Searching \ for '[OT] Multiple Dimensional Processors. (was [OT]Sol' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=
Search entire site for: 'Multiple Dimensional Processors. (was [OT]Sol'.

Exact match. Not showing close matches.
PICList Thread
'[OT] Multiple Dimensional Processors. (was [OT]Sol'
1999\03\06@200714 by Wagner Lipnharski

picon face
Gerhard Fiedler wrote:
> i guess the question is not whether to plot it inside out or from the
> outside, it is how to project a sphere's surface on a flat surface. since
> this can't be done without =some= distortion, the question is =what= to
> distort... in doubt, use a globe. usually from the outside :)
> ge

You see the problem?

If you are in the middle of the globe, it is not distorted in any way,
any direction you look has the same distance to the surface (well, more
or less), so you don't need to plot it in a flat surface, and this is an
error done long ago when it was difficult to think in tri-dimensional
plotting.  It is the same if you think about a 30 x 90¡ cinema screen,
it
is very difficult to plot the curved screen in a flat surface, since the
projection is not flat, but the center of this semi sphere is, so if you
are at that point, the audience, you see it at the right perspective.

This common error to try to plot a curved area in a flat surface brought
problems that makes everyone thinks our continents are in the shape we
see in the map, when in real they are not. Take this "flat world map"
and put three vehicles, top, center and bottom, traveling at the same
speed, from one side to another in the map, the top and bottom guys can
finish the travel forth and back several times while the center guy does
just once, because they are traveling different diameter circles.
The "flatted" map gives us a wrong concept of distances.

The observer at the center of the globe would never do that mistake,
since any circle would has the same diameter (in a perfect sphere), with
the circles centers at the observer point.

Unfortunately in a tri-dimensional world, our representation concept is
bi-dimensional. A great, great mistake done long ago, and it looks like
will not be solved soon. The 3D-TV would be a start, but...  

I think the biggest mistake was developing uni-dimensional electronic
technology. A multidimensional processor would do several routines at
the same time, millions, with a incredible processing power, getting a
little bit closer to our brain processing way, that runs at the
fantastic speed of only 12 Hz, consuming few milliWatts of power.

Suppose a PIC processor with a hundred or more subprocessors inside, and
each subroutine would be dispatched to a secondary processor. All the
processors need to finish the processing exactly at the same time, so if
one feels it will take more time, it dispatch parts of its code to other
idle processor.  The single dimension processing gives us wrong ideas
about how to build a software routine, we did learn to do it in a
sequential way, the previous routine needs to be finished before it
starts the next one, and it is very difficult to make our mind *see* a
way to do it differently, but not in our life. When you plan your life,
college, job, everything else, you are doing a multidimensional plan,
*this is not multitasking*, because multitasking does not intend a
synchronous intermediate ending steps, multidimensional processing does.

In a multidimensional processor, it takes only one clock cycle to do a
long processing sequence, but it is done by thousands of subprocessors.
Our *wrong way* to develop uni-dimensional processors,  needs several
thousand clock cycles to do the same in just one processor.  

Interesting analogy is about how you mentally do calculations, when I
ask you how much is 8 times 4 you use multidimensional processing and
locate the answer in your memory, you do not any math. But if you need
to do a long sum of number in a paper you will make your brain do a
uni-dimensional activity, because it was the way you learn how to do
it, so you will sum number after number in a sequence, even that this
summing activity uses multi-dimensional processing activity to get the
several small answers from the memory (... of from the fingers).

Any solar panel helmet to increase brain capacity?

I apologize for the out of topic subject, but I think it is something
that can make people "travel very far" thinking about it.
--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:   http:/http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\06@215235 by Andy Kunz

flavicon
face
>Interesting analogy is about how you mentally do calculations, when I
>ask you how much is 8 times 4 you use multidimensional processing and
>locate the answer in your memory, you do not any math. But if you need
>to do a long sum of number in a paper you will make your brain do a
>uni-dimensional activity, because it was the way you learn how to do

My great-grandfather was a Calculator during WWI and WWII for the US
military.  His job was to do math for an engineer who was working on
military hardware.  Actually, he worked for several men because he could do
the job faster than they could ask the questions.

He had the job because he was able to do math very quickly naturally,  His
game was to tell the cashier at the grocery store the bill (including tax)
before she could do it with her register.

The advantage of Calculators over adding machines was the speed of
multiplication and division.  Addition was fast enough by machine, but
Calculators could usually do it faster.

They also had the advantage of being able to suggest to the engineer or
professor an alternate method of solving the same problem, which suggests
that they were working on the math in an "alternate dimension" (not like
the X-files).

If only compilers could do that <G>.

Andy

1999\03\07@132450 by ryan pogge

flavicon
face
>Did you already tried to use the PC mouse with the opposite
hand? or
>reverse it in 180¡

180 isn't too hard....
try 90 degrees
and harder yet... 225 degrees!!!! im trying and it just does
not make sense to me :-)

1999\03\07@145946 by Gerhard Fiedler

picon face
At 11:09 03/07/99 -0500, Wagner Lipnharski wrote:
>Gerhard, very nice site about the subject (maps). I think if instead
>humans as the intelligent (well...) species on the planet, ants would
>create 3d maps.

since you're touching the subjects "intelligence" and "ants" and we're
quite OT anyway :)  i think our notion that we're actually intelligent (as
a species) comes more from the fact that we don't have a clue. as a species
we're as dumb as a log of wood. from what you say, you probably have
already sat over an ant hill (is that the right word?) for a while and
looked at them. (who's never done this, it's a very illuminating exercise
in self-reflection.) what they do is a whole lot more "socially
intelligent" than what we do. maybe that's what we call "freedom," but it
sure makes life a lot less funny for most of us -- but we can't help it.
the "critical mass" for that type of life style is not there. (yet?)

>Barefoot walking distances are almost always flat.

not where i come from :)  i grew up right at the border of the alps in
germany, and there almost =no= distance is flat... :))


>The "spatial vectoring" used by any "jet dogfight" plotting, takes long
>mind exercises for simple mortals to understand it, while "mosquitos" do
>it naturally with their insignificant brain.

i guess if we put to-be jet fighters at the age of four or five in a
simulator similar to those they have for the jets, the "real" ones, which
simulate acceleration and stuff, they'd probably create the brain
structures to process those informations more easily.

>Did you already tried to use the PC mouse with the opposite hand? or
>reverse it in 180¡?

like somebody else said, a straight angle is somewhat doable, but just
slightly off may be a lot worse. that's usually one of things i tell
beginners about the use of the mouse. i had quite a number of people who
miraculously augmented their mouse capabilities from one moment to the
other just by holding the mouse in a right angle to the screen... :)

ge

1999\03\07@152054 by Gerhard Fiedler

picon face
At 11:15 03/07/99 -0500, William M. Smithers wrote:
>Most of the
>users are guys who have spent their careers writing
>iterative code in FORTRAN, and trying to get them
>to twist their heads into the mindset of programming
>for fine-grained parallelism was virtually impossible.

um, yes, that illustrates the notion that many of the higher-end workers
are not essentially that more "intelligent" than the lower-end workers,
they just happen to have aquired a different skill set, but are not much
more flexible in expanding it... :)

i guess this would be the correct situation for the use of the expression
"a new programming paradigm." they've misused this term for almost every
new api, but actually there hasn't been much of a new "paradigm" since the
first formulation of sequential programs and the basic control structures.
maybe you can call object encapsulation a paradigm, too. but parallelism
sure would be a =new= one, and one that probably would require us to learn
programming almost from scratch, since it wouldn't be restricted to just
having a function (like the square root) process much faster, that's not
it. the real new thing is that they require completely different flow
control structures.

ge

1999\03\07@170150 by Alan King

picon face
Wagner Lipnharski wrote:
>
> Underground tunellers and exploration divers must use a different way to
> represent their tracks.  Just thinking about that makes our little
> friends, "ants", much more advanced than us in terms of planning and 3d
> mentalization/visualization (not saying about organization and
> participation).

