🚀 go-pugleaf

In article <kalessinDFuBt1.66y@netcom.com>, kalessin@netcom.com (Adam Megacz) writes:
|> >Or another way to look at is economically.  Suppose in the future memory
|> >system prices drop to 1 cent per megabyte.  Then 64 bits worth of physical
|> >memory costs $160 billion.  Carrying it further, at .01 cents/megabyte it's
|> >still $1.6 billion.
|> >
|> >Forget your children's college tuition costs.  Start saving now for the
|> >memories of the future!

There are good reasons for having a lot bigger logical address space
than the possible physical address space.  Even 25 years ago, it was
nice having 31-bit addressing on machines with 21-bit physical spaces.
In the future, I can think of reasons why it might be nice to have
32 extra bits rather than 10 extra bits.  Might as well go to 128-bit
addresses once you decide to go beyond 64.  Space obviously isn't going
to be a problem when machines get that big, but, you hate to slow down
address arithmetic unless you really need to.

|> Yeah, but you can never have enough adress space - just look at how Intel 
|> botched the 8086 (or IBM botched by putting it in the PC). You can use all 
|> that extra adress space for memory mapped files, virtual memory, and mapping 
|> the I/O space into the memory space. I think that 64 bits will be enough for 
|> a *long time*, but we will outgrow 32 bits (4GB) before 2010.

Ahem.  You can buy machines today with 8 GBytes.  You won't have
to wait until 2010.  In 2010, you will probably be able to buy a
*PC* with 8 GBytes.  Recently, Cray demonstrated a database of
1.6 TeraBytes of actual data on a Cray Superserver (CS6400).




-- 
  Hugh LaMaster, M/S 233-18,    Email:       Please send ASCII documents to:
  NASA Ames Research Center     Internet:    lamaster@ames.arc.nasa.gov
  Moffett Field, CA 94035-1000  Or:          lamaster@george.arc.nasa.gov 
  Phone:  415/604-1056          Disclaimer:  Unofficial, personal *opinion*.