der Mouse wrote:
>> But is not LIFE on the molecular level some thing like a bunch
>> computing machines that computes structure rather than logic.
>
> Something like, maybe. But since chemical interactions involve
> randomness randomness coming from the underlying quantum mechanics and
> thus being, as far as we can tell, true randomness...nothing
> deterministic is an accurate model.
To make another semi-random connection here, von Neumann was
very into the emergent behavior of simple automata known as
cellular automata. IIRC he was the first to prove that there
existed self-replicating CAs.
> And that includes nondeterminstic Turing machines, as I studied them,
> since they don't pick one possible transition randomly but instead take
> all possible transitions (which is why the power set construction works
> to determinize such a machine).
Interestingly there does exist a model called a probabilistic automata
that operates in just that way. In fact, if you define what accept
means the right way for the model, it can be shown that there are
non-Turing
languages that can be accepted by a PA.
Brian L. Stuart
Hello all,
I recently got a bare Wavemate Bullet board, and would like to get it up and
running. I received some 8" floppies with it, and they appear to be boot
disks for the board. All but one was readable on another CP/M system.
I'd like to get docs for it, or pointers to more information on the board.
It's a Z-80 based board, and has the CPU, DART, CTC, DMA, and PIO chips.
Looks like it has some RAM (16 MB8264), and some glue logic, plus a
Fairchild MB8877.
It is labeled "Wavemate (c) 1981 Rev. B Hawthorne, CA. Made in U.S.A."
There are two 10-pin headers marked "J3" and "J4", two 50-pin headers marked
"J1" and "J6", and two 34-pin headers marked "J2" and "J7". I suspect J3
and J4 are serial I/O, as the traces go to a set of 1488/1489 chips. J1 and
J2 are next to each other on the right of the board, and J7 and J6 are next
to each other on the top of the board. There's also another 34-pin header
near the top right marked "J5". There's an 8-position DIP switch labeled
"SW1". Power appears to be from a plug that loks like a typical PC-style
drive power connector. I'd like to verify the pinout, though, if possible
:-)
Curiously, there's no EPROM, nor even a socket for one. My luck, it's some
missing part :-(
I'd appreciate any help!
Thanks,
Rich B.
I have an 8" drive that I intend to connect to a catweasel card for the
purpose of getting the bits off of a lot of old PT-DOS disks for my Sol
computer archive project. If I can accomplish that, then I'll extend my
Sol emulator to allow running PT-DOS on the emulator. Supposedly the
floppies have a lot of interesting stuff on them, including source code
to a lot of PT software.
After an hour of googling, I am unable to find what power to feed it.
The connector on the back is two 3-pin molex connectors and looks like this:
+-------+
| X X X |
+-------+
| X X |
+-------+
Yes, the middle one of the lower pair is missing. I assume the power is
+5, -5, +24, and GND, maybe +12 or -12 too. But guessing isn't good enough.
Does anybody know for sure what power to feed each pin? I have found
information on how to build a 34-pin to 50 pin cable to mate the 8"
drive to the catweasel, so I think I have what I need there.
Thanks for any help.
re: Infinity UC-1800
I'm reasonably sure it's a "trainer", eg. an RCA 1802 in a box, hex
keypad and switches and LED display with which to poke in "demo" toy
programs. It seems to operate, inscrutably. (Here you can see I try to
enter an 8080 CALL instruction to no avail) (kidding). It has an
"expansion" connector available in the back. There was a leaking NiCd
battery pack I clipped out and tossed, but otherwise it's complete (sans
and accessories it came with).
Anyone have any documentation?
http://wps.com/temp/inf1.jpg and
http://wps.com/temp/inf2.jpg
While there have been other follow-ups to this thread, I'm picking
this one as the point I jump in. This is a bit long and for those
of you who don't see a point in this sort of "angels on the head of
a pin" discussion feel free to delete. If you want to criticize
me for engaging in such a discussion because there's no ROI there,
then you can just kiss my big, white... :-) Sorry, too much bean
counter frustration at work lately. Another way of looking at this
message is that it was suggested that a professor be asked. Well,
be careful what you ask for, you might get it:-)
>> There are several architectures for computers, with all being the
>> equal of the Turing machine.