 Ah, but you are giving them abilities they don't have, planning and
visualization.  The output set (three dimensional nest) appears complex,
but that doesn't necessarily imply much high-level thinking/processing
is going on.  Think of the Mandelbrot set.  Very nice looking/complex
seeming output, but it is generated by a simple set of rules, about 10
lines in basic and you can generate it and zoom etc..  They are
following simple rules, and the 3d nature arises from the ease of doing
3d in dirt.  They probably don't even 'remember' where the parts of the
nest are, they leave scents for everything.  They have simple
adaptations that turn what we see as a 'complex' undertaking into a
simple chore.  Walking on two legs is a much more complicated problem,
and yet we don't even think about it, it maps into the motor network of
our brains.  Much easier to walk, than to conceptualize it properly and
build a robot to do it.  Likewise, much easier to just build a 3D nest
than to visualize one and build to plan..
 Their low level brain (what they're thinking/remembering is much more
close to what they're actually doing, no real abstraction like we have)
is why they work so much more efficently together than most higher
animals.  No worker revolts or conflicting goals etc.
 This is remotely pic/robotics related discussion too.  I'm working on
a walker robot, and it has become clear that most walkers don't work or
work well because almost everyone conceptualizes walking incorrectly.  I
have a very different way of seeing the problem, and we'll see where it
takes me soon..

1999\03\07@173943 by Gerhard Fiedler

picon face
At 17:01 03/07/99 -0500, Alan King wrote:
>Wagner Lipnharski wrote:
>> Just thinking about that makes our little
>> friends, "ants", much more advanced than us in terms of planning and 3d
>> mentalization/visualization (not saying about organization and
>> participation).
>
>  Ah, but you are giving them abilities they don't have, planning and
>visualization.

if you look at a (human) city from the same perspective you look at an ant
hill, it's very hard to imagine that there actually is some planning going
on. :)

good luck with your "walker," and keep us posted how it owrks out.

ge

1999\03\07@175358 by Graeme Smith

flavicon
face
GRAEME SMITH                         email: spam_OUTgrysmithTakeThisOuTspamfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Sun, 7 Mar 1999, Gerhard Fiedler wrote:

{Quote hidden}

Actually yes, in order to define a fine grained approach to programming
that is NOT just a variation of serial programming, and doesn't depend on
speeding individual commands that still do serial programming, you would
need different flow control mechanisms.

The most advanced languages I have seen for parrallelism seem to be
limited to a basic split command, that essentially recreates the existing
state in another processor, and does two variations of the same code.

I personally have defined about 16 different types of split commands that
I would like to see implimented to get true parrallel programming
capability.

However, most of these could be implimented in a "structured Programming"
style, as variations of the loop commands, where instead of running a
loop, you split off parrallel threads and recombine them.

Thinking of it, has brought up one interesting thing, I have often
wondered whether there is a limit to practical parrallelism, a sort of
serial constant, that acts like the conservation of data rule, to conserve
the number of serial steps that MUST be taken within code.

My work with Neural Networks has suggested that even self-organizing
systems such as Neural Networks, need to have a certain "DEPTH" to them in
order to achieve proper function (Hence the hidden layers).

A Language the uses fine parrallelism (below the command level) then
allows structured commands that translate out to parrallelism at the
command level, and then Multi-tasking/Multi-threading at a level above the
command level, would probably make better use of Parrallelism than a
language that only approaches parrallelism at one level.

                       GREY

1999\03\07@193050 by Gerhard Fiedler

picon face
At 15:51 03/07/99 -0700, Graeme Smith wrote:
>Actually yes, in order to define a fine grained approach to programming
>that is NOT just a variation of serial programming, and doesn't depend on
>speeding individual commands that still do serial programming, you would
>need different flow control mechanisms.

>However, most of these could be implimented in a "structured Programming"
>style, as variations of the loop commands, where instead of running a
>loop, you split off parrallel threads and recombine them.

yes, the approaches i've heard of seem mostly to be either vectorized
sequential programs (using the same old sequential flow control techniques,
just spreading vector calculations over a vector of cpus) or some kind of
multi-tasking spread over different processors (using essentially the same
synchronization techniques used in conventional multi-tasking). nothing
really revolutionary in terms of parallelism.

>Thinking of it, has brought up one interesting thing, I have often
>wondered whether there is a limit to practical parrallelism, a sort of
>serial constant, that acts like the conservation of data rule, to conserve
>the number of serial steps that MUST be taken within code.

i guess this depends on the actual low level computing. as long as this is
strictly sequential, there can't be much different results, and IIRC there
has been some research on the amount of synchronization between processes
and when this gets inefficient, ie. takes over more time (as a sum of the
time each of the processors in the systems spends) than the one added
provides additionally.

i think the real advance in parallelism lies probably in completely
different hardware structures, something different from the actual
inter-connected sequential calculators. this would have to emerge slowly,
the theory and the hardware and the related software techniques going in
parallel, like it has been with sequential computing -- if it comes.

ge

1999\03\07@204735 by Wagner Lipnharski

picon face
I believe that a neural system, pure hardware with each
one of the distribution nodes doing few decisions based
on other surrounding nodes decisions, is probably a start.
Our brain works somehow like that, but impossible to
imagine such magnitude of multidimensional processing.
Another astonishing thing is that our brain can build
new bridges between those nodes, and eliminate others,
as if doing a self reprograming. Some medical research in
the area show that a person that lost part of the brain,
with sub sequential lost of mechanical functions, can get
the functions back in time, sharing and splitting the
remaining nodes to control the disconnected functions,
if repetitively trying to make those lost movements,
pure exercises.  Sounds that our neuron programming is
sole physical and is a result of pure necessity, as when
a child takes months to learn mechanical controls.

It would be nice to teach mathematics to a learning
computer just showing it the wrong moves it makes.

When I left IBM in 1993 I heard about an IBM research
about an organic processor, based on something like that,
but I never got any additional info about.
--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:   http:/http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\08@134543 by John Payson
flavicon
face
|Thinking of it, has brought up one interesting thing, I have often
|wondered whether there is a limit to practical parrallelism, a sort of
|serial constant, that acts like the conservation of data rule, to conserve
|the number of serial steps that MUST be taken within code.

For different problems, there are "hard" limits to how well they can
be rendered into "parallel" algorithms.

For example, even ignoring the fact that adding more processing units
will cause them to be further apart, addition of two large numbers will
require time proportional to the log of the number of digits, no matter
how many processing units you have to "throw" at the problem.  The norm-
al algorithm (which is even used in a lot of, e.g., 32-bit adders) is to
define--for any adder of any size, the terms "P" and "G" to indicate
whether the carry-out of the adder will be /propagated/ [i.e. true if
there was a carry-in] and/or /generated/ [i.e. true regardless of whether
there was a carry in].  For a 1-bit adder, P will be true if either of
the inputs is set, while G will be true if they are both set.  To combine
two adders into a larger one [P0 and G0 being the P and G of the one
handling the less-significant bits], compute P=P0 and P1; G=(G0 and P1)
or G1.  Once all the P's and G's are computed, it is then possible to
use those to compute the carry-in for all the individual stages and then
to compute the actual results.

Multiplication of two large numbers requires more hardware to achieve
maximum parallelism, but it too takes time proportional to the log of
the number of digits.  The trick here is to use what's called a "carry
save adder" [a useful trick to also know, btw, when doing 'vertical
addition'].  A carry-save adder is a simple device whose propagation
delay is independent of the width of the input, which will take in
three numbers and output two numbers whose sum equals that of the orig-
inal three.  Since multiplication is nothing more than the addition of
a lot of numbers [the multiplicand shifted by different amounts] it's
possible to use carry-save adders to reduce the number of terms to be
added to two (which are then summed using a conventional adder).

As of a few years ago, division was the one type of basic arithmetic
operation which was thought not to be effectively parallelizable.  This
is in fact not the case, however; if you check the patent web site you
will find a patent owned by Cyrix which describes a very good way of
performing division quickly.  Unfortunately for me, my independent dis-
covery of basically the same algorithm was a couple years too late [oh
well... otherwise I might be a millionaire.  But then I might never have
met the woman I later married.  So maybe it's all for the best... >:*3 ]

1999\03\08@140624 by Gerhard Fiedler

picon face
At 12:46 03/08/99 -0600, John Payson wrote:
>For different problems, there are "hard" limits to how well they can
>be rendered into "parallel" algorithms.

but only if you still insist on using parallel sequential processes, which
are not "the real thing". for example when we try to figure out how much 5
times 5 is, we usually don't propagate any carry-ins, we just associate.
that's a whole different process, and is probably not bound to the limits
you state.

ge

1999\03\08@142246 by Andy Kunz

flavicon
face
>performing division quickly.  Unfortunately for me, my independent dis-
>covery of basically the same algorithm was a couple years too late [oh
>well... otherwise I might be a millionaire.  But then I might never have

And if Procomm had used the "browser" I developed in 1988 which very
closely parallels Netscape's, maybe we'd both be rich.

