4.2. How Memory Works

Memory chips were originally attached directly to the system board of a computer. As the need for more memory increased and the space on the system board started to become scarce, memory chips were mounted on cards called memory modules. Figure 4-6 shows a typical dual inline memory module (DIMM). Notice the parallel rows of memory modules and the gold pin connectors.

Figure 4-6. DIMM.

The single inline memory module (SIMM) was one of the first types of memory modules developed. Memory chips were mounted on one or both sides of a SIMM, but electrical signals from both sides were routed to the same set of gold pins. SIMMs could handle only 32 bits of data at a time.

Today's HP ProLiant servers are built to handle 64 bits of data at a time, so SIMMs were eventually replaced by DIMMs. DIMMs have memory chips on both sides, and each side has separate pins for a total of 144 pins. Some DIMMs have 168 pins. The extra pins handle error checking and correcting (ECC) capabilities. (ECC is explained later in this chapter.)

The electrical traces from the memory chips are routed from the memory chips to gold pins located at the bottom of the memory module. The pins snap into slots on the system board or a memory board.

An application-specific integrated circuit (ASIC) known as the memory controller sits on the system board. The memory controller controls the memory bus, the pathway between the processor and the memory modules. The memory bus has two parts: the address bus and the data bus. Figure 4-7 illustrates how the memory controller utilizes the address bus and data bus.

Figure 4-7. The memory controller uses the address bus and data bus to communicate with memory.

In addition to the address and data pins, a typical memory module has other pins that it uses to communicate with the memory controller. The row address strobe (RAS) and column address strobe (CAS) are pins used to communicate the address of the cell the memory wants to set. Typically, they carry a high voltage. The memory controller communicates with the memory chip by changing the voltage of the RAS and CAS.

The memory controller uses the Write Enable (WE) pin to indicate whether an operation is a read or a write.

To read or write data, the memory controller sends a row address to the RAS and a column address to the CAS through the address pins.

For a write operation, data is sent through the Data In pin to the cell at the intersection of the addresses.

For a read operation, a sense amp measures the charge of the cell at the intersection. If it is charged, the sense amp indicates a 1 on the Data Out pin. If it is not charged, the sense amp indicates a 0 on the Data Out pin.

A memory cell can hold 1 bit of data, but a processor is designed to work with data in bytes. When the processor needs to store data in memory, it sends 8 bytes of data and an address through the system bus to the memory controller. The memory controller breaks each byte of data into 8 bits and uses the address that the processor sent to determine a row and column address. The memory controller sends each of the 8 bits to a different memory chip on the module. All 8 bits will have the same address, just on different chips.

When the processor requests data, the memory controller sends the same request to each chip, receives the bits of data, and reassembles them into the requested byte, as illustrated in Figure 4-8.

Figure 4-8. How the memory controller distributes data to the memory chip.