64-bit sounds magical doesn't it? Full of the promise of faster and more powerful computing. After all, 64 is two times 32, so there has got to be a performance increase right?

Well, yes and no. When we refer to 32-bit or 64-bit processors, what we are talking about is primarily the size of the registers, tiny, high-speed memory areas built right into the chip itself, closer even than the level 1 cache memory. You can think of them as the hands of the processor, in that they hold values to be operated on or combined.

As a vastly simplified example, say you want to add two integers together such as 2 + 3. Both values must be stored in separate registers until the adding operation can be performed, at which point the resulting value of 5 is deposited in a third register for transfer out. Any data to be processed must at some point end up in a register, therefore the number and size of the registers helps to determine the overall performance of the processor.

Ok, so 64-bit processors have 64-bit registers instead of 32, so they are faster, yes? Hmm, there is a catch though. When performing mathematical operations with integers (whole numbers), having a 64-bit register only helps performance if the integers being worked on are numbers larger than 32 bits, which is rather unlikely in current software. If you are adding 2 + 3 or 4,987,606 + 5,000,065 you are still going to need one register for each value and one for the answer, and the system doesn't care whether they are 32-bit or 64-bit registers, because you can only store one value in them at any given time.

When operating on floating point (decimal) numbers, a 64-bit register would seem to give added precision, as it can hold more numbers after the decimal place, but the x86 architecture, on which all current 32-bit Intel and AMD CPUs are based already provides for 64-bit floating point registers (actually 80 bits internally) so no advantage is gained there either. So what's the point? Well, there is one, and it's a good one. Memory.

To access an area in the computer's physical memory (RAM) to store or retrieve data, the processor needs the address of that location, which is an integer number representing one byte of memory storage.

Suddenly, having 64-bit registers makes sense as, while a 32-bit processor can access up to 4.3 billion memory addresses (232) for a total of about 4GB of physical memory, a 64-bit processor could conceivably access over 18 petabytes of physical memory.

This is the one area that clearly shows why 64-bit processors are the future of computing, as demanding applications such as databases have long been scraping on the 4GB memory ceiling, and although Microsoft and Intel have combined to enable servers using the 32-bit Xeon processor and certain versions of Windows 2003 Server to utilize more than 4GB of memory, the amount that can be accessed per-application is still less than 3GB.

If you are a business with a database of a terabyte or more of information, the 64-bit AMD Athlon64 processors look pretty good right now.