DRAM vs SRAM

In simple terms

There are two main technologies for the fast, volatile memory inside a computer, and they make opposite trade-offs. DRAM (dynamic RAM) is cheap and dense — it’s what your gigabytes of main memory are made of — but it’s relatively slow and has to be constantly “refreshed” to avoid forgetting. SRAM (static RAM) is much faster and doesn’t need refreshing, but it’s bulky and expensive, so it’s used only where speed is critical and the amount is small — chiefly the cache right next to the CPU.

More detail

The difference comes down to how each stores a single bit:

• DRAM stores each bit as charge in a single tiny capacitor (plus one transistor). Capacitors leak, so DRAM must be refreshed thousands of times per second — rewritten before it forgets. The upside: one capacitor + one transistor per bit is tiny, so you can pack billions of bits cheaply.
• SRAM stores each bit in a latch of six transistors that actively holds its value as long as it has power — no refresh needed, and access is very fast. The downside: six transistors per bit means it takes far more chip area, costs much more, and can’t reach the same densities.

This is exactly why the memory hierarchy exists. You can’t make all memory SRAM (too expensive and bulky) or all of it as slow as DRAM (too slow for the CPU). So computers layer them: a small amount of blazing-fast SRAM as cache, backed by a large pool of cheaper DRAM as main memory, backed in turn by even cheaper, slower storage.

Both are volatile — they lose their contents when power is removed — which distinguishes them from flash memory.

Why it matters

The DRAM/SRAM trade-off is the physical reason computers are built as a hierarchy of memory rather than one big uniform pool. It explains why CPUs have only a few megabytes of cache but gigabytes of RAM, why cache misses are so costly, and why memory access patterns matter so much for performance. Understanding it demystifies a lot of what otherwise looks like arbitrary hardware design.

Real-world examples

• The “16 GB RAM” in a laptop is DRAM; the “32 MB L3 cache” on its CPU is SRAM.
• A CPU’s L1/L2/L3 caches are SRAM precisely because they must keep up with the processor’s clock.
• DRAM refresh consuming a small but constant amount of power is part of why idle memory still draws energy.

Common misconceptions

• “SRAM is just better than DRAM.” It’s faster, but far more expensive and less dense — which is why we use a little of it (cache) and a lot of DRAM (main memory), not all SRAM.
• “SRAM is non-volatile because it’s ‘static.’” “Static” only means it doesn’t need refreshing; it still loses everything when the power goes off. Both DRAM and SRAM are volatile.

Learn next

This trade-off is the reason for the memory hierarchy; SRAM is what the CPU cache is built from.