If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

What is computer memory?

What is the limit of computer memory? Created by Brit Cruise.

Want to join the conversation?

Video transcript

Voiceover: When we perform calculations with a pen and paper, we often need to save intermediate results. And we may do this with, say, scrap paper, and in this case, the paper is acting as a form of external memory. And memory no matter the form, takes up physical space. Computers contain memory, we can think of it as the scrap paper for the computer. And, say, when you construct an array to store values in your program, you require memory. And, at the lowest level, computers read and store all instructions as a string of numbers. But, how do you store numbers in a machine? This was a very difficult problem originally, especially when you need computers to hold their memory after the access to power is lost. This is known as nonvolatile memory. The easiest difference for a machine to detect is simply a presence versus an absence of something. And this is how old punch cards would work. Along the top, we have some data and the vertical columns contain a series of punched holes which represent each character. So, computers really have 2 fingers, base 2, same as a light switch being "on" for 1, and "off" for 0. This is the smallest amount of information, a single difference, which we call a bit. But bits are powerful for storage because the amount of unique states grows exponentially as we add bits together. Remember, one light switch is one bit and it can store 2 states, but 2 light switches can store 4 unique states. And 8 light switches or 8 bits can store 256 unique states. And space is measured in bits, but the physical size of a bit depends on your method of storage. So how do computer store zero's and one's internally? (gentle music) Man in Uniform: Modern data processing systems like these use thousands of magnetic cores. What are magnetic cores? They are tiny rings of nickel alloy or other magnetic materials. They have replaced vacuum tubes for many important functions in data processing systems. Voiceover: And it allowed computers to store bits as clockwise versus counter-clockwise magnetization direction. Because the each core could be magnetized in 2 different ways, depending which direction the current was applied. Man in Uniform: Because a bit can be represented by any bi-stable device and a magnetic core is a bi-stable device. Voiceover: Later on, this was done using thin film magnetic disk where we can think of as each bit as a tiny magnetic cell, which can be charged to store either a 1 or a 0. So, long story short, the size of a bit has been rapidly shrinking since the days of punch cards. A hard drive in a modern computer can be thought of as billions of tiny magnetic cells. Now, you may wonder, well how small can these little magnetic cells be? And current research at IBM is pushing this to the atomic level where they have shown 12 iron atoms can work together as a stable magnetic unit, where they are able to store a 1 or a 0, depending how they are oriented. And this is approaching a theoretical limit where we would hold a single bit on a single atom! And interestingly, IBM estimates that we can put around one quadrillion bits of information in a handheld device, the size of an Ipod, with atomic storage. And, let's call this a super drive, it doesn't even exists yet, as a hypothetical example. A small handheld super drive using atomic storage would hold one thousand terabits, which is one thousand trillion switches or more commonly known as 125 terabytes in the palm of your hand, or to use an example everyone can understand, 125 terabytes is the same as having a 1250 kilometer long book shelf in the palm of your hand. And this is what the future of memory looks like, or we ever be able to store a bit on something smaller than an atom?