Skip to main content
ScienceVerse
TechnologyDifficulty 2-4

Memory

Computer memory holds the instructions and data a processor is using, while storage keeps information when the power is off. Working memory (RAM) is usually volatile: DRAM stores each bit as a tiny charge on a capacitor that leaks and must be refreshed. Computers combine small, very fast memories near the processor with larger, slower ones further away, and measure capacity in bytes using decimal (GB) or binary (GiB) prefixes.

Sign in to save this concept.

Locality and caching

Caches work because programs show locality: recently used data tends to be used again soon (temporal locality) and data near a recent access tends to be used next (spatial locality). Caches therefore move data in fixed-size blocks and keep the most frequently useful ones close to the core.

tavg=thit+m×tmiss penaltyt_{\text{avg}} = t_{\text{hit}} + m \times t_{\text{miss penalty}}

Average memory access time: hit time plus miss rate m times the miss penalty.

Example (our calculation): with a 1 ns hit time, a 5% miss rate and a 60 ns penalty, the average access time is 1 + 0.05 × 60 = 4 ns. Halving the miss rate to 2.5% gives 2.5 ns, a bigger gain than shaving the hit time.

  • DRAM: one transistor and one capacitor per bit; dense and cheap per bit, but the charge leaks, so cells need refresh.
  • SSD storage: persistent solid-state memory, used for data that must survive power-off.
  • ROM: pre-recorded and not writable in normal operation.
Common misconception: 'Volatile' describes data retention without power, not speed or reliability.
Full explanation — the complete reference version every reading depth is based on

Memory versus storage

'Memory' in everyday computing usually means working memory: the fast random-access memory (RAM) that holds the program and data in use. RAM is volatile — it loses its contents when power is turned off or lost. Storage, such as a solid-state drive (SSD), keeps data persistently, so your files are still there after a restart.

A hierarchy of memories

  1. Registers: a few ultra-fast locations inside the CPU core.
  2. Cache (L1, L2, L3): high-speed memory near the core holding frequently used instructions and data.
  3. Main memory (DRAM): much larger working memory for running programs.
  4. Storage (SSD or hard disk): persistent, largest capacity, slowest to reach.
  5. Read-only memory (ROM): pre-recorded contents that can be read but not written.

How DRAM stores a bit

In 1966 Robert Dennard at IBM conceived a memory cell that holds one bit with a single transistor, storing the bit as a charge on a capacitor; the patent was issued in 1968. The capacitor's charge leaks away over time, so each cell must be read out and rewritten again and again — the refreshing that makes the memory 'dynamic'. For several generations of DRAM the standard refresh interval has been 64 milliseconds.

Counting capacity

1 GiB=230 B=1 073 741 824 Bvs1 GB=109 B1\,\text{GiB} = 2^{30}\,\text{B} = 1\,073\,741\,824\,\text{B} \quad\text{vs}\quad 1\,\text{GB} = 10^{9}\,\text{B}

Binary (IEC) and decimal (SI) prefixes give different sizes for similar-sounding units.

Worked example (our calculation): a drive sold as 500 GB holds 500 × 10⁹ bytes. Dividing by 2³⁰ gives 500 × 10⁹ ÷ 1 073 741 824 ≈ 465.7 GiB, which is why software that counts in binary units shows a smaller number. No space has gone missing; the two units are simply different sizes.

How we know

The definitions of volatile memory, ROM and SSD come from NIST's glossary; the history of the one-transistor cell comes from IBM's own heritage account; the 64 ms refresh figure comes from a peer-reviewed 2012 computer-architecture study of DRAM refresh; and the unit prefixes are fixed by the SI Brochure and by NIST.

Assumptions and limits

Not every memory is volatile and not every storage device is slow: ROM keeps its contents without power, and solid-state storage blurs the line between memory and storage. The 64 ms refresh interval is a conservative standard set by the leakiest cells; most cells hold their data far longer.

Common misconception: Misconception: 'more storage makes a computer faster'. Storage capacity decides how much you can keep; speed depends far more on the processor and on how much fast working memory and cache it has.
Info: Connections: memory is one of the five parts of a stored-program Computer, feeds the CPU, and is built from semiconductor devices (transistors and capacitors on silicon).

Ask ScienceVerse

Still curious about Memory? Ask a question, get hints, take a short lesson or try a challenge. The tutor answers only from this concept's approved sources, and says so when it has none.

Ask the tutor about this concept on the full tutor page.

Connections

Prerequisites

Understand these first:

Guided learning path

See everything to learn before this, in order, with your progress:

Related concepts

Try the experiment

Put this concept into practice with a hands-on activity (each shows its supervision requirement first):

Check your understanding

Take a quick check of two to five questions, with an explanation for every answer:

See the neighbourhood of Memory in the Knowledge Galaxy

Sources and methodology

  • Volatile memory loses its contents when power is turned off or lost. (awaiting scientific review)
  • Read-only memory (ROM) is a pre-recorded storage medium that can be read from but not written to. (awaiting scientific review)
  • A solid-state drive (SSD) is a storage device that uses solid-state memory to store persistent data. (awaiting scientific review)
  • Dynamic random-access memory (DRAM), conceived by Robert Dennard at IBM in 1966 and patented in 1968, holds each bit with a single transistor and stores it as charge on a capacitor that must be refreshed repeatedly. (awaiting scientific review)
  • A DRAM cell's capacitor leaks charge over time, so every cell must be periodically read out and rewritten (refreshed); a 2012 study reported that the standard refresh interval had stayed at 64 ms for several DRAM generations. (awaiting scientific review)
  • A processor's cache (levels L1, L2 and L3) is high-speed memory near the processor core that stores frequently accessed instructions and data. (awaiting scientific review)
  • SI prefixes always denote powers of 10, so one kilobit is 1000 bits rather than 1024 bits; the binary prefixes kibi (Ki), mebi (Mi) and gibi (Gi) denote 2¹⁰, 2²⁰ and 2³⁰. (awaiting scientific review)
  • One gibibyte (GiB) is 2³⁰ = 1 073 741 824 bytes, whereas one gigabyte (GB) is 10⁹ = 1 000 000 000 bytes. (awaiting scientific review)

Claims marked “awaiting scientific review” cite the sources listed but have not yet been signed off by a scientific reviewer.

Content status: published 1 October 2026.

  • Scientific review: this version has not yet been signed off by a scientific reviewer.
  • The Advanced explanation has not yet been reviewed for age suitability.