Billy Holmes posted <[EMAIL PROTECTED]>, excerpted below,  on
Wed, 07 Dec 2005 09:52:01 -0500:

> Conway S. Smith wrote:
>> assumption on this list, then you automatically get support for well
>> over 4GiB of memory (I can't remember exactly how much is supported, but
> 
> If we low ball it and just assume 48-bits are used to address memory, then
> that's
> 
> 256 Terabytes of RAM :)
> 
> I think we're safe for the time being.

The current AMD spec says the CPUs offer 40-bit physical memory
addressing, 48-bit virtual memory addressing.  So, 256 TB virtual memory,
but only a terabyte physical memory, in a flat-address configuration. 
However, given that terabyte disks are only now becoming feasible, and
thinking back to where memory was when gigabyte disks were at that stage,
we still have several year's worth of progress before a typical
configuration has to start worrying about that.  I'm sure AMD and Intel
will both have long addressed the problem by the time it appears,
increasing usable physical address bitcount to 48 or higher.  It's not
like there's another barrier of similar significance to 32-bit, out until
64-bit, which is far enough beyond current technology that it's no use
planning beyond that, because we don't know what the rest of technology
will look like by then, so we can't predict what the solution once that
point is reached, might be.

In the mean time, even a single TB of physical memory will be enough for
most uses, for several years yet.  In the extremes where it's not, the
techniques for non-flat memory are well developed from 32-bit land, and
can be used where needed.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman in
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html


-- 
[email protected] mailing list

Reply via email to