Andy


  \-----------------/
   \     /---\     /
    \    |   |    /          Andy Kunz
     \   /---\   /           Montana Design
/---------+   +---------\     http://www.montanadesign.com
| /  |----|___|----|  \ |
\/___|      *      |___\/     Go fast, turn right,
                              and keep the wet side down!

1999\03\08@150620 by Wagner Lipnharski

picon face
I think that we, the old vicious sequential process workers, simple
*can not* see an easy way to do it.  It is deed to young children
develop it, without any old ways interference.  I wouldn't dare to
try to motivate a kid to try it, not even show the direction of it,
since my old and not efficient sequential programming knowledge
would interfere severely in kid's mind, in a disastrous way.

I already saw some inexperienced students, fresh new ones, come up
with some very strange solutions, just because they didn't know
the common tracks, they used their own imagination.

In this new multidimensional processing systems, carry bit would
be as useless as an iron horse shoe in a Formula-1 pitstop,
or at NASA launched, except if used for good luck.

When a very small kid take two toys with both hands, one in each
hand, this is the baby's mind sharing mechanical useful motion,
and it is not doing mathematics or logic, simple and plain
adaptability, based on the past right and wrong experiences.

The human capacity to grow and create progress is based in the
fact that when we born, just the basic survival microcode comes
along with our brain, nothing else, so we need to learn and
this is the key for the progress. If we would born like ants,
with a complete microcode running, we would have no evolution.

If we really want to pursue that hiper-powerful capacity of a
MD processing, we need to understand that it will be so vast
that will be almost impossible to be programmed at the factory,
it needs to "learn" as we do.  We also need to accept that this
"machine" will do mistakes. A machine like that will require a
new profession, the "psique-programmers", someone that will
helps the machine to find ways, by itself, to cut wrong
connections and make good ones in its neural processing system.

It is impossible to create a concept of how it'll works, do
math or decisions, it would be a complete and different
processing world. We can't even dream about it, not with
our "less-than-100-years-of-data-processing-knowledge".

Wagner

Gerhard Fiedler wrote:
{Quote hidden}

1999\03\09@053131 by Stefan Sczekalla-Waldschmidt

flavicon
face
Hi,

John Payson wrote:
>
> (...)
> As of a few years ago, division was the one type of basic arithmetic
> operation which was thought not to be effectively parallelizable.  This
> is in fact not the case, however; if you check the patent web site you
> will find a patent owned by Cyrix which describes a very good way of
> performing division quickly.  Unfortunately for me, my independent dis-
> covery of basically the same algorithm was a couple years too late [oh
> well... otherwise I might be a millionaire.  But then I might never have
> met the woman I later married.  So maybe it's all for the best... >:*3 ]

The internet fools me - I tried to find the patent but I can't find it.
Can you eventually point me to the web page or post the algorithm ?

Kind regards,

       Stefan

1999\03\09@133301 by John Payson

flavicon
face
> As of a few years ago, division was the one type of basic arithmetic
> operation which was thought not to be effectively parallelizable.  This
> is in fact not the case, however; if you check the patent web site you
> will find a patent owned by Cyrix which describes a very good way of
> performing division quickly.  Unfortunately for me, my independent dis-
> covery of basically the same algorithm was a couple years too late [oh
> well... otherwise I might be a millionaire.  But then I might never have
> met the woman I later married.  So maybe it's all for the best... >:*3 ]

|The internet fools me - I tried to find the patent but I can't find it.
|Can you eventually point me to the web page or post the algorithm ?

Sure.  Go to http://www.patents.ibm.com/advquery, then search for
"cyrix" in the assignee field and "division" in the abstract field.
This will yield six patents; the last three seem to be slight var-
iations on the same idea.

Basically, the trick is to multiply the divisor by an approximate
reciprocal so that for some base (e.g. base-256 if you want to do
8 bits at a time) it is of the form "10xxxxx".  Once this has been
done [and it's pretty easy] the first "digit" of the result either
equals the first digit of the dividend, or the first digit of the
dividend minus one.  This digit is ten multiplied by the scaled div-
isor and subtracted from the dividend and the cycle is repeated.

For example, in base 1000, let's divide 150,146,497,025,115 by
281,120.  We first multiply the divisor by 3558 to convert him
to the suitable form [1,000,224,960].  Now we settle down to do
the division:

               150,112 r 112,025,195,520
              --------------------------
1,000,224,960 | 150,146,497,025,115
               150,033,744,000
               ---------------
                   112,753,025,115
                   112,025,195,520
                   ---------------
                       727,829,595

At this point, we multiply the quotient by the same amount as we scaled
the divisor [3558] yielding 534,098,496, divide our remainder into the
original divisor yielding 2,589, and add that to the quotient yielding
534,101,084.

Cute, eh?

1999\03\10@091738 by Graeme Smith

flavicon
face
GRAEME SMITH                         email: .....grysmithKILLspamspam@spam@freenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Sun, 7 Mar 1999, Gerhard Fiedler wrote:

> yes, the approaches i've heard of seem mostly to be either vectorized
> sequential programs (using the same old sequential flow control techniques,
> just spreading vector calculations over a vector of cpus) or some kind of
> multi-tasking spread over different processors (using essentially the same
> synchronization techniques used in conventional multi-tasking). nothing
> really revolutionary in terms of parallelism.
>
Oh.... And here I thought it was just my own limited exposure ;)

{Quote hidden}

Well, some of the techniques that have been developed, get into
parrallelism at the micro-code level, and of course Neural Networks are an
essentially analog technique, but a limited exposure to Anatomy, suggests
that even "Evolutionary Algorythms" have to retain a certain amount of
serial "DEPTH", which suggests a limit to parrallelism.

Pipelining, allows you to "Cheat" the serial limitations, but you still
have a need for a certain amount of serial structure in your system.

Which is why I think that there is an essential conservation of serial
steps, in computing.

>
> i think the real advance in parallelism lies probably in completely
> different hardware structures, something different from the actual
> inter-connected sequential calculators. this would have to emerge slowly,
> the theory and the hardware and the related software techniques going in
> parallel, like it has been with sequential computing -- if it comes.
>
> ge
>
I don't see any reason to believe that it won't come...

After all, I am working on the theory for some new techniques right now...

But, like everything else, it takes time, to rethink old theories, and its
much easier, to rehash old work, than to create completely new work, and
much easier to sell when you do, unless you have already built your
prototype. Which of course takes money to design.. which you can't get
until you get backing which you can't get until you prove your theories
which you can't do until you have a prototype... ad nauseum

                               GREY

1999\03\10@094210 by Graeme Smith

flavicon
face
GRAEME SMITH                         email: grysmithspamKILLspamfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Sun, 7 Mar 1999, Wagner Lipnharski wrote:

{Quote hidden}

I recently had a problem with my right eye....

the sphincter muscle in the iris, refused to close.

The doctors, assured me it was a relatively benign condition caused by a
virus, selectively destroying the synapses of one type of neuron in the
nerve. They suggested that the eye would attempt to "Repattern" itself,
by using the same neurons that focus the eye, or hold open the eye-lid.

For about a month, I had trouble focusing my eye, especially in bright
sunlight.... Now, sometimes when I look in the mirror, my eye seems to
be able to close the iris, and other times it doesn't. In other words,
Yes, the Neurons can repattern... but not quickly, and not efficiently.

Recent Medical Comments about Post Polio Syndrome, suggest that polio,
which affects a much larger population of neurons, essentially creates
the same sort of effect, it forces the body to "repattern" the nerves,
and people who have overcome the parrallysis of polio, are devastated
when a predictable time later, their parrallysis re-emerges. due to
Neural Fatigue killing off the repatterned neurons.


What most computer engineers see as really valuable about a system that
can reconnect itself, is that over time, it will degrade gracefully,
rather than just deciding one day to bite the dust.

