To ensure that it will win this sort of trademark
war, Intel purchased all of the existing ``* Inside'' trademarks. They have been serious about this issue for years.
Bull-Honeywell's GCOS machines still use 9-bit bytes. C was designed to run on these machines
(Kernighan's
Programming in C begins ``C is a computer language available on the GCOS and UNIX operating systems...''). The size of various types is intentionally left flexible to allow for these machines.
A 36-bit word on a machine with limited address space allows pointers to individual bits.
Those who do not know their own history are doomed to assume that it was lived only by `backward' peoples?
Anyhow, what little I've seen emphasis speed, but doesn't show the downside. That would appear to be the cost of converting *complex* applications, and the ability to coexist with other things.
Compatability is maintained through libraries that
implement POSIX, OS X, win32, etc. system calls.
Most applications that take advantage of the Exokernel features will just use predesigned
libraries for customized OS features, much like
replacing malloc with a specially tuned version.
Tux might be considered to be the Linux attempt
at doing what an Exokernel is intended to do.
The latest Inferno/Plan 9 newsletter mentions a port of the Inferno operating system to the iPAQ. If you pay for an Inferno
subscription, then you can get the port `for free'.
I believe that the source would be available to
you.
Vita Nuova was able to hack the iPAQ thoroughly; if you can't get the information that you want, you
could use Inferno source code as documentation or
the basis of your own hacks. If I wanted to
set up a wireless mobile network with custom
software on it, this would be the easiest
prepackaged solution of which I know. (I would rather use Erlang than Limbo for a distributed system, but it would take effort to get it on an iPAQ.)
Have you ever wondered why C uses octal? Or why Unix (and therefore chmod(1))
takes octal numbers for permissions? It's because C and Unix were initially
developed on 36-bit DEC machines. A 36-bit word has four 9-bit bytes, each
represented by three octal digits.
According to a
long list of computers, the
PDP-7 was an 18-bit machine and the PDP-11 was
16 bits. Not mentioned on the list, the GE 635
had a 36-bit word.
Even if they were using "commodity" PC parts, it would still be more than your bare bones
home PC, and I certainly couldn't put something together for under $500 (Fast CPU,GPU
and memory). Does anyone know how much a console really costs?
As mentioned in another reply, the machine will
sell for US$300. Actual cost of the hardware is
expected to be $500-$600, I believe. The developer machine will be sold at about hardware
cost, but will either play unsigned games or come
with software for signing games just for that machine.
Mass-producing identical hardware and not needing
to give profit to n levels of resellers
and shippers for each component lowers cost. The
GPU might be equivalent to a $300 card on its own.
I'm curious what the swap-in GPU upgrades will
cost.
At the last finals, the Regional Contest Directors for the contest were asked to vote on removing Pascal from the contest finals. The motion was not passed because many of the European teams are from countries in which it is not practical to run regional contest on modern machines. A posting for this article indicates that in the previous two years, St. Petersburg used Pascal and did well at the finals.
In my own region (East-Central North America, home of Waterloo, CMU and Toronto teams), teams using Pascal tend to do very poorly even though the problems are set with the limitations of Pascal in mind. Unfamiliarity with the region's supported versions of Pascal probably contributes to this problem, but I think that the main issue is that the teams choosing to use Pascal are less experience programmers than the other contestants.
I am not familiar with the advantages that Delphi provides, but I feel that teams who are extremely familiar with C++'s STL and Java's bignum libraries will have improved chances at winning.
Michael Van Biesbrouck, ECNA Regional Contest Director, 1999
If you write `AA->B' as `B->AA' then this problem becomes a fairly straight-forward parsing problem. You can use any general-purpose CFG algorithm for this without taking advantage of the nicer aspects of the problem specification.
I don't see why you have a circular order ordering on the letters.
The contest has been sponsored most generously by IBM for the past three years. Microsoft software was never required at the regionals (my region has used Unix for all of its contests in recent memory). When I went to Microsoft-sponsored finals (1996) the version of Pascal was from Borland. The following year the Pascal at the finals was a Microsoft beta product.
Michael Van Biesbrouck, ECNA Regional Contest Director 1999
Slashdot Mirror
To ensure that it will win this sort of trademark war, Intel purchased all of the existing ``* Inside'' trademarks. They have been serious about this issue for years.
Bull-Honeywell's GCOS machines still use 9-bit bytes. C was designed to run on these machines (Kernighan's Programming in C begins ``C is a computer language available on the GCOS and UNIX operating systems...''). The size of various types is intentionally left flexible to allow for these machines.
A 36-bit word on a machine with limited address space allows pointers to individual bits.
Those who do not know their own history are doomed to assume that it was lived only by `backward' peoples?
Most applications that take advantage of the Exokernel features will just use predesigned libraries for customized OS features, much like replacing malloc with a specially tuned version.
Tux might be considered to be the Linux attempt at doing what an Exokernel is intended to do.
The latest Inferno/Plan 9 newsletter mentions a port of the Inferno operating system to the iPAQ. If you pay for an Inferno subscription, then you can get the port `for free'. I believe that the source would be available to you.
Vita Nuova was able to hack the iPAQ thoroughly; if you can't get the information that you want, you could use Inferno source code as documentation or the basis of your own hacks. If I wanted to set up a wireless mobile network with custom software on it, this would be the easiest prepackaged solution of which I know. (I would rather use Erlang than Limbo for a distributed system, but it would take effort to get it on an iPAQ.)
According to a long list of computers, the PDP-7 was an 18-bit machine and the PDP-11 was 16 bits. Not mentioned on the list, the GE 635 had a 36-bit word.
Early history of Unix according to Ritchie.
As mentioned in another reply, the machine will sell for US$300. Actual cost of the hardware is expected to be $500-$600, I believe. The developer machine will be sold at about hardware cost, but will either play unsigned games or come with software for signing games just for that machine.
Mass-producing identical hardware and not needing to give profit to n levels of resellers and shippers for each component lowers cost. The GPU might be equivalent to a $300 card on its own. I'm curious what the swap-in GPU upgrades will cost.
Numbers come from memory, but originate with an audio presentation previously featured on Slashdot.
In my own region (East-Central North America, home of Waterloo, CMU and Toronto teams), teams using Pascal tend to do very poorly even though the problems are set with the limitations of Pascal in mind. Unfamiliarity with the region's supported versions of Pascal probably contributes to this problem, but I think that the main issue is that the teams choosing to use Pascal are less experience programmers than the other contestants.
I am not familiar with the advantages that Delphi provides, but I feel that teams who are extremely familiar with C++'s STL and Java's bignum libraries will have improved chances at winning.
Michael Van Biesbrouck, ECNA Regional Contest Director, 1999
I don't see why you have a circular order ordering on the letters.
Michael Van Biesbrouck
Michael Van Biesbrouck, ECNA Regional Contest Director 1999
Michael Van Biesbrouck, 1999 ECNA Regional Contest Director
Michael Van Biesbrouck, ECNA Regional Contest Director (1999)