CSS Optional — Paper-II, Computer Organization & Architecture
1. Definition
Computer generations classify the evolution of hardware technology by the dominant switching component used to build logic circuits, from vacuum tubes to today's VLSI microprocessors.
2. Classification — The Five Generations
| Generation |
Technology |
Period |
Notes |
| 1st |
Vacuum tubes |
1940s–50s |
Huge, hot, unreliable (e.g., ENIAC) |
| 2nd |
Transistors |
Late 1950s–60s |
Smaller, faster, more reliable |
| 3rd |
Integrated Circuits (IC) |
1960s–70s |
Many transistors on one chip |
| 4th |
Microprocessors / VLSI |
1970s–present |
Entire CPU on one chip |
| 5th |
AI-based / parallel processing |
Ongoing |
Multicore, GPUs, AI accelerators |
3. Moore's Law
- Definition — the observation that the number of transistors on a chip roughly doubles every 18–24 months, driving decades of exponential performance growth.
- Limitation — physical constraints (heat dissipation, quantum leakage at nanometre scales) have slowed the pace since the mid-2000s, a period often called the "power wall."
Transistors(t) = Transistors(0) × 2^(t / doubling_period)
4. Programming Language Evolution (parallel to hardware)
- Machine language — raw binary, tied to one specific hardware.
- Assembly language — mnemonic instructions, one-to-one with machine code.
- High-level languages — FORTRAN, COBOL — hardware-independent, compiled.
- Structured languages — C, Pascal — control-flow discipline.
- Object-oriented languages — C++, Java — encapsulation, reuse.
- Modern languages — Python, Rust, Go — safety, concurrency, developer productivity.
5. Performance Scaling Shift
- Before ~2005 — performance scaled mainly with clock frequency.
- After ~2005 — the power wall made frequency scaling impractical; performance gains now come mainly from multicore parallelism, not raw clock speed.
🗺️ Architecture Diagram — Generational Timeline