Re: 32bits and 3G memory limit
On 06/21/2010 10:09 AM, Aaron Toponce wrote:
On 6/20/2010 10:32 AM, T o n g wrote:
Is the 3G memory access limit is the natural one, or something
superficial imposed by M$? I mean, does 32bits Linux (the i386
architecture) has such 3G limit as well?
I'm not familiar with "M$", but if you're referring to "MS", short for
"Microsoft", then no. Sorry, pet peeve, but "M$", "Micro$oft", and the
like, come across as quite immature, and very fanboy.
Very 1990s, I'd say.
A 32-bit system, is a system that can address at most 2^32 bits of
memory for any given process.
The traditional definition of a CPU's "bitness" is the width of it's
registers, not the amount memory it can access.
Most 32-bit kernels these days, however,
can address much more thanks to the physical address extensions (PAE),
typically 64 GB. But that still means that each process can only address
2^32, or 4GB of RAM.
Because "address lines" are not the same as "data lines". (Unless
they are multiplexed, but conceptually they are still different.
Anyway, engineers only do that when die space is more important that
performance, and that hasn't been the case on the "desktop" is 20
Seek truth from facts.