Imagine the Mars Rover lasting hundreds of times as long as expected
rather than merely 10...

>
> It would be nice to teach mathematics to a learning
> computer just showing it the wrong moves it makes.

The problem with this approach, (And we can already to some extent achieve
it with neural networks), is that we don't know how to teach the computer
how to "Understand" its mistakes, so it isn't enough really to just show
it that it is wrong, what good is getting the computer to get the right
answer, if, it never learns enough to figure out WHEN it should cheat a
little?


>
> When I left IBM in 1993 I heard about an IBM research
> about an organic processor, based on something like that,
> but I never got any additional info about.

Hush hush... top secret... Frankensteinian Black Project...
Best not to say too much about it.... eh?  ;)


                                       GREY

1999\03\10@102650 by Graeme Smith

flavicon
face
GRAEME SMITH                         email: .....grysmithKILLspamspam.....freenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Mon, 8 Mar 1999, Gerhard Fiedler wrote:

> At 12:46 03/08/99 -0600, John Payson wrote:
> >For different problems, there are "hard" limits to how well they can
> >be rendered into "parallel" algorithms.
>
> but only if you still insist on using parallel sequential processes, which
> are not "the real thing". for example when we try to figure out how much 5
> times 5 is, we usually don't propagate any carry-ins, we just associate.
> that's a whole different process, and is probably not bound to the limits
> you state.
>
Well... suppose even associative techniques, still require some
co-ordination... Then, despite the fact that you can associate in
parrallel quite quickly, you would still have some serial dependency.

For instance, take the formula  (5+1)*6

There is an implied serial dependency that separates that from
the formula

                               5+1*6

In fact, if you want to express it in sum of products terms the equivalent
is more like

                               (5*6)+(1*6)

No matter how you try to combine these functions, (addition and
multiplication) if you don't deal with the sequential dependency, your
answer will be different.

You can say that the formula is in an essentially parrallel form, but the
sequential dependency, is defined by the syntax.

You simply cannot get the right answer, except by doing two separate
steps in sequence.

That is why I believe there is an essential conservation of sequential
dependency. Not because some techniques require more sequential steps than
others.

In the example I used, (compression) different levels of compression are
possible, using different compression models, and different coding
techniques, but there is a theoretical limit to compression, that has yet
to be overcome without losing data.

It is true that you can cheat, in specific cases, and get much greater
compression, but it is not the RULE. The RULE is that any general
compression algorythm, cannot exceed the theoretical limit.

The difference is that in specific cases, you can imbed data, that is
known by both reciever and transmitter and therefore does not need to be
sent. But when you do this, you limit the number of recievers that can
recieve the data.

So, such a program would be unable to make a useful general compression
algorythm, and might be mistaken for encryption.

                               GREY

1999\03\10@124306 by Gerhard Fiedler

picon face
At 08:24 03/10/99 -0700, Graeme Smith wrote:
>Well... suppose even associative techniques, still require some
>co-ordination... Then, despite the fact that you can associate in
>parrallel quite quickly, you would still have some serial dependency.
>
>For instance, take the formula  (5+1)*6
>
>There is an implied serial dependency that separates that from
>the formula
>
>                                5+1*6
[...]
>You simply cannot get the right answer, except by doing two separate
>steps in sequence.

i don't see the difference here as parallel vs. serial. of course, if you
parse the expression serially, or if you think in serially formulated
parsing, you have a serial dependency. but it seems to me that expressions
may just as well "parsed" as a whole -- don't see why not, but don't see
exactly how, though :)  the parentheses of course imply serial parsing,
since they are a serial definition ("first this, then that"). but this
might just mean that the notation is (or the underlying principles are) not
adequate.


>It is true that you can cheat, in specific cases, and get much greater
>compression, but it is not the RULE. The RULE is that any general
>compression algorythm, cannot exceed the theoretical limit.

i guess as long as you talk about algorithms, you're right, because they
are essentially serial. and maybe it's so fundamental that it won't ever be
overcome, maybe having to do with us being kind of "stuck" in time. but so
far most rules had their limits, it's just a matter of time to find them.

thanks for the insights (they enhanced my very limited exposure :)

ge

1999\03\10@131813 by Wagner Lipnharski

picon face
Graeme Smith wrote:
> Well... suppose even associative techniques, still require some
> co-ordination... Then, despite the fact that you can associate in
> parrallel quite quickly, you would still have some serial dependency.

A thousand parallel processors will run serial routines, this is
not multidimensional data processing.  We are talking about just
one processing cycle done by zillions of processors.  For sure
it doesn't mean just one electronic cycle, it can be a cascade
or sequential logic gates, but it will not involve the need to
finish something to do the next step. Parallelism involves the
thinking that everything is being done at the same time, what
again, several things running serially in parallel, this is not
the mdp idea. We don't know how to do it, yet.
MDP is somehow comparable to understand what is the physical
movement of every molecule of water in the Atlantic, track it
down, without analyze it in a matrix, serial or parallel ways,
just if zillions of tinny processors logging it, and somehow
you could get the information from any processors far or close.
If the distance (time) is not an issue you can forget about
the "serial" expression. Just a silly example.
--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:   http:/http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\10@143459 by Marc

flavicon
face
> MDP is somehow comparable to understand what is the physical
> movement of every molecule of water in the Atlantic, track it
> down, without analyze it in a matrix, serial or parallel ways,
> just if zillions of tinny processors logging it, and somehow
> you could get the information from any processors far or close.

Probably you can view at the zillions of water molecules _being_
zillions of "tiny processors" designed for performing the Atlantic
ocean problem in realtime, with least energy consumption! Only
the space requirements of this super-computer is a little bit
high. :-)

1999\03\10@170644 by Sean Breheny

face picon face
At 08:22 PM 3/10/99 +0100, you wrote:
>> MDP is somehow comparable to understand what is the physical
>> movement of every molecule of water in the Atlantic, track it
>> down, without analyze it in a matrix, serial or parallel ways,
>> just if zillions of tinny processors logging it, and somehow
>> you could get the information from any processors far or close.
>

It seems to me that MDP processors would be inherently suitable to certain
tasks and inherently unsuitable to others. For example,in any case where
you want to track lots of processes happening at once,as in your
example,MDP would be VASTLY better than any serially-arranged method.
However,aren't many of the tasks we ask computers to do inherently serial?
Somebody else gave a good example in a simple mathematical expression. You
often must at least begin to calculate one part of it before you can begin
the others. There are dependencies where one processor would have to wait
for another to finish before beginning its own calculation. It seems to me
that this will occur in any case where the input is single dimensional
and/or the output is single dimensional. In the water molecule case,you
inherently want a vector output (the position/momentum of each molecule,for
example) and you certainly take a vector input (the initial conditions for
each molecule).

Also, some list members have suggested that our brain does something
wonderful when we are able to say instantly "3*4=12". It seems to me that
this is a simple lookup table,neural network style.A PIC can do this really
fast,too,if it uses a lookup table! Don't get me wrong,the human brain is
an AMAZING thing,and I don't think we will EVER fully understand it. Most
of the things that it does which are so amazing,however,somehow involve
inherent parallelism. Biological brains are the best image processing
computers,hands down. They can track dozens or even hundreds of visual
targets at once and filter out the unneeded information in a fraction of a
second. However,this is because the brain is MASSIVELY PARALLEL and it is
dealing with a massively parallel situation (lots of moving objects). It
doesn't excel in processing an inherently serial input,unless that
processing involves pattern matching or creativity,two areas at which
computers are kinda challenged.

Sean

|
| Sean Breheny
| Amateur Radio Callsign: KA3YXM
| Electrical Engineering Student
\--------------=----------------
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
EraseMEshb7spam_OUTspamTakeThisOuTcornell.edu ICQ #: 3329174

1999\03\10@183403 by John Payson

flavicon
face
|Also, some list members have suggested that our brain does something
|wonderful when we are able to say instantly "3*4=12". It seems to me that
|this is a simple lookup table,neural network style.A PIC can do this really
|fast,too,if it uses a lookup table!

