Most software that are available is not 64-bit yet. So there's not much benefit migrating to a 64-bit OS.

Just because some technology is released, and it's better, doesn't mean that everybody will go ga-ga and straightaway use it.

Software makers won't release 64-bit softwares yet since the demand is less from 32-bit ones, and most people are still using 32-bit.

Hardware device makers might release 64-bit drivers for their new and upcoming hardware, but they won't spend the resources porting the 32-bit drivers for their old and obsolete hardware to 64-bit. And people still are using old and obsolete hardware. Believe me.

As far as Windows is concerned, it might not support as many devices in 64-bit than in 32-bit, so this alone discouraged many to move on to 64-bit.

But 32-bit to 64-bit move _WILL_ happen. It's all in the matter of time. And as history shows, it will be progressively slow for economical reasons.

Like it happened from our 16-bit, segmented memory model (286 and below) to 32-bit, protected mode, flat memory model (386 and above)

The 32-bit technology we use now began in the 1980s. It might predate some of us here

---

Added on November 1, 2008, 10:03 pmCalculation performance has nothing to do with the OS, but the processor itself. And adding two 64-bit numbers will take more than 2 cycles, in any present processor.

The main difference is the size of the pointers. In 32-bit, pointers are 4 bytes, whereas in 64-bit, 8 bytes.
Another thing is, the general-purpose registers will have an extended 64-bit versions (RAX, RBX, RCX, RDX, ...)

What does this mean?

In a flat memory model, you could address 4,294,967,296 (2 ^ 32, or 4GB) memory locations in bytes, in 32-bit mode. That's why you hear things about 32-bit OS only supports up to 3GB ram, etc, etc. (3GB because your video card RAM has its place there, too. Not to mention your peripherals, USB, PCI cards, all will have its section of memory) It's not that you can't use your 2x2GB sticks in your computer running 32-bit OS, it's just that the remainder RAM can't be used by the OS.

For a 64-bit, (2 ^ 64, calculate it yourself) you could just imagine how much memory will it be able to "see." Windows XP 64-bit for instance, supports up to 128GB of RAM.

However, the pointers will take double the size in RAM (8 bytes instead of 4 bytes in 32-bit.) So the same application in 32-bit, when recompiled for 64-bit will take a bit more memory, and I have reasons to believe that it'll take a bit more disk space. But this is compensated by other 64-bit features (more general-purpose registers, etc.)

DOS was a 16-bit OS, but it supports up to 640KB of memory. It does this by using segmented memory model (10 pages of 64KB.)

But I feel we won't be using the segmentation model to enable access to more than 4GB RAM since 64-bit processors are available and they are affordable (in contrast to computers during the 1980-90s)



This post has been edited by Deslack: Nov 1 2008, 10:03 PM

If your motherboard supports it, I can't see why not.

Well, the system will run at the lowest speed available, anyway. So the 1066 will run at 667 max.

Hm, there's this factor called the Frontal Side Bus. Your RAM might be able to support higher speed but it's quite useless if the speed of the processor's access to RAM is not that high. If the processor's FSB is running at 667MHz, I think even using only 1066MHz RAM would gain no benefit over 667MHz RAM.

Unless... you're thinking of overclocking the processor and raise its FSB.

DDR667 supports up to 1333MHz quad-pumped Intel FSB.

I don't think you need to downclock the RAM. Leave it at 1066MHz.

But do stress-test your system, just in case. I could be wrong.