Dynamic Random-Access Memory
Subject: Computer and Technology noun /daɪˈnæm.ɪk ˈræn.dəm ˌæk.sɛs ˈmɛm.ə.ri/ 1 definition
- computer-hardware
- computer-architecture
- volatile-memory
- semiconductor-memory
-
Definition (noun)
Entry #222 · UpdatedDynamic Random-Access Memory (DRAM) is a high-density, volatile semiconductor memory used as the primary system memory in modern computers, servers, and mobile devices. Each memory cell stores one bit as an electrical charge in a tiny capacitor paired with an access transistor, enabling direct, non-sequential access to any memory address at much higher density and lower cost per bit than static memory. Because charge leaks from the capacitors within milliseconds, a memory controller must periodically read and rewrite every row—a process known as memory refresh—to retain data while powered, and all stored contents are lost once power is removed.
Common Types
Synchronous DRAM (SDRAM): Synchronizes memory operations with the system clock to enable command pipelining; its double data rate (DDR) successors transfer data on both clock edges in standard desktops and servers.
Low-Power DDR (LPDDR): Reduces operating voltage and adds deep sleep states to minimize energy consumption in smartphones, tablets, and thin laptops.
Graphics DDR (GDDR): Trades slightly higher access latency for wide, high-frequency channels optimized for parallel throughput in discrete GPUs.
High Bandwidth Memory (HBM): Vertically stacks DRAM dies via through-silicon vias (TSVs) on an interposer beside the processor, providing ultra-wide buses for AI accelerators and data-center GPUs.
Example
When a workstation compiles a large codebase using 64 GB of DDR5 DRAM, the CPU accesses active symbol tables and build buffers in tens of nanoseconds while the memory controller continuously issues refresh cycles every few microseconds to prevent capacitor charge loss. If the system loses power before saving changes to its SSD, every unsaved intermediate state held in DRAM disappears immediately.