I think it's somewhat more of a "results cache".  Although many
people are indeed tought the multiplication table (through 9x9
at least, and the 10x's and 11'x are really easy) a lot of "code-
heads" can answer certain types of slightly larger multiplications
(e.g. what's 96*256) almost reflexively, despite having never had
to explicitly memorize a list of multiples of 256.

|                         Don't get me wrong,the human brain is
|an AMAZING thing,and I don't think we will EVER fully understand it. Most
|of the things that it does which are so amazing,however,somehow involve
|inherent parallelism. Biological brains are the best image processing
|computers,hands down. They can track dozens or even hundreds of visual
|targets at once and filter out the unneeded information in a fraction of a
|second. However,this is because the brain is MASSIVELY PARALLEL and it is
|dealing with a massively parallel situation (lots of moving objects). It
|doesn't excel in processing an inherently serial input,unless that
|processing involves pattern matching or creativity,two areas at which
|computers are kinda challenged.

The brain is an excellent association machine; given a problem it can
often quickly grab the knowlege it posesses associated with that problem.
In that way it is very much a parallel machine; all of the neurons assoc-
iated with various memories work almost simultaneously to retrieve the
needed information.  The foreground processing of such information, how-
ever, is largely sequential.

What is perhaps most interesting, though, about the brain is its ability
to adaptively program its motor control networks to perform amazingly
complex tasks.  Most people who have much experience driving a car can
do so with very little foreground thought; it's possible to be busy think-
ing about other things with the foreground and yet still manage all "norm-
al" driving-related activities in the background.  [note: this can be dan-
gerous as the foreground processor may take a few hundred miliseconds to
context-switch when the background process chimes in with an "unexpected
object in vehicle path" interrupt].  Really quite amazing.

1999\03\10@194456 by Wagner Lipnharski

picon face
> Also, some list members have suggested that our brain does something
> wonderful when we are able to say instantly "3*4=12". It seems to me
that
> this is a simple lookup table,neural network style.A PIC can do this
really
> fast,too,if it uses a lookup table! Don't get me wrong,the human brain
is
> an AMAZING thing,and I don't think we will EVER fully understand it.
Most
> of the things that it does which are so amazing,however,somehow
involve
> inherent parallelism. Biological brains are the best image processing
> computers,hands down. They can track dozens or even hundreds of visual

> targets at once and filter out the unneeded information in a fraction
of a
> second. However,this is because the brain is MASSIVELY PARALLEL and it
is
> dealing with a massively parallel situation (lots of moving objects).
It
> doesn't excel in processing an inherently serial input,unless that
> processing involves pattern matching or creativity,two areas at which
> computers are kinda challenged.
>
> Sean

It was me, who said about the XxY=Z and about the molecules, now, pay
attention to a working neuron, it just connect two points because and
just
because the influence of the neighborhood, and few other non
compressible
details yet.  This is pure memory, look ahead tables.  You can at gross
mode
compare with a huge e2prom with zillions of bits, but with the
difference that
they can interact between each other, and form very complex MD gates.

The molecules in the Atlantic example, is an example where (6+1) * 5 is
done in
thousand of different ways (at the same time) and only at the end the
right one
is decided to be the correct.   As the molecules, you can decide (at the
end)
which ones would be the ones to indicate the correct big mass movement
or
pressure changes and then plot a possible hurricane or something.  In
this
fashion, things doesn't need to be done in a sequential way, so several
calculations at the same time could take place, as 6x1,  6x5, 1x5, 1x6,
6+1,
5+1, several answers would be available, and just one will be the
correct.  I
think the word here is judgment, and this is why a young children can
not be
responsible, his(her) neuro-network is not yet "operational" for that
kind of
judgment about what is right or wrong.  I hnow a person that can "judge"

calculation results, in some great percentage he is right, he doesn't
know
exactly how he "knows" the correct answers.  I believe that this is what
makes
great engineers, doctors, scientists, artists, and so on, his "judgment"
about
the subject is better than the majority.

Don't ask me how it is done, as I said before and at the beginning of
this
subject, we don't know the answers, we are not prepared to understand it
yet.

I think a psychiatric or "shrink" can answer better those questions. We
are
trying to duplicate a MD neuron processor, our brain, and using the same
brain
to do it, it is similar to a PIC microchip trying to understand its own
code
and  "modus operandus", at least this attempt is ridiculous.

Wagner.

1999\03\11@211646 by Graeme Smith

flavicon
face
GRAEME SMITH                         email: grysmithspamspam_OUTfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Wed, 10 Mar 1999, Wagner Lipnharski wrote:

> A thousand parallel processors will run serial routines, this is
> not multidimensional data processing.  We are talking about just
> one processing cycle done by zillions of processors.

OK, but all you will get is a single transform. You might get all the
data in that transform, and you might be able to do multiple transforms
at the same time, (in parrallel) but all you get, is one mapping of
datasets from one point to another. Instead of looking at it as parrallel
processing, you can look at it as a very complex mapping algorythm, doing
one very complex transform.

Good as far as it goes, and Hey, it can process a wide set of things all
at once, but, it ain't going to compete with the human brain, that uses
that technique, PLUS some Sequential Routing, to create an even more
powerful effect.

You may recognize a face in an instant, but to put a name to it, you have
to first recognize that you recognized it, and THEN search for the name.

In other words, to do anything more than process the data, you NEED a
serial relationship between the data, and its end product.


> For sure
> it doesn't mean just one electronic cycle, it can be a cascade
> or sequential logic gates, but it will not involve the need to
> finish something to do the next step. Parallelism involves the
> thinking that everything is being done at the same time, what
> again, several things running serially in parallel, this is not
> the mdp idea. We don't know how to do it, yet.

No? Isn't that what the guys that are trying to do Quantum Computing are
describing? Potential Parrallelism that is virtually unlimited until you
get to the point of trying to sample the answer, then suddenly it settles
into a single state?


> MDP is somehow comparable to understand what is the physical
> movement of every molecule of water in the Atlantic, track it
> down, without analyze it in a matrix, serial or parallel ways,
> just if zillions of tinny processors logging it, and somehow
> you could get the information from any processors far or close.
> If the distance (time) is not an issue you can forget about
> the "serial" expression. Just a silly example.

Well, that "Silly example sounds like a single transform to me... ;)

mapping some characteristic of every molecule of the sea to data, then
sampling the data. But try to consider sampling the data, before it is
mapped, and you get an idea of why seriality is impossible to completely
remove from the situation. On a pic, try an experiment, instead of waiting
for your data to settle, try to feed a value into the pic before it
completes its reset while the data is still in an unknown state.

Are you going to get any valid response? Not likely!


                               GREY

1999\03\11@223431 by Wagner Lipnharski

picon face
Grey wrote:
[snip]

In a world where we are used to see things in a traditional
way, as separated entities as "memory", "ALU", "priorities",
"output after input" and others, MDP doesn't fit.

The recognition of a face would be done empirically by everyone
of those photosensors processing circuits, not the sensor
itself. So again comparing with the neuron function, each
sensor understand its single bit pattern according to its
neighbors reaction, and this modeling of "transformation"
as you said, generates a final substance that we can call not
only the "vision" but also the "judged" interpretation of
what means that vision.  In real our "vision" works somehow
like that, we can recognize a pattern in a fraction of a
second, even that we don't know what is it.  Our visual
memory associated to each sensor in the retina doesn't make
part of our deep memory.  The "association" we make between
the image of a dog and the name "Snoopy" makes part of a
high skilled memory banks fetching process.  And this is
why it is very common to see somebody and recognize as
someone you know, but you don't remember nothing else.

This visual high speed identification makes part of our
defense system, developed from thousands of years, to
very fast recognize our enemies and predators, much before
trying to identify "what is the name of that lion?".

The females needed to develop this ability much more than
the male, since it was the female who protected the prole,
so "she" needed this sensoring intensified, and this is
why police prefer to have female witnesses, and this is
why females are attracted to colors, reflective things
(not necessary gold :), and high busy visual fields.

This visual processing is bit mapped and cross several
brain fields in a very interesting reading matter for
a raining night. It is not serially processed, since
it is faster than our brain processing speed.  So, it
looks like to be a sophisticated MDP.  It took millions
of years to achieve the actual state, and it is the most
accurate sensor all predators use for hunting.

Just as a test, put a dollar bill in the middle of a book
and run fast the pages with your thumb, and try to stop
when you find the dollar. You will stop dozens of pages
later. This is not only the mechanical muscle delay, it
has a visual post processing timing that doesn't make
part of the recognition process.  This neighborhood
patterns makes our visual system much more sensible to
certain colors, representing the real danger, as certain
animal colors or even blood. It was not only intensified
in the fast recognition process but also by intensifying
the sensoring in the retina for those light waves.

