Question

I am reading some C text at the address:

https://cs.senecac.on.ca/~lczegel/BTP100/pages/content/compu.html

In the section: Addressible Memory they say that “The maximum size of addressable primary memory depends upon the size of the address registers.“

I do not understand why is that.

Can anyone give me a clear explanation, please?

Thanks a lot.

Answer 1

If you have 32-bit registers, then the highest address you can store in a single register is 2^32-1, so you can address 2^32 units (in modern computers, units are almost always bytes). A larger number simply won’t fit.

You can get around this by using memory addresses that are larger than a single register can hold (and some CPUs/operating systems have features for doing so), but using addresses/pointers will be slower because it has to fiddle with multiple registers.

As an example, suppose you have 32-bit registers but 64-bit pointers and want to increment a pointer to find the next item in an array of char (++p). Instead of performing a simple increment instruction, the processor will have to

1. Increment the lower 32 bits;
2. check if the result is zero (overflow);
3. increment the upper half as well if overflow occurred.

Simplifying a bit, this means it has to perform a branch (if-then-else) instruction, which is one of the slowest and most complex instructions a modern CPU performs.

(See, e.g., x86 memory segmentation on the Wikipedia for a multi-register addressing scheme used in Intel processors.)