Phase III: Game Consoles and Personal Computers
Electronic games, either for home systems or arcade machines, were an instant success. Different manufacturers such as Coleco (short for Connecticut Leather Company, a clear example of extreme diversification), Fairchild, and many others introduced their own home gaming systems, and competition fostered innovation. Game cartridges were introduced in 1976, trackballs in 1978, and so on.
Game Consoles and Game Developers
It was during this fast growth period, sometime in the 1970s, that the distinction between console manufacturers and game developers was established. Console manufacturers would invest large sums of money in creating the hardware platform and selling it at an attractive price to consumers. Sometimes, the price would hardly cover (or even be below) the real cost of the hardware. Manufacturers would also build tools (very primitive at this stage) that allowed outside companies to create games for the console. The reason why other companies were given access to the business was obvious. As game consoles began to compete against each other for market share, being able to offer many games became the main competitive advantage, and hence the importance of the developer. No manufacturer could ever have the bandwidth to deliver the amount of games required to make its console platform attractive.
Thus, outside companies began coding games for console systems. At this stage, little or no quality control was performed by the console manufacturer; quantity was valued over quality. For each game sold, the developer agreed to pay a royalty fee to the console manufacturer to cover the console development costs and thus generate revenue. Some console manufacturers focused on the hardware business only, whereas others chose to act as developers as well, sometimes causing conflicts of interests. Did the outside developer have the same information as the internal team at the manufacturer? Sometimes first-party teams had access to better information than third parties, which caused complaints. With some variations, this production model still applies today. The main difference with today's model is that consoles are sold way below their manufacturing price, especially when they debut. This is a risky decision, because the hardware manufacturer has to sustain significant losses for some time. Then, as the console manufacturer begins receiving royalties from developers, it is able to recover from the initial losses and finally make money.
The king of programmable consoles at the time became the Atari VCS, also known as the Atari 2600, which was introduced in 1977 with a price tag of $249. Because it became the de facto standard, I will use it to exemplify what coding for a console meant in the 1970s. The console featured a 6507 CPU equipped with 128 bytes of RAM, which were used to store state variables such as the life and ammunition levels. The program was stored in an external, interchangeable cartridge, which was inserted into the console. One side of the cartridge was full of electrical connections. Thus, inserting it into the console integrated the data chips in the cartridge as part of the console's hardware, usually as memory banks where the program code and data were stored.
The memory cartridge method was used with minimal variations until the Nintendo 64 and the Gameboy Advance, whereas more modern systems employ different kinds of disks as storage media. Storing games on CD-ROMs and DVDs makes them much cheaper to manufacture, at the cost of increasing the risk of piracy and having to enhance the memory capabilities of the console to store the program. But for the Atari 2600, 6KB of ROM were usually built into the cartridge and held the game code and data together in a memory chip. The 6507 CPU ran at an incredible 1.19 MHz and had an address space of 8KB.
Assisting the CPU were the television adapter, known as TIA or Stella, and the I/O chip, called RIOT. Stella was accessed through registers and had the extreme limitation of not having a frame buffer or physical memory to hold the screen's contents. Modern adapters (from CGA onward) have a memory region that holds the screen data pixel by pixel, so copying data to that region effectively paints the screen. The Atari 2600 did not have such a buffer. So, the screen was drawn by reading some Stella registers serially, synchronized to the electron beam from the TV set. The CPU had to synchronize itself with the electron beam and write those registers at exactly the right speed, so it looked correct and produced no flicker. As an example of the limitations imposed by the hardware, here is the sequence of a game loop for the Atari 2600:
Start the vertical blanking interval Start the vertical sync interval Here is space for 80 micro instructions End vertical sync Perform game computations here End vertical blank Now the screen rendering starts... Send each line to the register 6 instructions can be fit here Loop lines until the screen is rendered Go to first step
The 2600, as all consoles were at that stage, was coded in machine-specific assembler. Program data and source code were part of a single memory block. Data was just bytes of information trailing after the code, which program code never jumped to. An untested program that began to execute data addresses would simply go mad. Here is an excerpt of a combat game for the 2600, showing how code and data were part of a whole:
B15A7 STX D2 LDX #03 B15AB LDA L1765,Y EOR D1 AND D2 STA COLUP0,X STA D6,X STA D8,X INY DEX BPL B15AB RTS J15BD LDA #00 B15BF INX STA A2,X BNE B15BF RTS L15C5 .BYTE $0E ,$0A ,$0A ,$0A ,$0E ; 0 .BYTE $22 ,$22 ,$22 ,$22 ,$22 ; 11 .BYTE $EE ,$22 ,$EE ,$88 ,$EE ; 22 .BYTE $EE ,$22 ,$66 ,$22 ,$EE ; 33 .BYTE $AA ,$AA ,$EE ,$22 ,$22 ; 44 .BYTE $EE ,$88 ,$EE ,$22 ,$EE ; 55 .BYTE $EE ,$88 ,$EE ,$AA ,$EE ; 66 .BYTE $EE ,$22 ,$22 ,$22 ,$22 ; 77 .BYTE $EE ,$AA ,$EE ,$AA ,$EE ; 88 .BYTE $EE ,$AA ,$EE ,$22 ,$EE ; 99 L15F7 .BYTE $F8 ,$F7 ,$F6 ,$06 ,$06 .BYTE $06 ,$16 ,$17 ,$18 ; $15FC .BYTE $19 ,$1A ,$0A ,$0A ; $1600 .BYTE $0A ,$FA ,$F9 ,$F8 ; $1604 .BYTE $F7 ,$F6 ,$F6 ,$06 ; $1608 .BYTE $16 ,$16 ,$17 ,$18 ; $160C .BYTE $19 ,$1A ,$1A ,$0A ; $1610 .BYTE $FA ,$FA ,$F9 ,$E8 ; $1614 .BYTE $E6 ,$E4 ,$F4 ,$04 ; $1618 .BYTE $14 ,$24 ,$26 ,$28 ; $161C .BYTE $2A ,$2C ,$1C ,$0C ; $1620 .BYTE $FC ,$EC ,$EA ; $1624
Most of the time, games were built by a single person who laid down the memory map, wrote the program code, designed the graphics, and even provided the sound. Sometimes an artist helped out, but doing graphics is a relatively straightforward task on such limited hardware. Code reuse, so popular today, was virtually nonexistent. At most, programmers would borrow sequences of microinstructions from a previous title so a new game could be coded faster. But generally speaking, coding for such a machine involved lots of craft and skill, as anyone who has coded assembly programs for a living knows.
While consumers were busy playing with game consoles, the personal computer revolution was about to begin. Computers appeared as a by-product of the discovery of the integrated circuit by Texas Instruments in 1959. The first computer, the Digital PDP-1, appeared in 1960 and featured a keyboard and a monitor. It debuted at $120,000, clearly positioning computers in the corporate market: Accounting, engineering, and databases were their main uses. By 1964, the BASIC language appeared, allowing intuitive programming.
Douglas Engelbart invented a two-dimensional pointing device called the "mouse," which is now widely used. By 1968 (one year before humanity reached the moon), the first home computer, the Honeywell H316, was introduced at $10,600. The Apollo Guidance Computer, developed specifically for the Apollo 11 spaceship, was an impressive feat. It ran at 2MHz, had 64KB of ROM and 4KB of RAM, and was capable of performing 50,000 additions per second.
In 1972, when Atari and Magnavox became successful, computers were gaining ground. The C programming language was introduced, the first ideas about laptop computers appeared at Xerox's Palo Alto Research Center (PARC), and Intel introduced the 8008, a 16KB, 200kHz, 60,000 instructions-per-second, low cost chip. From this moment on, computers advanced at blazing fast speed. In 1974, Intel unveiled the 8080, a 2MHz, 16KB, 640,000 instructions-per-second chip. That same year, the MITS Altair 8800 was announced in Popular Electronics for about $400.
One year later, Microsoft was founded. Its flagship product was a BASIC interpreter for the Altair. It was the first programmable language for a personal computer. Years later, Microsoft would create the operating system for the IBM PC, and the rest is history.
Then, in 1976, one of the most popular garage startups surfaced, and the personal computer industry was born. Apple was founded by Steven Jobs and Steve Wozniak. Jobs, with an eye on product design and business, had been at Atari since 1974. Wozniak, the hardware genius, worked for Hewlett-Packard. Their first product, the Apple I, sold in kit form for $666, and was introduced at the Homebrew Computer Club. Some simple games began to appear for the Apple I, but they were mainly clones of those existing on game consoles.
But it was with the advent of the Apple ][ in 1977 when games began to pour in quickly. A decisive factor in this direction was the Apple ]['s CPU: a 6502, extremely similar to the one running Atari's own 2600. The computer featured new 16KB RAM modules to reach a theoretical 48KB of total RAM, which was larger than anything on the market at the time. The bundle also included a BASIC interpreter to code your own programs, a QUERTY keyboard, a cassette interface, and a dedicated game I/O connector for paddles, joysticks, and so on. It could display up to 280x192 pixels in 16 colors, making it an ideal delivery platform for games.
The Apple ][ was state-of-the-art technology, clearly surpassing the products of the competition, including game consoles. It could be programmed, used for games and office applications, had a large memory area, and offered "full" color support (the full 16 colors in the palette, that is). The downside was, obviously, the cost. The base kit cost $600, whereas a full-featured, 48KB "supercomputer" was $2275, which was mainly due to high memory costs of those early days. But Apple ][s were sold by the truckload, and hundreds of games were coded for it—a notorious example being the first installments of the popular Ultima series.
Five years later, IBM unveiled the IBM PC, which ran on an Intel 8086 CPU and Microsoft's Disk Operating System (MS-DOS). The main advantage of the PC was that the architecture was soon to become open. This enabled other companies to design and sell their own PCs, which were all compatible with the original IBM model. Competition fosters innovation, and an open platform is the best way to enable evolution. The personal computer era had begun.