This is a very interesting subject not explored by the
technology, the "Neighborhood Effect in MDP".
The closer "NE" I already saw in technology is the Parity
and Zero flag in the 8051 chip. It is done by hardware,
in nanoseconds with no cpu clock dependency.
--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:   http:/http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\12@232919 by Graeme Smith

flavicon
face
If the visual Mechanism were a pure MPD, it would be a single layer of
cells. Instead, we have a single layer of cells, called the retina,
connected by a cluster of fibrous connecting cells, called the optical
nerve, feeding into a FANOUT, system, that then connects to a number of
Parrallel processing Lobes in the brain, and also through an inversion/
swap that projects the image projected on the retina, upside down, and
backwards so that it seems right side up, and correctly oriented despite
the fact that the body image is mirror inverted in the brain.

simultaneously (in parrallel) with the vision, a part of the brain is
constantly adjusting the light levels, and the focus, to create the best
vision constant possible with our eyes focal length.

At the same time (also in parrallel) another part of the brain is
constantly monitoring the moisture level on the surface of the eyeball,
and adjusting it, when it gets too low, by blinking the eyes. Further at
some level of the brain, the blinking is virtually edited out of our
experience, so that only when we actually think about it, do we notice our
eyes going black for a split second in order to blink.

Because we are unconciously editing all this detail out of our stream of
conciousness, we are not aware of the large number of serial tasks that
are happening in parrallel. That however does not change the fact that
blinking takes time, and focussing takes time, and adjusting the iris
takes time, etc. etc.

The structure of the brain, is complex, not because of its
multi-dimensional processing, but because it can't always use MPD to get
the job done, and so it has to build serially organized networks, and
processes to achieve the full range of function necessary to survive.

For instance, to recognize that you have "Recognized" a tiger, and
trigger the "Jump" that will get you out of the way, the brain
actually needs to process at least 23 serial steps, only 6 of which are in
the CORTEX which is the closest we come to a Multi-dimensional processor
in the brain.

I really think you need to spend a little more time thinking about the
limitations of processing, even in a Quantum Processor, and less time
fantasizing about not needing serial steps.

                               GREY


GRAEME SMITH                         email: @spam@grysmithKILLspamspamfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.

1999\03\13@111105 by Wagner Lipnharski

picon face
All the parallel and serial effects you said about the visual system are
purely mechanical, moisture,
iris, those are simply support systems that have not a high priority on
the system.  Few other mechanical has
higher priority as for example "focus". This also has nothing to do with
MDP, this is simply multi-task based
on cause/effect. (We almost don't "see" the blinking because it is not
100% of the times we blink that we close completely the eyes, 4 in 5 we
just conduct lacrima to the center of the eye. This is another defense
system to keep us alert. By the way, we blink even with external
moisture applied to the eye... The predator birds never obscure their
vision, except when sleeping).

I was talking purely about the image caption and recognition mechanism,
that has nothing to do even with
the interpretation of the image.  When you see a car crash photo with
blood and other things, it is repulsive
much before you recognize exactly what it is.  This is what I am talking
about, and it happens without the
correct light level, moisture and cause/effect (serial processes).  That
photo can be repulsive upside
down, or at any angle, and an alert will be created much before the
translation is done. If you learn about
the "neighborhood process" you will see it.

Remember the "8051 parity flag", it could be hardware connected to drive
a port pin, much before the next
instruction would be fetched from the code memory to understand it. A
silly example of neighborhood effect.

Think about the "limitations" is very difficult when we don't understand
not even a little part of the system,
or not even "dream" about how to replicate it, but there are no limits,
except in out rational way to think.

Serial steps make part of a universe that uses time for the cause/effect
process, I am not talking about it, and we will never reach MDP or some
mysteries of the brain or space warp, while thinking in serial steps.

Some time ago I said here that the word "impossible" need to be
re-spelled again as "difficult", because we are entering a human phase
where technology allow us to see, to learn, think and conclude better.
But not in front of a MTV channel, "that" is pure fantasy and leads to
the absolute nothing (except for the marketing results).

--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:  http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\13@134109 by Gerhard Fiedler

picon face
At 11:10 03/13/99 -0500, Wagner Lipnharski wrote:
>Remember the "8051 parity flag", it could be hardware connected to drive
>a port pin, much before the next
>instruction would be fetched from the code memory to understand it. A
>silly example of neighborhood effect.

and not a good one, too, because this one is =clearly= a serial process,
even if it's high speed logic :)  you have to apply the data at the input,
and, some defined propagation time later, the output has the appropriate
state. if you build a computer based on this process, you definitely have
serialization of the calculation.

ge

1999\03\13@142754 by Wagner Lipnharski

picon face
Gerhard Fiedler wrote:
>
> At 11:10 03/13/99 -0500, Wagner Lipnharski wrote:
> >Remember the "8051 parity flag", it could be hardware connected to drive
> >a port pin, much before the next
> >instruction would be fetched from the code memory to understand it. A
> >silly example of neighborhood effect.
>
> and not a good one, too, because this one is =clearly= a serial process,
> even if it's high speed logic :)  you have to apply the data at the input,
> and, some defined propagation time later, the output has the appropriate
> state. if you build a computer based on this process, you definitely have
> serialization of the calculation.
>
> ge

Remember that MDP never ruled to process the information
in time=0, but in just one clock cycle.  Yes, pure hardware
gets very close to the MDP idea, while serial programming
is totally the opposite, but you can imagine the size and
power consumed by a pure hardware circuit to be able to do
calculus with 128 bits...  A serial process involves more
than two steps (start and end), involves  middle point(s),
while the parity bit is developed based on logic in one
simple propagation time into a 8 bits exclusive or.

Remember the serial cascade chips we were talking about,
the Q-1 being able to propagate to Q+1 at the same clock?
if you go very close to that possibility, you can almost
eliminate the middle points of the process, and somehow
propagate all the bits at once, in just one clock cycle.
Remember that in MDP there is no reference to clock, since
"clock" itself is the master of any serial process.
Clock cycles, or MDP cycles, creates the dominion.
Yes, you can say the dominion is a serial thing.
To avoid that, or with the fear of need to live with that
problems, we created the serial programming, when step
*a* needs to ends before we starts step *b*.  In some
theories of space-warp, step *b* needs to be accomplished
first to allow step *a* to start.

A very crude example is that we are developing technology
and applying it at the Hubble telescope today, to see what
happened to the stars millions of years ago.  So we are
anticipating ourselves for the "past to come".  :)
It is almost the same as if today you are doing plans and
deductions trying to find out what you did yesterday, but
you don't know it yet, you will find out it tomorrow....
strange huh?

So, in the space-warp, (of course, caused by the light speed
delays) you just desconsider if that "thing" happened already
or will happens. Time is just a point of view, so clock or
any other references to it, simply can not be used.

Well, we can not do it with PIC, 8051 or any other, not
for a while :).
--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:  http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\13@210618 by Andy Kunz

flavicon
face
>the interpretation of the image.  When you see a car crash photo with
>blood and other things, it is repulsive
>much before you recognize exactly what it is.  This is what I am talking
>about, and it happens without the

All this just goes to prove that only an Infinite Engineer could design a
system so simple yet so complex and implement it into the smallest of
creatures.

No wonder David, King of Israel could write, "I am fearfully and
wonderfully made."

Andy

1999\03\13@212523 by Sean Breheny

face picon face
At 08:55 PM 3/13/99 -0500, you wrote:
>All this just goes to prove that only an Infinite Engineer could design a
>system so simple yet so complex and implement it into the smallest of
>creatures.
>
>No wonder David, King of Israel could write, "I am fearfully and
>wonderfully made."

So true!

And here is another thing to ponder: When you hook up a battery to a
resistor,what fraction of what is REALLY happening there can you,or
anyone,explain? Think of how many interactions are going on in just that
simple,simple system. Think of all the atoms in the copper and carbon which
are being excited in various states,producing atomic scale oscillations and
strange,beautifully odd shaped charge distributions. Not to mention the
nuclear strong force which is holding the nuclei and quarks and who knows
what else together.

Humbling isn't it?

