On Wed, Jan 05, 2000 at 03:37:06PM -0600, Bill Wendling wrote:
> Also sprach Dominik Kubla:
> } On Wed, Jan 05, 2000 at 03:32:53PM -0500, Richard B. Johnson wrote:
> } ...
> } > I think that before 38 years is up, none of us will be using 32-bit
> } > machines so, even if I was 10 years old, I wouldn't bother 'fixing'
> } > 32-bit machines. Even Intel's new stuff is 64 bits.
...
> I'll put my foot squarly in my mouth and predict that, in 38 years,
> people will think that running anything on a machine less than 1GHz speed
> for a production machine is nothing short of insanity.
No, the range of Linux things is wider than web-servers
or computing farms (or workstations). I myself would not
put 1GHz processor on machine which would then be sitting
idle for 99.5% of its time -- or more importantly, if 64-bit
1GHz processor eats 30 Watts of power, while a 32-bit one at
100MHz eats 1 Watts, my energy constrained applications would
definitely go for the 1 Watt approach.
CMOS power consumption is dependent roughly linearly with the
number of switching elements (double from 32-bit to 64-bit),
and clock speed difference (100 MHz to 1000 MHz). Add there
that very high speed chips will likely need a large number
of clock distribution circuitry which slower ones don't need,
and you rather trivially get that factor 30 ...
(I am apparently quoting roughly what Digital described about
motivations behind the StrongARM developments some years ago.)
So yes, while 64-bits and perhaps even 128 bits will appear
sometime into the mainstream, the 32-bit platform will not
disappear for quite a while.
What I would like to see, though, is that when ia64 goes out,
its *native* syscall APIs will use all possible scalar fields
which now are 32-bits OR LESS at width of 64-bit.
The main reason (which I applaud IBM doing with their 64-bit AIX)
is that CHANGES TO APIS IN THE FUTURE ARE ALWAYS PAINFULL!
Even if you don't today figure a reason for doing 64-bit value
space for something, it doesn't mean such won't appear a month
from now. (Like why UIDs there should be 64-bit when at no
other platform they are anything but 32 ?)
> But, then, there are always the hobbyists...but, then, they can "fix" the
> kernel themselves? :)
>
> Bill
/Matti Aarnio <matti.aarnio@sonera.fi>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Jan 07 2000 - 21:00:04 EST