The Future of Memory

Without memory, all the computation, all the cool graphic cards and dongles in the world would be rendered entirely useless. This article provides an overview of trends in memory—and some future developments that just might change computer memory as we know it entirely.

A quick refresher before we begin. Computer memory is written in binary code, that is, in a series of 1s and 0s. Each of these digits is referred to as a binary digit, or bit. Memory can be stored on any physical system that can be easily turned from one easy-to-differentiate state to another, or a bistate, with one state representing 1s and one state representing 0s. So, the medium through which memory is stored can vary a lot.

From here, there's one main division in types of memory: volatile and non-volatile.

Volatile memory is memory that requires power to retain stored information. Volatile memory is vastly more expensive than its counterpart, but much quicker to access, hence its use as temporary memory. It will generally consist of a series of extremely small electronic switches that can be turned on and off, usually with capacitors and transistors.

Non-volatile memory does not require power to sustain it. This tends to make up most of permanent memory, which will use more physical ways to store memory, such as pitted plastic disks or tape with magnetic coating. Non-volatile memory tends to be much cheaper, so it is used whenever speed is not of the essence.

Essentially, the trend with memory can be summarized as such: more memory, smaller size, cheaper prices. That's it, no matter what type of memory you're speaking of. This simple trend does not look like it'll be stopping anytime soon, either. There's a sort of positive feedback cycle involved: as more memory becomes more cheaply available, programs become more memory intensive, which in turn requires more memory, so more memory is made more cheaply available... and so on and so forth. Because the turnover rate is so quick, even last month's products can be incredibly cheap: a mere few weeks seeing the cutting edge of memory and storage speeds, power use, and size, move forward. We've a ways to go before hitting any physical ceiling that'll put a stop on memory development.

No matter what memory storage device you bring to mind, from flash drives to computer hard drives, memory capacity has been increasing, often doubling over the course of a mere few months. Whole terabytes are now available in external hard drives. 64-bit architecture means that there is no 4GB limit to RAM, and computers that take advantage of this are growing in popularity.

The physical size required for such devices is decreasing as well. Those multi-terabyte hard drives won't be much bigger than a slice of toast or two, and they're only set to continue shrinking. Mostly, the increased capacity and smaller size are made possible by increasing the density of memory. Physically, this is due to ever improving techniques to put the electronics involved closer and closer together. Without new advancements in the field, a practical barrier will at some point be met with regards to increasing memory density even further—we just don't know where yet. Think Moore's Law, which is having more problems with economic rather than electronic issues.

Everyone needs memory. Despite demand being so high, manufacturers are more than meeting the requirements. While the very cutting edge of memory will always be a bit pricey, even just buying a few weeks after any given product has entered the product can result in dramatically reduced prices.

Of course, this is all working off of more of the same. Some of the technologies becoming available for use in memory promise to revolutionize the field considerably, once they make it into the commercial mainstream market.

One such example is an amazing terabyte-per-square-inch permanent memory, coaxed into existence into latter 2008 at the Max Planck Institute of Microstructure Physics. Essentially, they created incredibly tiny platinum capacitors on a ceramic surface, which isn't all that different from current technology—just a whole lot smaller. Whether this is commercially viable or not remains to be seen, but a terabyte per square inch is going to be tempting for any computer user.

Holographic memory is another possibility. Seems like out of Star Trek? Well, everyone gets their ideas from somewhere. Essentially, holographic memory functions by having two beams of light intersect, leaving an interference pattern along a plane. By adjusting these beams of light, you can create slightly different interference patterns, which can be the medium through which data is store. In a 3-D object—such as a holographic crystal—there can be thousands or millions of such planes by only slightly adjusting the angles of the light for the different interference data, and thus, memory. Still not commercially viable, but the technology is there, waiting to be exploited.

Biological memory is another possibility, though at this point only a theoretical one written about in sci fi novels such as Blood Music. Storing data at a molecular or even atomic level, such as within complex proteins, promises to store more data per square inch than anything else we've been able to manage, and in a remarkably stable and easy-to-access form. Researchers have been able to manage about 800 MB-per-square-inch using this molecular memory thus far, but greater memory density is a sure promise for the future—though widespread commercial use might take a bit longer.

Most of these outlined technologies require the existence of some sort of individual dot for each bit of data. Why not have them placed on a line—or rather, on a super-speed racetrack? Researchers have just managed to use billions of tiny nanowires to quickly transfer and store data, and this is just in two dimensions for the moment. It's volatile, meaning that it will require power to retain at all times, but it promises to use less than current volatile memory devices.