Yet we can do so much! How much MORE complex is a PIC than a resistor!

Its quite a universe. Must be quite an Infinite Engineer :-)

>
>Andy
>
|
| Sean Breheny
| Amateur Radio Callsign: KA3YXM
| Electrical Engineering Student
\--------------=----------------
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
KILLspamshb7KILLspamspamcornell.edu ICQ #: 3329174

1999\03\13@212526 by Wagner Lipnharski

picon face
Andy Kunz wrote:
> No wonder David, King of Israel could write, "I am fearfully and
> wonderfully made."
> Andy

Yep, nature took million of years to produce and adjust it,
and we dream to replicate it with a bunch of silicon gates.
But we did a lot of progress in less than 60 years with the
same silicon technology, perhaps in the next 600...

Wagner

1999\03\13@215507 by Wagner Lipnharski

picon face
Sean Breheny wrote:
{Quote hidden}

.. just because I can not see that wonder of nature, in retaliation, I
will anymore connect a resistor to any
power supply.... thinking about that, what would be the
lightning-show-dream of any resistor, AC or DC?

Hey, question;

Considering the chain movement of the electric current in a conductor to
be close to the
speed of light (298,000 km/s), what would be the max frequency the
conductor can transport, without create
collision or attenuation?

Just move your finger up and down, once a second, now twice a second,
now 5 times a second, 10, 20, you
already got a frequency where your muscle can not answer anymore, and
your finger will be stopped,
right? Then you feel a muscle in your arm get tense, this is circuit
overrun, total attenuation.

How we calculate the electric circuit overrun based on the max possible
conductor stimulation?

Wagner

1999\03\14@001254 by Sean Breheny

face picon face
Hi Wagner,

At 09:54 PM 3/13/99 -0500, you wrote:
>.. just because I can not see that wonder of nature, in retaliation, I
>will anymore connect a resistor to any
>power supply.... thinking about that, what would be the
>lightning-show-dream of any resistor, AC or DC?

I wasn't claiming that you could actually SEE anything...That's part of the
wonder of it,how much is going on BEHIND THE SCENCES that we cannot see,yet
know must be happening in order to produce the phenomena which we can
measure or see (heat generated by the resistor, current limiting
effect,noise voltage across it,shot noise at low currents,etc).

>
>Hey, question;
>
>Considering the chain movement of the electric current in a conductor to
>be close to the
>speed of light (298,000 km/s), what would be the max frequency the
>conductor can transport, without create
>collision or attenuation?

Well, that is an area in which I have little knowledge. Copper and other
common condutive metals have relatively the same behavior even up to light
frequencies (i.e.,the reason why you can make a mirror out of a piece of
polished copper or aluminum is because they are good conductors and there
is good electron mobility even at such high frequencies). You will run into
skin effect,which will cause the speed of propagation to decrease greatly
and the efficiency of the conductor to go down since only the very surface
is being used for conduction.

>
>Just move your finger up and down, once a second, now twice a second,
>now 5 times a second, 10, 20, you
>already got a frequency where your muscle can not answer anymore, and
>your finger will be stopped,
>right? Then you feel a muscle in your arm get tense, this is circuit
>overrun, total attenuation.
>

Yes, I can see that.

>How we calculate the electric circuit overrun based on the max possible
>conductor stimulation?

Just think of shining a flashlight at a piece of wire. The wire reflects
the light very well,which in this case corresponds to some conduction. If
you take xrays to the wire,I think some frequencies would be absorbed, some
refracted at specific angles (Bragg refraction - due to scattering by
electron clouds of atoms,sorta still like conduction,the electrons giggle
when struck by an xray photon),and some transmitted. Similar to xray
crystalography. As for gamma rays, probably just absorbed.

Gee, the title of this message is MDPs and Solar Panels?!! <G>

>
>Wagner
>

Sean

|
| Sean Breheny
| Amateur Radio Callsign: KA3YXM
| Electrical Engineering Student
\--------------=----------------
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
RemoveMEshb7TakeThisOuTspamcornell.edu ICQ #: 3329174

1999\03\14@005024 by Wagner Lipnharski

picon face
Sean Breheny wrote:
[snip]
>... (i.e.,the reason why you can make a mirror out of a
> piece of polished copper or aluminum is because they are
> good conductors and there is good electron mobility even
> at such high frequencies)...

also any heavy mass (density) as crystals including glass,
some polymers, high viscosity as oils, well, not good
electric conductors, but also has a great light reflector
matters.

[snip]
> ...how much is going on BEHIND THE SCENCES that we cannot
> see, yet know must be happening...

if anyway possible, I would like to see the electric reaction
that happens in a crystal receiving a pressure hit, because it
would resonate during some time, with zillions of interactions
between the shock waves and electric responses, until it stops.
It should looks like (in a poor example) the concentric waves
in the water, but in 3D reaction with polarized fronts...

... and we complain when a crystal needs 33pF capacitors...
hehe
--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:  http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\14@005854 by Sean Breheny

face picon face
Hi again Werner!

At 12:48 AM 3/14/99 -0500, you wrote:
>
>also any heavy mass (density) as crystals including glass,
>some polymers, high viscosity as oils, well, not good
>electric conductors, but also has a great light reflector
>matters.

Yes,but I think this has to do with permitivity. All of the things you
mention have quite high dielectric constants relative to air,so the
molecules are highly polarizable,and the electron clouds distort when the E
field of the light impinges upon them,and this electron motion opposite to
the impinging field produces the reflected wave in the opposite direction.
Still sorta similar to what happens when the light hits a conductor.

>
>[snip]
>if anyway possible, I would like to see the electric reaction
>that happens in a crystal receiving a pressure hit, because it
>would resonate during some time, with zillions of interactions
>between the shock waves and electric responses, until it stops.
>It should looks like (in a poor example) the concentric waves
>in the water, but in 3D reaction with polarized fronts...

Yeah,that would be a neat thing to see! I'm sure someone has done a decent
simulation of that. The piezo effect is under great study in the fields of
biomechanics and geology.

>
>... and we complain when a crystal needs 33pF capacitors...
>hehe

Hehehe,yeah!

>--------------------------------------------------------
>Wagner Lipnharski - UST Research Inc. - Orlando, Florida
>Forum and microcontroller web site:  http://www.ustr.net
>Microcontrollers Survey:  http://www.ustr.net/tellme.htm
>

Sean

|
| Sean Breheny
| Amateur Radio Callsign: KA3YXM
| Electrical Engineering Student
\--------------=----------------
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
spamBeGoneshb7spamBeGonespamcornell.edu ICQ #: 3329174

1999\03\14@210301 by Andy Kunz

flavicon
face
At 09:23 PM 3/13/99 -0500, you wrote:
>Andy Kunz wrote:
>> No wonder David, King of Israel could write, "I am fearfully and
>> wonderfully made."
>> Andy
>
>Yep, nature took million of years to produce and adjust it,

Better yet - It took God 6 days to speak it into existence in a perfect,
mature state, and here it's taken us 150 years to just begin to scratch the
surface of seeing the design of what He _implemented_ in that short atime.

Andy

1999\03\14@222203 by Gerhard Fiedler

picon face
At 20:58 03/14/99 -0500, Andy Kunz wrote:
>Better yet - It took God 6 days to speak it into existence in a perfect,
>mature state, and here it's taken us 150 years to just begin to scratch the
>surface of seeing the design of what He _implemented_ in that short atime.

well, i guess we don't know what his days are exactly... since he's not on
earth (probably, physically speaking), they're probably the rotation of
some other planet (if they are literal "days" at all), likely some other
solar system... might be milleniums or billeniums... (here we go again:
american or imperial? guess it doesn't matter here... :)

and then i wouldn't call it a "perfect, mature" state, looking at tv or
politics, for instance... (this thread as =two= OTs in the title, i guess
that's ok then to mention this un-word :)  more like a nice try :)

ge

1999\03\14@230009 by Wagner Lipnharski

picon face
Gerhard Fiedler wrote:
[snip]
> ... (this thread as =two= OTs in the title, i guess
> that's ok then to mention this un-word :)  more like a nice try :)
> ge

This is the dangerous of an open thread where mind floats, we ended
talking about religion or politics... not good, not good.
Two of the most explorive subjects... I avoid those as fire.
I think there is nothing else to talk about this subject, at
least for a while... :)
Wagner

1999\03\15@101413 by Andy Kunz

flavicon
face
>well, i guess we don't know what his days are exactly... since he's not on
>earth (probably, physically speaking), they're probably the rotation of

Actually, we do know.  The word used was YOM, which in the Hebrew texts
_always_ meant a 24-hour period.  They used other words to denote longer
periods.

