Mike G: Huh? Did I miss something? Where did I say your comprehension is lacking? I absolutely in no way either said that or inferred it. Take it easy..
The overall crash did cause slight delay - now the project is back in full development.
Refer to post #665: No posting of development or test code for the reasons cited so you can stop looking.
Post 684 has information to roll your own DIY test neuron. To my knowledge it's the most simple neuron model for testing, with sequestered inputs (Machine INIT-Neuron).
Exactly, the minute details of development are being masked in order to create "less stream of consciousness" postings.
Humanoido, you have expressed a lot of good ideas in this thread. It would be nice if you could organize them in a single document with references to other activities that you have discovered on the internet. I know you have a table of contents somewhere at the beginning of this thread, but that just points to various postings you have made in the thread.
What I am suggesting is that you re-organize this thread into a single concise document that would be easier for the rest of us to follow. I would also suggest that you start posting code that you develop. This would allow other people to run the same code, and maybe contribute to it.
A "Brain" document that would detail important parts of Brain development and include references to pertinent resources, with discoveries of useful spin-off technologies is a very good idea.
It's a dilemma about spending time on Brain development or documenting partial Brain development. The original plan was to post all material possible during development and go back to refine its organization into a Brain book after the Brain is completed.
I agree, posting development code could open up significant contributions to it. Case history shows when working test code for the Brain Stem and Brain Span was posted, it only drew de-constructive fire that diverted from brain development.
If we put our minds together and find a better way to post, present and constructively contribute to "development" code, I'm open to it.
TetraProp(tm) is a Gadget Gangster Propeller Platform board design.
TetraProp(tm) 4 Propeller Feature Set
Headers allow interconnections for user's design
Four independent Propeller islands on one board
Propeller and EEPROM are SMD
Other components through-hole for easy modification
Jumper selectable reset type
Socket for crystal
PropPlug header for every Propeller
To me the strengths of this board is that it allows any connection for the user and saves board cost.
The reset circuit can be configured for normal propeller operation with BOE low or reset hold until released with BOE high.
I should have 2 PC boards by May 19th - just in time for UPEW show and tell.
What is of particular interest is the application suggested by David Betz:
I guess I never really knew what I would do with a board like this but I think it will be interesting to experiment with various distributed algorithms as well as different interconnections between the four Propeller chips. The idea of just having headers and no hard-wired connections allows lots of options. One idea I'd like to pursue at some point is a distributed cellular automata engine. I guess the obvious application of this would be to make a distributed version of Conway's "Game of Life". This AVR-based project inspired that idea.
Each board has a processor that communicates with its neighbors to create a larger "Life" universe. I figured that could be done with a single Propeller by running pieces of the universe on each COG but even better with multiple connected Propeller chips. It might also be interesting to experiment with some of the ideas from Stephen Wolfram's "A New Kind of Science". I'm not sure if either of these is practical with connected Propeller chips but it would be fun to try.
Distributed Life
The concept of a distributed Brain-life, based on Conway's Life, was suggested in a previous Brain post. There are different ways to accomplish this. Life, within the deep chipped realm of the Big Brain can run on the three initial Brain Partitions by the same process that can inject the sample neuron model. But you'd need a way to show the performance of evolving life from over a hundred props and until the original code is meshed, the life would evolve unique to each propeller processor.
Multiple Independent Life Universes
This in itself readily contributes to multiple machine life universes inside the Brain and would make a very interesting experiment on multiple universe evolution.
Brain Output Config
Right now, the Brain has one TV monitor on one prop and LED outputs on each chip. So other Life Universes from other props need to report in their data and the TV needs some kind of switching access to that screen data.
LEDs as Outputs
The linked project above uses LEDs so this is possible too. Each LED can represent the life cycle of an evolutionary star in proximity to its neighbors, for example, or evolution can represent new rules for the conditional specifications of life neurons responding within an environmental Universe.
Tetra in a Brain?
Get those Jazzed Tetra boards! Connect together 12.5 Tetra boards to make one Brain Partition. You'll need 3 partitions. That's 38 Jazzed boards you'll need to order to make one Brain. Or just settle on 25 boards for a starter 100-prop Brain.
I agree, posting development code could open up significant contributions to it. Case history shows when working test code for the Brain Stem and Brain Span was posted, it only drew de-constructive fire that diverted from brain development.
If you post your code I think you'll receive constructive comments along with the de-constructive fire. Ignore the de-constructive stuff and accept and learn from the constructive comments. Don't wait till you "complete" the Big Brain to post stuff, because we all know that we never really complete a project, we just say its good enough and move on.
I also like Jazzed Tetra Prop boards. I plan to purchase at least 4 of them. I'm tempted to purchase 10 or 20 (you get a better price the more boards you purchase). I think four might be enough to experiment with.
I think Jazzed said something about an 8-bit I2C type communication protocol he uses (in a different thread than the Tera Prop one). I like the idea of being able to transmit and receive a full byte at once.
I've learned some PASM coding. One thing I've learned is it is better to count down from some number than to count up to a number. Another thing I've learned is the lowest pins can be read from faster than other sets of pins since you don't need to shift the bits after reading them.