>
> That's interesting, because not one of the devices I have called
> computers, used as computers, or heard/seen called computers, is even
> theoretically equivalent to a Turing machine. In particular, they all
> have finite storage.
It is indeed correct that in an absolute sense physical computers
can be modeled by finite automata and don't need a Turing machine
to model them. And it's correct that the reason for this is that
the physical computer is finite and a Turing machine's tape is
unbounded. (Almost no authors highlight the distinction between
unbounded and infinite. But keep in mind that for any recursive
function, the Turing machine must finish in a finite amount of
time and therefore can only use a finite amount of tape. However,
there is no a priori bound on that finite size.)
>> Frankly, the Turing machine is the definition of a computer,
>
> I don't know where you got _that_, but it certainly doesn't match my
> understanding, usage, or experience of others' ditto. As I remarked
> above, there aren't any computers by that definition.
Now here I do agree with the desire to define a computer in terms
of machines that can compute functions that are computable in a
Church-Turing sense. Of course in doing so we have to keep in
mind that we are only looking at bounded approximations. In other
words, there are some recursive functions that we can never
completely implement with a finite physical computer.
So if physical realizations of the computing ideal will always
come up short of the models of the Turning machine or the lambda
calculus, then why do we ever study these models and why do we
make distinctions between recursive and recursivly enumerable?
As I see it, the answer lies in the enormous size of the finite
automaton that a physical computer realizes. Take even a small
(by today's standards) machine with 1MB of memory. Ignoring
secondary storage, the machine has 2^8388608 states. Since there
are only about 2^34 nano-seconds per year, it would take 2^8388574
years to go through all the states at a rate of one new state
per nano-second. Now, of course, one doesn't have to visit all
of the states of a finite automaton to recognize an input string.
But this observation suggests that regular languages make a poor
model to describe the physical computers we build. Add to that
the fact that the way that deterministic models (like RAM or
deterministic TMs) handle those regular languages most naturally
reognized by non-deterministic FAs is by exploding the number
of states in a similar way. So the bottom line is that the
regular expression model does a poor job of describing the
types of interesting problems we use computers for.
Now if we flip the perspective and ask about what interesting
things can be done on a Turing machine, we find a similar issue.
In order for the answer to do us any good, we must get it in
our lifetime which implies that there is an upper bound on
the size of the tape we will use in an interesting computation.
Of course, this gets us right back to the same issues we have
with the physical computers. Once I set an a priori bound,
the set of languages recognizable is a subset of the regular
languages.
But once I say that, then all of the structure in the language
hierarchy collapses. However, we view the question of Church-
Turing computability and the questions of NP-Completeness as
something more than just mental self-gratification. Therefore
(you knew I had to get there eventually), the Turing machine
is a powerful model to describe physical computers just as
it's a powerful model to assess the limits of computation in
the abstract. Furthermore, it's a more useful model than
those that are actually theoretically equivalent.
Those who would argue that I've shifted the discussion away
>from the semi-original issue of defining a computer have a
good point. Just because the Turing machine is one of the
best ways to abstractly model the physical computer doesn't
necessarily mean that it should define said computer. So
how would I attempt to couch the definition of a physical
computer in terms of a Turing machine? I'd say that if you
can implement a simulation of a universal Turing machine
so that the range of it's computations is limited only by
the amount of addressable memory, then you have a real
computer. I am aware that Cohen defines the term computer
in a more abstract sense and I understand why he does so.
(By the way, his is one of my favorite books on the subject.)
