Often memory is confused with storage as the names are similar. However, they are indeed different. The computer's memory is known as volatile, which means that when the power is turned off the data within the memory is erased. The computer's storage is non-volatile, when the power is turned off the data remains. Another difference is speed. Memory is solid-state circuitry. This means that there are no moving parts within memory. The computer's storage, or hard drive (sometimes known as a hard disk) is not solid state. This means that the memory is as fast as the electrical current inside of it whereas the storage is as fast as it's slowest moving part. Another fundamental difference is how the data is accessed. Memory is known as random-access (or RAM, which stands for Random-Access Memory) and storage is known as read-only (or ROM, which stands for Read-Only Memory). These concepts are important to know because otherwise they are easily confused.
There are many different types of memory, ranging from older RAM types such as FastPage or Extended Data Output (or EDO, for short) which are older memory types to Synchronous Dynamic RAM (or SDRAM, for short)and the exotic Double Data Rate SDRAM (or DDR, for short) and Rambus Direct RAM (or RDRAM, for short). All of these memory types are important. In order to understand them fully you must first understand how memory, or RAM, works.
When you use your computer, certain programs are running all of the time. From the operating system to drivers, these programs are constantly sending instructions to the CPU (for an explanation on the Central Processing Unit, or microprocessor, click here). Instead of running these programs from the much slower hard drive, they are stored in the system memory. This is a much faster area of storage. This speeds up the computer tremendously. Also, these programs are transferred back and forth between an area on the hard drive known as virtual memory (or the swap file, or the page file). This is where most of the confusion between storage and memory comes from. This file on the hard drive is usually much larger than the available system memory and is used when the computer runs out of memory for a desired application. For example, lets say you wanted to type a letter. In this example we will say you have 5 units of memory. Your computer's operating system uses 2 units of that memory, and we will say your drivers use another 1 unit.That leaves you with 2 free units of memory. When you type that letter it requires 3 units of memory to run. So, the operating system sends the drivers to the virtual memory, which frees up another unit of memory. You now have 3 units available for you to type that letter. This is obviuosly a very basic example but it demonstrates the point. Now, if in that same situation you had 7 units of memory in your computer it would have run much faster because the operating system would not have to send the drivers onto the hard drive's virtual memory file, making the whole computer much faster.
In older computers, some or all of the RAM was physically soldered right onto the motherboard because back then a whole lot of RAM was not neccesssary. In today's computers, wich have ultra-fast processors and programs requiring huge reserves of system resources (system resources can be loosely related to the amount of available free memory) to run, memory is critical in maintaining overall system speed. In today's computers, the RAM is mounted on the motherboard via slots which allows users to upgrade the amount of RAM inside the computer keeping their computer up-to-date. Each memory type has its own slot design which can make identification difficult when determining what RAM your computer has. Older computers with either EDO or FastPage RAM are not extremly upgradeable. The motherboards that these computers have usually will not take any more than 128MB of RAM, and not all motherboards can go up even that high. Another unfortunate attribute of these older memory types is scarcity and cost. Because the computer develops so fast older technologies are often discarded in order to usher newer, faster technologies. Because of this it is hard to find EDO or FastPage memory and then they are expensive because they are not manufactured extensivly. (if you think you have a computer with either EDO or FastPage memory, or any other type of memory and you wish to upgrade it, click here for the latest prices)
Newer memory types offer a better course when deciding to upgrade. Most modern computers since the Intel Pentium II (and some later Pentium Pro's and Pentium processors) will use SDRAM. SDRAM was initilized by Intel in order to keep RAM architecture up to speed with the increasing speeds of the microprocessor. SDRAM comes in three varities based on its speed: PC66 (or 66MHz), PC100 (or 100MHz) and PC133 (or 133MHz). This speed is actually the speed of the frontside-bus, or the data path between the memory, CPU and the video card (in newer systems). In systems with SDRAM it does not matter which SDRAM speed you buy as long as it is at least faster than your bus speed. For example, if you have an older computer with a 66MHz frontside-bus (sometimes refered to as FSB) then you can safely use (some motherboards do not support this but if they don't no damage will befall your computer)either PC100 or PC133 (you can also use PC66). Your computer will not gain any additional speed as the RAM will synchronize (hence the "S" in SDRAM) with whatever your FSB is.
Another very important aspect of memory is its speed. Obviously, the more memory the better, however speed is very important. From a pure numbers standpoint, RDRAM is the fastest. In its fastest trim, PC800, RDRAM has a theortical maximum data transfer rate (the rate at which the memory can move the data stored within it)of 3.2GB/sec. This is tremendous speed. However, it is not always what is on the paper that counts. Even with RDRAM's fast speed, it has a very narrow bus (a bus is a pathway between two or more components within a computer), it is only 16 bits wide. This means that in order to achieve fast data transfer rates it must be clocked very high. This disparity between bandwidth and speed results in a phenomonon known as "latency." Latency simply means that the data has to wait before it can be transfered because it has to be broken up into smaller 16 bit peices. The data is processed very quickly in RDRAM but because of the narrow bus, the data ends up waiting to be sent. To overcome this, RDRAM is a DDR technology. This means that it can send information on both the falling and rising edges of each clock cycle. Another DDR RAM technology is actually called DDR SDRAM, or DDR for short. DDR is a very fast RAM type because it has a large data bus of 64 bits and fast speeds because of its DDR technology. This means that it can achieve fast transfer rates without resorting to exotic, complicated and costly technologies, like RDRAM. In its fastest trim, DDR can achieve data transfer rates nearing 2.7GB/sec. Even though this is still 500MB/sec. slower than PC800 RDRAMs dual-channel trickery, PC2700 DDR is not dual channel and does not have the same latency effect as RDRAM. Both memory types are extremely fast as benchmarks have shown. Each also has certain pros and cons. RDRAM, with its extremly fast memory clock is very good at processing multimedia information quickly and DDR, with its fat pipe to the CPU and the GPU (for more information on the Graphics Processing Unit, click here) is very good at processing information that has to be utilized very quickly, such as rendering high framerates (a framerate is typically used to describe how fast the graphics card can process graphical information and then send it to the monitor to be drawn onto the screen, the more framerates, the smoother the image appears) during a videogame. I am not here to tell which is better as I am not reviewing these different technologies. They are simply different from each other. However, RDRAM is costly as it has to be bought and installed in pairs, which means that you have to have at least two or four sockets filled at a time with RIMMs (Rambus Inline Memory Modules, which refers to the actual stick of RAM, itself)and then you have to have devices known as continuity modules installed in the remaining sockets. Continuity modules are RIMM-shaped devices that are pluged into the RIMM sockets in order to keep the electrical current flowing. Why? Unlike other memory types, RDRAM is configured like your christmass lights; if one bulb burns out, the whole strand does not work. Rambus, Inc. is also having problems finacialy, as it is involved in a variety of law suits involving patent and licenseing infringement. Rambus, Inc. also charges royalties to memory manufactures that wish to manufacture it which raises the price of it above other memory technologies. DDR, for example, is simply an enhanced and much faster version of SDRAM and requires no exotic manufacturing techniques, is generally available to memory manufactures and does not have the heat issues that RDRAM does.
