32 bit OS limitations

Joe

I am curious about the limitation of 32bit architectures. From what I understand, a 32 digit (bits) binary number can only go up to 2^32 = 4 294 967 296. It is then easy to figure outthat if a 32 bit processor uses 32bits to store data address in memory, it can only go as high as 4.2 billion which is where the 4GB limitation comes from.

Now here's my problem. I come from a C environment and I know that such a problem is usually fixed by encoding the address on multiple adjascent '32bit-spaces'. So I'm guessing I have the whole thing figured out wrong.

Can anyone explain (or at least point me in the direction of a post/article/tutorial that does) the limitations of 32 bit operating systems as well as the actual need for 64 bits, except that we're all attracted to shiny high numbers?

If anyone also cares to comment about the use of 32 bit OS on 64 bit proc and whether there's a loss in performance.

Thanks ;)

Kassem

Yeah actually I'd like to know more about this too. I am not really good when it comes to hardware and all this bit/bytes and binary stuff (I know the basics though). But I've read in a blog post that some software which come in 32-bit and 64-bit have huge differences in performance, like the Photoshop 64-bits is much faster and more efficient than the 32-bit counterpart. So yeah, any info about the subject would be appreciated.

P.S: everything before Dual Core was 32-bits and now almost every processor is 64-bits, is that right? Also, what makes it different is the number of pins the processor has? (that's what my "Introduction to Computing" professor said but I cannot take her words for granted because she's literally stupid)

nuclearcat

There is no significant performance difference. 64-bit have more cache/memory trashing (64-bit long and pointers), but faster 64-bit math (also even atomic?) and etc.
32-bit can address up to 64GB over PAE, but with insignificant performance drop.

Significant for me and for changing to 64 is problem to address for single process more than 2-3 GB or memory (usually userspace have 3GB max adressable, depends on OS).

kareem_nasser

64 bits can take or address up to 2^64 of physical RAM but as for performance it is the Application or OS that should be optimized to such architectures in order to work more efficiently and in our native CPU usage we wont notice this much performance difference but in Matrix and other astronomical or mathematical equations the performance is of course better. The PS2 had a 128-Bit CPU or it is marketed this way for unknown reasons becasue no game uses this but it could be used for guided missiles.(Like Sony did with Psones GPU it had some unknown specs that were not released to the public check this.

The custom instruction set consists of 107 instructions for operating on four 32-bit, eight 16-bit or sixteen 8-bit integers simultaneously. Instructions defined include: add, subtract, multiply, divide, min/max, shift, logical, leading-zero count, 128-bit load/store and 256-bit to 128-bit funnel shift in addition to some not described by Sony for competitive reasons.