Skip to content
Tutorials Center
My learning

Topics

Lesson 2 of 55 ยท Meet the Computer

Generations and History of Computers

Walk through the five generations of computers, from vacuum tubes and transistors to microchips and AI, and test Moore's law with a Python simulation.

Every computer you use today is the result of about eighty years of rapid change. A machine that once filled a room, burned as much power as a small factory and made a few thousand calculations per second has been replaced by a phone that fits in a pocket and does billions. Historians and textbooks group this progress into generations, and each generation is defined by one key technology.

This lesson tours those generations, explains why each jump made computers smaller, cheaper and faster, and then tests the famous rule of thumb called Moore's law with a small simulation. For the short timeline from gears to chips, see What Is a Computer?; here we go deeper into the electronic era.

Four hardware generations

1st: vacuum tubes Room-sized, hot, unreliable.
2nd: transistors Smaller, cooler, cheaper.
3rd: integrated circuits Many parts on one chip.
4th: microprocessors A whole CPU on one chip.

Generation boundaries are approximate, and historians draw them in slightly different places.

1 / 2

First generation: vacuum tubes (about 1940 to 1956)

The earliest electronic computers used vacuum tubes, glass bulbs that act as electrical switches with no moving parts. A tube can turn a current on or off thousands of times per second, which is far faster than the relays and gears it replaced. The price was high. A machine needed thousands of tubes, each one produced heat and used a lot of electricity, and tubes burned out often, so the computer might spend more time under repair than running.

Programs were written directly in machine language, long lists of numbers, as described in Machine Code and Assembly. Input came on punched cards or paper tape, and results were printed. Only governments, universities and a few large companies could afford such a machine, and each one was looked after by a team of operators.

Second generation: transistors (about 1956 to 1963)

The transistor, invented in 1947, does the same job as a tube but is made of solid material (a semiconductor) rather than glass and a heated filament. It is smaller, uses a small fraction of the power, produces little heat and almost never wears out. Replacing tubes with transistors made computers more reliable and much cheaper to run.

This generation also brought the first widely used high-level languages, such as FORTRAN for science and COBOL for business, so people could write programs in something close to English and math notation. Operating systems for batch processing appeared: jobs were queued on tape and the machine ran them one after another. See Types of Operating Systems for how that style compares with today's.

Third generation: integrated circuits (about 1964 to 1971)

Engineers learned to build many transistors, wires and other parts together on one slice of silicon, called an integrated circuit (IC), or simply a chip. Instead of wiring thousands of separate transistors by hand, a factory could print them all at once. That made computers smaller, faster and cheaper again.

Software grew up too. Operating systems could now run several programs at once and let many people share one machine through terminals with keyboards and screens, rather than handing in a stack of cards. This is the time-sharing idea you met in Processes, Threads and Scheduling.

Fourth generation: microprocessors (about 1971 to today)

In 1971 the first microprocessor, the Intel 4004, put the whole central processing unit on a single chip (see The CPU). A computer built around a microprocessor could be small and cheap enough for a desk, and the personal computer was born in the late 1970s. Graphical interfaces with windows, icons and a mouse arrived in the 1980s. Networks grew, and in the 1990s the web connected the world's computers.

The same chips now sit in laptops, phones, cars and washing machines. Chip makers keep packing more transistors into the same area, which is why a modern processor has billions of them. This is also the generation where memory became cheap enough for gigabytes of RAM in ordinary computers.

A fifth generation?

Many books describe a fifth generation that is defined less by one hardware part and more by capabilities: artificial intelligence, speech and image recognition, massively parallel processors such as GPUs (see Motherboard, Buses and the GPU), cloud computing and mobile devices. The label is fuzzy. A phone assistant that understands speech runs on the same kind of transistor-based chips as a fourth-generation computer, only far more of them working together. Quantum computers are sometimes described as a future generation, but today they are experimental machines for narrow problems, not replacements for everyday computers.

Moore's law

In 1965 Gordon Moore, who later helped found Intel, observed that the number of components on a chip was growing at a regular, exponential pace. In the form most people quote, the number of transistors on a chip roughly doubles every two years. It is not a law of physics, only an observation that became a goal the industry worked hard to meet. Let us compare the rule with real chips. The program below starts at the 4004's 2,300 transistors in 1971, doubles every two years, and lists some well-known processors with approximate transistor counts.

Python Moore's law versus real processors

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The prediction tracks reality surprisingly well over almost fifty years: the real counts stay within a factor of a few of the rule. Doubling every two years for 47 years means more than 20 doublings, about a 12-million-fold increase. Exponential growth is hard to imagine, which is why a modern phone is so much more powerful than the room-sized machines of the first generation.

Now let us turn the idea into something tangible: how long would a hypothetical task take if the same speed gain applied to it?

Python What exponential growth does to time

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

In 1971 the job takes about six weeks; by 2003 it takes under a minute, and by 2019 a fraction of a second. Real speed-ups are not that clean, because clock speeds stopped rising around 2005 as chips ran into heat limits, so designers added more cores instead of a faster core. The doubling has also slowed, since transistors are now only a few dozen atoms wide. That is why newer chips depend on specialized hardware and smarter design, topics covered in Computer Organization.

Moore's law describes transistor counts, not speed or usefulness. More transistors only help when engineers find ways to use them: more cores, bigger caches, special circuits for graphics or AI.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Which technology defines the second generation of computers?

    Transistors replaced vacuum tubes, making computers smaller, cooler and more reliable.
  2. 2. What did the first microprocessor, in 1971, put on a single chip?

    A microprocessor is a CPU on one chip, which made small, cheap personal computers possible.
  3. 3. What does Moore's law say, in its common form?

    It is an observation about transistor counts per chip, not about speed, price or software.
  4. 4. Why did the third generation (integrated circuits) lower costs?

    Printing many transistors and wires at once is far cheaper than wiring separate parts by hand.

Summary

  • First generation: vacuum tubes. Second: transistors. Third: integrated circuits. Fourth: microprocessors. A fifth, defined by AI and parallel hardware, is a loose label.
  • Each step made computers smaller, cooler, cheaper, more reliable and faster.
  • Software developed alongside hardware: machine code, then high-level languages, operating systems and graphical interfaces.
  • Moore's law, a doubling of transistors about every two years, held for decades but is slowing, so designers now use more cores and specialized chips.

Continue with Hardware and Software.