microcode, compilers, and supercomputer architecture

From: Bill Yakowenko <yakowenk_at_cs.unc.edu>
Date: Wed Apr 7 22:09:14 1999

On Tue, 6 Apr 1999, ard_at_p850ug1.demon.co.uk (Tony Duell) wrote:
] > To put this slightly on topic, we all know that computer systems have
] > become more reliable in the last 20 years but does anyone have any real
]
] I am not totally convinced. Computers have become a lot more complicated
] in the last 20 years, and IMHO some of that complexity has led to
] unreliablity (cf the comparative crash rate of say Windows and CP/M).

I am totally unconvinced. :-)

The number of times per day that I run into something that doesn't
work has steadily gone up over the years. Years ago, although the
machine might crash from time to time, it would generally work pretty
well. Nowadays, even when the machine is "working", I have to wait
for the characters that I type to show up on my screen, and programs
that I run every day will often refuse to work because some server
that I've never even heard of is down or unreachable.

Prediction: between ten and twenty years from now, somebody will
discover "standalone" computing; that a computer can be made to
function even when not attached to a network. It will be a
revolution, paving the way for systems that keep working once
they are set up, even with evil sysadmins reconfiguring everything
they can get their hands on three times a week.

] Also, I have had _far_ more built-to-a-price PC parts land on my bench to
] be repaired than (say) DEC PDP11 and PDP8 parts. For all I have had to
] maintain many _more_ of the latter type of machine.
]
] -tony
Received on Wed Apr 07 1999 - 22:09:14 BST