Understanding Data Storage: RAM, DRAM, and Memory Controllers

In this article the author, a senior architect, introduces the fundamental problem of data storage in computers and explains how memory modules work.

1. How to Store Data

Data must be represented by stable electrical states; a constant high voltage can represent a binary 1 and a constant low voltage a binary 0, with the ability to switch between them.

Memory modules (RAM) provide this capability.

2. RAM

RAM (Random Access Memory) allows arbitrary read/write of any location. Two circuit approaches have been used historically:

Static SRAM : each bit is stored in a latch made of multiple transistors, keeping a stable state without refresh but requiring many transistors per bit, making it costly and large.

Dynamic DRAM : each bit is stored as charge on a capacitor; this is much simpler and denser, but the charge leaks, so the cell must be refreshed periodically (typically every 64 ms). This scheme is called Dynamic Random Access Memory.

Images illustrate the two cell designs.

3. Memory Controller

To read a specific bit, the system must specify chip, bank, row, and column addresses. The memory controller abstracts these details, providing a simple interface for the CPU.

The controller also manages the periodic refresh of DRAM cells, ensuring data integrity.

It translates a single CPU address into the appropriate chip, bank, row, and column locations, and nowadays it is integrated into the CPU itself.

Because CPU speeds have outpaced memory, CPUs also include small caches to store frequently accessed data, reducing the need to access main memory.

Images show the controller and address mapping.

The author invites readers to discuss and share viewpoints, and provides links to additional resources and community groups.