Re: 64-bit

["Leif Svalgaard" <leif@xxxxxxxx>]

From: Vernon Hamberg <vhamberg@attbi.com>
> I'm wondering, seriously, where do we find the distinction? Single-level
> store uses 64-bit addressing, plus some extra bits, right? You discuss this
> in your eBook.
>
> But, in a way, EVERYTHING is handled with 16meg segments at the SLIC level.
> User spaces, a very handy construct in XPF, can automatically grow to that
> limit. Data files are built up from multiple dataspace segments. We don''t
> generally have access to assembler - level stuff (except as you describe
> using Display/Alter in SST)
>
> I don't think even the 400 is truly segment-free, although Storage
> Management hides the details. But the term segment turns up all over in
> PEX, with SAR (segment address resolution), etc.
>
> Even teraspace is probably allocated with requests to Storage Management
> for contiguous segments.
>
> Anyway...
>
> Back to Hollerith cards, eh?  :-)

===> well maybe not all that way back, but we should just not
fool ourselves into believing that we do 64-bit computing. We
do not, and the fact that the hardware can do it is that Tom points out
irrelevant and hidden. The only real gain we have is the 64+ bit
wide memory bus that makes access quicker for some operands
(clearly not for BIN(2) integers!).
One a 32-bit unix box each process can get a virtual 32-bit private
address space (IIRC) and we can talk about 32-bit computing.
To talk about 64-bit computing we by analogy would expect
each process to be able to get a 64-bit private address space.
But wait, isn't there a 64-bit Linux? and does it run on our beloved
box? or is that only the 32-bit version? So, when we follow
IBM's wishes and convert everything to Linux, we may be back
to 32-bit computing even on the AS/400....