Synchronous dynamic random access memory (SDRAM) is made up of multiple arrays of single-bit storage sites arranged in a two-dimensional lattice structure formed by the intersection of individual rows (Word Lines) and columns (Bit Lines). These grid-like structures, called banks, provide an expandable memory space allowing the host control process and other system components with direct access to main system memory to temporarily write and read data to and from a centralized storage location.
When associated in groups of two (DDR), four (DDR2) or eight (DDR3), these banks form the next higher logical unit, known as a rank. 2GB DDR3 Dual Inline Memory Modules (DIMM) are undoubtedly the most popular density choice among today's enthusiast users. Most new parts of this type are configured as two identical ranks of eight banks each; one side of the DIMM housing those ICs that make up Rank 1, with Rank 2 populating the opposite face of the module. For this reason, single-sided DIMMs typically comprise only a single rank of addressable memory space.
Figure 1. Typical functional arrangement of SDRAM memory space. One Bank only is shown for clarityFigure 1 shows the typical functional arrangement of SDRAM memory space. In the case of our example dual-sided dual-rank unbuffered 2GB SDRAM DIMM, the fully populated module contains a total of 16 ICs, eight per side. Each IC contains eight banks of addressable memory space comprising 16K pages and 1K column address starting points with each column storing a single 8-bit word. This brings the total memory space to 128MB (16,384 rows/bank x 1,024 columns addresses/row x 1 byte/column address x 8 stacked banks) per IC. And since there are eight ICs per rank, Rank 1 is 1GB (128MB x 8 contiguous banks) in size, with the same for Rank 2, for a grand total of 2GB per module.
If each row contains 1K (1,024) column address staring points and each column stores 8 bits (1 byte), this would mean each row (page) is 8,192 bits (1,024 x 8 bits) or 1K bytes per bank. It's important to understand that each page of memory is segmented evenly across Bank n of each IC for the associated rank. For this reason, each page is in actuality 8KB (1KB x 8 contiguous banks) in size. So when we talk about IC density we are referring to eight distinct stacked banks and the total memory space therein, whereas when we talk about page space, we are really working with Bank n spread across the total number of ICs per rank. In the end the math comes out the same (8 ICs versus 8 banks), but conceptually it's a critical distinction worth acknowledging if we are to really grasp the ins and outs of memory addressing.
We can now see why the DDR3 core has a 8n-prefetch (where n refers to the number of banks per rank) as every read access to the memory requires a minimum of 64 bits (8 bytes) of data to be transferred. This is because each bank, of which there are eight for DDR3, fetches no less than 8 bits (1 byte) of data per read request - the equivalent of one column's worth of data. Whether or not the system actually makes use of all 8 bytes of transferred data is irrelevant. Any delivered data not actually requested can be safely disregarded as it's just a copy of what is still retained in memory.