The 64-bit Question
64-bit and 32-bit computing compared
If you have either bought or have considered buying a new computer over the past couple of years, you have probably been faced with the question of whether to go for 64-bit hardware or stick with the old-fashioned 32-bit variety. Even if most of the computers you look at now have 64-bit hardware, you are still faced with the question of whether to use a 32-bit operating system (i.e. Windows) or a 64-bit OS.
So what is the advantage of the extra 32-bits? When computers were first available they used eight bit microprocessors. This means that the CPU dealt with data in eight but chunks. By the time the original IBM PC came out, 16 bit CPUs were cost-effective and this became the standard for business computing. But as computer users became more demanding, needing larger amounts of memory, it was obvious that 16 bit registers made it very difficult to write software that could handle this memory. As if it wasn't difficult enough to write reliable software, developers had to use all kinds of complicated trickery, and 32-bit computing was the obvious answer to this problem. But the transition was a painful one for the software industry.
If moving from 16- to 32-bit was a no-brainer, the advantages of moving from 32-bit or 64-bit are far less obvious. You really have to have programs addressing very large amounts of memory, and if you have less than 4Gb of memory there isn't really any advantage. Servers often have large amounts of memory and lots of processes running to use this memory. This means that while 64-bit operating systems have become the norm for servers, for desktop computing it is far less clear cut.
Just to be clear, there is no reason why you can't have 64-bit hardware and run a 32-bit operating system. That will give you the maximum flexibility to upgrade to a 64-bit later if this becomes necessary. And although the situation has improved, it is certainly difficult to find 64-bit device drivers for older hardware devices. That might be a deal-breaker if you are relying on some old printer or scanner or storage device.
But even if you have 64-bit hardware and have decided to use the 64-bit version of the operating system, you can still use 32-bit versions of applications. This is because the 64-bit version of Windows has the ability to run 32-bit applications (that is user software rather than low-level software like device drivers). This works in the same way that you used to be able to run 16-bit applications on 32-bit Windows.
So you still have the question of whether to install 64-bit versions of your applications, if they are available. There is, perhaps suprisingly, a price to be paid for shuffling around 64-bit data, which results in slightly poorer performance for 64-bit applications until the point at which the benefits of the extra addressing power kick in. At the moment the only software that can really benefit from this is something like a spreadsheet where you have very large worksheets, or perhaps image Processing Software that you were working with very large images. In the majority of cases running the 64-bit version of an application is nnecessary, and those extra bits come at a price.
So the conclusion is: if you are buying hardware - make sure it is 64-bit. If it comes with a 64-bit OS, good (unless you have some old and weird devices you need to use, in which case check that 64-bit drivers are available). Don't bother to upgrade an existing computer to a 64-bit OS. For applications, install the 32-bit version unless you are dealing with huge spreadsheets or images.