And if I'm wearing my theoretician's hat, then I'm likely
to use the terminology the same way. But when I'm wearing
my engineer's hat and am appreciating the design of classic
hardware and when I'm wearing my historian's hat and am
trying to classify early machines, then I'm more likely
to use a definition like I've suggested above.
If you've stayed with me this long, you have my condolances.
However, there's one more interesting point that comes out
here. This sub-thread was sparked by the question of whether
the potential of self-modifying code was necessary in order
to be a computer. Notice that with the definition above,
any computer can implement a universal Turing machine and
unless you engage in some unnatural limitations, such a
machine can modify it's own programming. (Remember that
a universal Turing machine interprets instructions read
>from the tape.) A more prosaic way of looking at this same
point is that I don't need a von Neumann architecture to
implement an interpreter which interprets self-modifying
code. And if computing theory teaches us anything, adding
or removing levels of abstraction doesn't affect the basic
power of the model.
My apologies for being so long-winded.
Brian L. Stuart
I had no idea they'd be so popular...
I live in Los ANgeles, the Cosmac jobbies are in Los Alamos NM, about
1000 miles from me. However, I go there twice a year anyways, so I will
buy them all, drag them home and email the list. I don't have any great
interest in them, and I'll only want to recover costs.
I may have taken photos of them, if so, I'll contact a local (to NM)
friend and see if they can pick them up and ship, etc.
Don't hold your breath, it will take a while.
being an engineer that worked in assembly language on many micros
before and after the 8086, the segmented architecture was not
that hard to handle and actualy had many side benefits such as
code size and more efficient use of the bus bandwidth. The pdp11
may have been better overall, but there is no comparison on terms
of price, availability, and being able to get the job done for
the millions of PC users.
best regards, Steve Thatcher
>--- Original Message ---
>From: "Eric Smith" <eric(a)brouhaha.com>
>To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk(a)classiccmp.org>
>Date: 11/13/03 3:24:07 PM
>
"Hans Franke" <hans.franke(a)mch20.sbs.de> wrote:
>> to me it's the way the memory is handled
>> that makes the 8086 the great CPU it is
>
>What, the 64K segments that alias on paragraph boundaries?
>Yecch! What a kludge! The PDP-11 had better memory management
>for a 64KB address space at least seven years earlier.
>
>
I do have a Rainbow 100 home page, but I haven't updated in a while.
Things have been extremely busy at home this past summer, but I will try
to update it more often now that I'm settled into my new house. I always
love any extra information. BTW, who scanned all those Rainbow Tech Docs
now available at ftp.update.uu.se? That was quite the task, and I was
glad to find them available!
Oh, and of course, the Rainbow 100 web page is at:
http://www.classiccmp.org/rainbow/
-Jeff
jba(a)sdf.lonestar.org
SDF Public Access UNIX System - http://sdf.lonestar.org
Whilst cleaning out a closet I found the following
items that may be of interest to people on this
list...
1. A pair of 8-bit (XT) Corvus Omninet Transporter
Cards. One is the older LONG card, and the other is a
Gate Array Card.
2. A Box of Miscellaneous Tops Hardware and Software.
3. Two copies of Lantastic Client Software, Complete.
I can also supply some 16-Bit NE-2000 Compatible
10megabit Ethernet Cards to go with these.
4. A box full of SCSI to Ethernet Adapters. Useful to
put SCSI Devices (like a Printer or an Old Mac) on an
Ethernet Network. I have Powersupplies for these, but
no SCSI Cables.
If interested, drop me a line.
Not looking for much. Hate to throw this stuff out,
especially if someone else has some use for it.
Al
__________________________________
Do you Yahoo!?
Free Pop-Up Blocker - Get it now
http://companion.yahoo.com/
> I intend to connect to a catweasel card for the
> purpose of getting the bits off of a lot of old PT-DOS disks
Aren't SOLOS discs hard-sectored?
The 848 user's manual is at www.bitsavers.org/pdf/tandon
