As the electrical current traveled along the wire, the core created enough current that the magnetic force traveled in the direction of the corresponding 1 or 0 value. The computer then performed specific tasks based on the given binary number. Though revolutionary at their debut, their cost, material usage, and speed inhibited continuous innovation.
Solid State Transistors
A transistor is a semiconductor-based device, which is a device made of materials that conduct energy only in certain conditions, such as germanium or silicon, that serve as an ideal medium to control the flow of electrical current. Transistors are used in the same vein as vacuum tubes. They amplify electronic signals or act as a switch in a circuit, allowing energy to flow in some conditions and blocking that flow in others.
In 1947, the first working transistors were invented at Bell Labs in a series of discoveries and tests by John Bardeen, Walter Brattain, and William Shockley. In 1954, Jean Felker and engineer James R. Harris designed and built a fully transistorized computer called the Transistor Digital Computer (TRADIC).
The invention of the transistor marked a significant step forward towards the development of the integrated circuit. While the transistor pushed engineers to design more complex electronic circuits, these circuits still relied on discrete components (such as transistors, resistors, and capacitors), each separately manufactured, and interconnected via wiring or hand-soldering onto circuit boards. This process was expensive, time-consuming, and unreliable as each soldered joint posed potential issues. The challenge became identifying a quicker, reliable, and cost-effective way to produce the discrete components and interconnect them.
Integrated Circuits: The Start of Silicon Valley
In 1959, integrated Circuits were independantly, but simultaneously, invented by Jack Kilby and Robert Noyce (soon to be Co-Founder of Intel).
In 1957, Jack Kilby proposed to the U.S. Army Signal Corps the idea of creating small ceramic substrates called micromodules for each discrete component of uniform size and shape, with the wiring built into the components themselves. The modules then piled on top of one another and snapped together to create an electronic circuit, eliminating the need for soldering the components together. This led to the establishment of the Micromodule Program sponsored by the U.S. Army Signal Corps. As the project gained momentum, Kilby left the U.S Army Signal Corps and joined Texas Instruments in June 1958.
At Texas Instruments, Kilby proposed a new idea that argued for the fabrication of all electronic components (i.e. transistors, capacitors, resistors, etc) onto the same board rather than stacking discrete modules on top of one another. On September 12, 1958, Kilby built and demonstrated the first integrated circuit prototype to managers at Texas Instruments. Kilby’s integrated circuit was composed of a thin slice of germanium to serve as a bulk resistor, a single bipolar transistor, four input/output terminals, a ground, and gold wires. This became the first integration of all electronic components onto a single substrate. Additionally, in order to keep the integrated circuit small, engineers relied on the use of transistors. Now designers and engineers could significantly increase the processing power of a single circuit board by adding multiple computer chips that each contained an integrated circuit, and controlling their input via transistors.
Robert Noyce and Jean Hoerni worked at William Shockley's small Semiconductor Laboratory in Mountain view, California. When the operation fell through due to mismanagement, Noyce, Hoerni, and six others went on to co-found Fairchild Semiconductors funded by The Fairchild Camera & Instrument Corporation of New York.#
Jean Hoerni later created the planar process of manufacturing in 1958, becoming an essential building block for the integrated circuit. Building on Hoerni's planar process, Noyce created a design for a monolithic integrated circuit. Noyce's design, like Kilby's, used semiconductive material for all components and placed all components on a single board. However, Noyce chose to use silicon rather than germanium. Silicon acted as an electrically neutral base, allowing it to serve different functions. For example, it can serve as the base, the collector, or the emitter of a transistor.
This multifunctional material allowed for the interconnection of multiple components on the surface of the chip. To accomplish this, a protective oxide overlayed the silicon allowing for the use of aluminum lines to poke through to connect the components. Noyce patented his invention in July of 1959, only a few months after Kilby patented his design. As their design processes hadn't been influenced by each other, hence the both of them being considered cofounders of the integrated circuit.
"Kilby is credited with building the first working circuit with all components formed using semiconductor material; Noyce with the metal-over-oxide interconnection scheme that yields a monolithic structure."#
The independent invention was contentious legally. According to the Computer History Museum, "Fairchild and TI engaged in litigation over I.C. patents for many years. The courts eventually ruled in Noyce's favor but by then the companies had already settled on a cross-license agreement that included a net payment to Fairchild."# Noyce and Kilby shared the National Medal of Science for the invention. Robert Noyce passed away in 1990 before Jack Kilby received the 2000 Nobel Prize in physics for integrated circuits, but Kilby mentioned Noyce as a part of the invention's success in his acceptance of the award multiple times.
In 1963 the first transistorized Static Random Access Memory (SRAM) using bipolar junction transistors was created by Robert Norman at Fairchild.# SRAM was a form of volatile memory that didn't require a sequential read but always required power to retain its memory. SRAM cells stored energy in a four transistor flip-flop circuit that connected to external components using two more transistors. Each cell contained one bit. A year after the invention, John Schmidt, also at Fairchild, created the first Metal Oxide Semiconductor SRAM.
The first Dynamic RAM (DRAM) was invented by Robert Dennard at IBM between 1966 and 1967.# A DRAM cell stored its energy in a capacitor which removed the necessity for the computer to stay connected to power. The capacitor connected externally with only one transistor. Though this memory was non-volatile, capacitors leaked electrons over time, so DRAM cells needed refreshing to continualy store data. The invention of DRAM became incredibly important to Regitz' creation of the 1103 DRAM chip in 1971.
Memory and computer processing have a symbiotic relationship where the advancement in one area can inspire advancement in another. The creation of DRAM, specifically its usage of metal oxide semiconductive material, directly influenced the creation of the first microprocessor, or as it's commonly called, the CPU. The Intel 4004 was generally considered the first CPU, invented by Gordon Moore and Robert Noyce in 1968. Moore and Noyce would later found Integrated Electronics (Intel) that same year.
The Early Computer Industry
