|
How does it work?
RAM stores information in a totally different manner than hard drives do, as would be reasonable to assume considering how much faster it is. In fact, at its base, RAM is no different from any other electronic chip in existence. It all starts out as an Integrated Circuit. In an IC, there are millions of memory cells, which consist of a single transistor and a single capacitor. This cell can either have the capacitor charged, giving it a value of 1, or have it be empty, giving it a value of 0. This represents a single bit of data. To add a bit (excuse the pun) of perspective, eight bits is equal to one byte. There are 1,073,741,824 bytes in a gigabyte. Therefore, your average killer gaming rig has more than eight billion memory cells in your RAM alone. Now, back to the topic. The capacitor in each cell drains over a period of a few microseconds so that a charge of “1” would drain to a charge of “0”, thus destroying the data. This is where, for the first of many times, the memory controller comes in to play. It recharges the drained capacitors thousands of times per second so that data isn’t lost while it’s in the middle of being processed. Memory cells are arranged in a two-dimensional plane, with columns that are called CAS and rows that are termed RAS. When RAM is operating, a charge is sent down a CAS to activate the transistors in every cell. Then the RAS either sends its own signal down the right row to charge the appropriate capacitor at the intersection, if it is writing, or it uses a signal amplifier to read what the charge is on the capacitor when it is reading. After it is done reading it recharges the capacitor to full ability. The time it takes for this entire process to be completed over all the cells in the chip is expressed in nanoseconds.

A graphical representation of a group of RAM cells currently being accessed.
Obviously, all these billions of cells would be worthless without some sort of system that kept them organized and kept everything running smoothly. Think of these circuits as the farmers and the memory cells are crops. They’re designed to take proper care of them. They identify every row and every column so that it knows where signals are supposed to go. They keep track of when the last refresh was for the charge-storing capacitors, and recharge accordingly. They amplify any signals read from the capacitors. Finally, they tell cell whether or not it should take a charge or not. And where do these farmer circuits get their information? Some of it is built in to the chip, but most of it comes from the memory controller, which is either part of the chipset, as it is in Intel and most AMD setups, or on the actual CPU, like with the Athlon64.
|
User Comments
- 2 Comments» This story has had 2 comments posted since April 23, 2005 at 10:39 PM EDT.