> Date: Wed, 10 Jan 2001 17:05:43 -0800
> From: Larry Dwyer <yy@xxxxxxxxxx>(by way of Larry Dwyer <yy@xxxxxxxxxx>)
>
> Based on the transition from 16 to 32 bits and the transition from 32 to 64
> bits, I predict that the transition to 128 bits will happen approximately
> 14 years from the first availability of a commercial UNIX platform that was
> 64 bits (Alpha from Digital). I believe this will be about 2009.
It sounds like you've made the common mistake of assuming that, since
32-bit Unix hosts showed up in force around 1978 and 64-bit hosts
around 15 years later, then 128-bit machines will start to arrive
around 15 years after that. But, even assuming continuing exponential
growth in machine capability, you're off by a factor of two: if it
took us 15 years to for us to outgrow a the 16->32 increase, then it
should take us *30* years for us to outgrow the 32->64 increase. This
would be around 2025.
For various other reasons (including Moore's Second Law), even 2025 is
a quite optimistic date unless you're assuming some sort of unexpected
software/hardware revolution -- and if so, please tell us more!
Hmmm, I see you published a similar estimate in 1996, in:
<http://www.uniforum.org/news/html/publications/ufm/jul96/infotech.html>
Hmm, nobody caught the error back then? For a hopefully-clearer
explanation of where you went astray, please see John Mashey's 1995
news article reprinted in <http://yarchive.net/comp/128bit.html>, in
particular his comment:
"Note however, that the common assumption that it took N years to go
from 32->64 means that it would take N years from 64->128 ... is
incompatible with the normal memory progress, i.e., 64->128 is 2N."
> Assume the transition to 128 bits will happen (our engineers are already
> considering how we will weather this transition).
Nice to hear from someone taking the loooong view.
|