>and then i wouldn't call it a "perfect, mature" state, looking at tv or
>politics, for instance... (this thread as =two= OTs in the title, i guess
>that's ok then to mention this un-word :)  more like a nice try :)

We are "after the fall of Adam."  Before Adam sinned, it was perfect.

Andy


  \-----------------/
   \     /---\     /
    \    |   |    /          Andy Kunz
     \   /---\   /           Montana Design
/---------+   +---------\     http://www.montanadesign.com
| /  |----|___|----|  \ |
\/___|      *      |___\/     Go fast, turn right,
                              and keep the wet side down!

1999\03\15@101416 by Andy Kunz

flavicon
face
>This is the dangerous of an open thread where mind floats, we ended
>talking about religion or politics... not good, not good.
>Two of the most explorive subjects... I avoid those as fire.
>I think there is nothing else to talk about this subject, at
>least for a while... :)

Some people don't like to go there because the implications are too
frightening (ie, there IS a God, and therefore I'm responsible for my
actions).  Others don't because it only shows their ignorance of that which
is truly important.  Still others don't because they don't want to deal
with it.

Andy


  \-----------------/
   \     /---\     /
    \    |   |    /          Andy Kunz
     \   /---\   /           Montana Design
/---------+   +---------\     http://www.montanadesign.com
| /  |----|___|----|  \ |
\/___|      *      |___\/     Go fast, turn right,
                              and keep the wet side down!

1999\03\15@135412 by Wagner Lipnharski

picon face
Andy Kunz wrote:
> >This is the dangerous of an open thread where mind floats, we ended
> >talking about religion or politics... not good, not good.
> >Two of the most explorive subjects... I avoid those as fire.
> >I think there is nothing else to talk about this subject, at
> >least for a while... :)
>
> Some people don't like to go there because the implications are too
> frightening (ie, there IS a God, and therefore I'm responsible for my
> actions).  Others don't because it only shows their ignorance of that which
> is truly important.  Still others don't because they don't want to deal
> with it.
>
> Andy
No Andy, I am talking that Religion and Politics are two subjects that
can lead very easily to fanaticism, and that is an uncontrollable out of
reality situation when you simply can not deal with.  A fanatic person
acts as with a blinder over his eyes, guided by something that is not
his own will, as a delusion world where there is no other option, except
the super polarized mind. A fanatic person is easily recognized because
he try to impose heavily and massively his belief. Those kind of people
exist all around, and you talk to them daily, so this is the reason I
avoid this subject, not because i am afraid or don't give the correct
importance, but purely because there is no sense to show yellow to
person that only wants to see green.  There are some religions around
that assume the World will end up at the end of millennium... I have a
neighbor, mature person, that affirms this belief looking straight into
your eyes... what they gonna do at January 2001? commit massive
suicide?  how can you deal with a person like that? Just 5 centuries ago
other religion ensured and applied penalties to whom that disagree that
the earth was the center of universe... At the political side, we could
se at TV a famous character asking "What do you mean by
-religion-?"...So as you can see, this is an endless and very dangerous
discussion that leads to nothing. I will not talk about this anymore
here, mainly because this is not the right place.
--------------------------------------------------------
Wagner Lipnharski - UST Research Inc. - Orlando, Florida
Forum and microcontroller web site:  http://www.ustr.net
Microcontrollers Survey:  http://www.ustr.net/tellme.htm

1999\03\15@152956 by Graeme Smith

flavicon
face
GRAEME SMITH                         email: TakeThisOuTgrysmithEraseMEspamspam_OUTfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Sat, 13 Mar 1999, Wagner Lipnharski wrote:

> All the parallel and serial effects you said about the visual system are
> purely mechanical,...

What about the organization of the neurons into collectives called nerves
and back out into cortex tissue etc?

I think that there would be more useful work done, if we didn't fly in the
face of ALL the constants, but accepted that there were some that are at
the very least, somewhat restrictive, and WORK AROUND them.

Instead of fighting time.... Why not consider it a Co-linear system,
each step takes time, but the process can be seen across many such time
lines, and the end result happens sooner, the more of it that can be done
in parrallel.

I think of it a bit like one of those conforming tools, that is made up of
a whole bunch of pins, that are allowed to slide back and forth in a
frame. A little pressure, against an object, echoes the relief of the
surface of that object....

If we accept that each step takes time, but that the co-ordination between
the steps is not rigid, and can flex to some extent, we get something a
whole lot more useful, than either a straight serial processor, or a rigid
parrallel processor.

>
> I was talking purely about the image caption and recognition mechanism,
> that has nothing to do even with
> the interpretation of the image.  When you see a car crash photo with
> blood and other things, it is repulsive
> much before you recognize exactly what it is.

How do you know its repulsive, unless you recognize aspects of it, that
"REPULSE" you? I think you are using too large a granulation on this,
Neurons are much less powerful than you seem to presuppose. Hebbian neural
networks require two or more layers, simply to react at all, HH neurons
may have a second order effect but only are valid in "Tissue" like
simulations.

The "Neighborhood" effect is not nearly as intelligent as you make it out
to be.


> Think about the "limitations" is very difficult when we don't understand
> not even a little part of the system,

??? WE DON'T?

You mean we have spent centuries, trying to understand something, and
FAILED COMPLETELY??? (Not Likely)


> or not even "dream" about how to replicate it, but there are no limits,
> except in out rational way to think.

I don't much like the "Pholosophy" of "Rationality" myself...

I think the GREEKS did us a bit of a disservice, by attempting to define
"Right Thinking" as some function of logic. While it might seem, well-
"Illogical" to redefine logic, a less formal system, is indicated, at
least in a functional manner by the relative successes of comercial
versions of "Fuzzy Logic". Micro-Chip, seems to value this contribution.

However there is no reason to get... well, "Irrational" about it, (he he)
and throw all the babies out with the bathwater... Lets consider that
there might be good reasons for the "Constants" if only to act as place
holders in our theories.... ;)

>
> Serial steps make part of a universe that uses time for the cause/effect
> process, I am not talking about it, and we will never reach MDP or some
> mysteries of the brain or space warp, while thinking in serial steps.
>

??? Why not? after all, serial steps are what got us here in the first
place, why can't it continue to carry us on? Remember the COLLECTIVE of
all scientists working on all the questions in the universe
simultaneously, is a good parrallel processor, complete with checks and
ballances, like this discussion.


> Some time ago I said here that the word "impossible" need to be
> re-spelled again as "difficult", because we are entering a human phase
> where technology allow us to see, to learn, think and conclude better.


And you don't think we are going to discover MORE CONSTANTS???

                               GREY

P.S. I think all this discussion about time is based on the illusion that
time is some secondary dimension... My personal beleif is that time is the
first dimension, so everything is imbedded in it, no matter how much it
depends on "Spacial" characteristics to achieve anything.

I would much rather have a physicist define motion in terms that do not
have to include mass, but still require energy, than to define space
without time.


1999\03\15@163011 by Gerhard Fiedler

picon face
At 13:27 03/15/99 -0700, Graeme Smith wrote:
>place, why can't it continue to carry us on? Remember the COLLECTIVE of
>all scientists working on all the questions in the universe
>simultaneously, is a good parrallel processor, complete with checks and
>ballances, like this discussion.

while it might be seen as a parallel processor, i wouldn't consider it as a
good one -- the communication is not up to it, yet. maybe we (that is, the
whole community) get there some day, but as of now, this "processor" works
highly inefficiently, due to lack of communication and community.

(i don't say there is =no= communication, but there is definitely less than
is desirable in an efficient system, and =much= room for improvement. a
technical system where the components communicate only as well as in most
social systems would probably not get over the first test stages... the
example of this list only supports my point. the mere fact that people --
including me -- regularily are stunned by the help they receive here shows
that it is an exception rather than the rule. and even this list is not yet
"the ultimate" :)

ge

More... (looser matching)
- Last day of these posts
- In 1999 , 2000 only
- Today
- New search...