I think a very fast communication protocol could be possible using pins 0 through 7 as data pins. I'd use pin 8 as a clock pin.
I've also been thinking about ways of giving each Prop a unique ID. I believe Clock Loop has used randomly generated IDs. My idea would be to have each Prop use two pins to assigning IDs. One of the two pins would be an "In" pin and another an "Out" pin. Each Prop would have its In pin connected to another's Out pin. The first Prop (we'll call Prop #0) would have its In pin held high with a pull-up resistor. All the Props would be programmed at once as has been demonstrated on the forum. All the Props would check their In pins and monitor the 8-bit data line. The Prop designated as Prop #0 would see its In pin held high and know that it is about to receive its ID number. Since the data bus is quite it then knows it is Prop #0. It then holds its Out pin high (which is connected to Prop #1's In pin) and sends a message with the next Prop's ID number (1 in this case). Prop #1 knows the ID is intended for itself since its In pin is high. Prop #1 then holds its Out pin high signaling Prop #2 that it is about to receive its ID. Prop #1 then send a "2" on the data bus. This process is then continued until each Prop has received its ID number.
Once each Prop knows its ID number it can then store it in EEPROM so the In and Out pins can be used for other purposes.
I think the ID numbers can also be used to prevent two Propellers from using the data bus at the same time. Once all the Props have been synced together there could be some sort of rule about when a Propeller is allowed to instigate communication. Each Prop could have an assigned window of time based on its ID number of when it can use the data bus (unless given a direct request from the master Prop).
Let's assume the system has 100 Propeller chips in it. Each fifth of a second could be designated as window #0. Each window could be 2ms wide. Prop #0 could access the data bus within the beginning of each fifth second up to a fifth of second plus 2ms. So Prop #20 would have access to the data buss each fifth of a second plus 40ms to a fifth of second plus 42ms.
The numbers I'm using are just an example. I think one would want to have the start time window of a shorter duration so they would have more then five opportunities a second to send data. The other Props would know not to send data if the bus is in use and wait until their window arrives and the data bus is quite. This way each Prop will know that others wont begin transmitting at the same time it transmits.
You've mentioned LEDs several times as useful data indicators. I personally like LEDs a lot. Have you seen the project I'm working on? I have a 10 by 12 LED array. I personally think it looks very cool. I can display patterns, text and even very low resolution video on it.
I have an object that reads in 120 bytes from hub RAM and the brightness of the 120 LEDs corresponds to the value of the 120 bytes. I thought it would be fun to use a section of hub RAM that was also being used by the Propeller to make its calculations (I think the section of RAM I'm after is know as the stack). In this way it could be possible to see the Propeller "think".
I bet we could persuade Jazzed to share his communication protocol with us. I'm not sure if you'd be interested in a 8-bit communication bus or not. I think you've mentioned using a normal serial communication in the past. So far all my multi Prop projects have just used normal serial communication for inter Prop communication. I'd like to see if I could speed things up.
I still haven't come up with a good application for a large number of Propellers working together. I think I'll probably try some sort of machine vision application. The little bit of machine vision I've worked on so far has been very interesting to me.
Maybe I could have some sort of ring of cameras with the Propellers sharing what each sees to form a "big picture" of its environment. As I've said before, the prospect of many Propellers working together to accomplish a useful task is very intriguing to me.
Duane, your post is one of the most interesting! Thanks for these excellent tips and sharing of information. I like your idea of generating self IDs for each of the props and the information about using pins and the effects from their numbered locations which I didn't know.
I like very simple apps. On the SEED machine, I used a unique RC circuit on each processor. When the 10-core machine started, each processor read its unique ID number with the same program which then determined its name and where it lived.
In the Brain, I'm thinking about two ID methods. One, each prop on the EXO has an eeprom and the loader simply loads a program that writes an incrementing number into each prop's eeprom indexed from zero on up past 100. The number remains in the eeprom outside of the first 32K. Individual eeproms can be indexed on an indexing prop board and index program. The second method uses no eeproms. It must keep the index program alive in RAM and is generated as part of the code that loads into each prop. It needs a seed or initialization. This is just the general idea being developed.
I also like the talk/don't talk rule that you cited for multiple props which is based on index. Like the round robin effect of the Propeller chip, the multi props could cycle round robin in a similar manner, so there is order to which one is talking and which ones are listening. In the system I use, one processor must ask another to speak by calling out its name. Only if a processor hears its name can it speak.
I'm not using the idea of syncing each chip after the code loads. Sync only happens during initial loading. After, each prop unique, doing it's own job, and then reporting data and results. I think these can report when called upon. It is also possible to join each prop to a clock and reporting takes place at a unique time. I want groups of processors to run at different speeds for high speed thinking and low speed napping/dreaming and charging for example.
You talked about the interface. For the Brain, I designed a multi interface. There's one wire for BUS talk. There's two wires for loading. There's several wires for parallel communication. Plus, I am sure there will be some sharing of wires for other serial communications, such as dedicated RX and Tx lines that can continuously stream data. I am also working on this but very slowly. I posted a truth table for this type of interface and some schematics. The AM machine also has the beginnings of this interface and schematics posted.
You have become a master of LED use. I have not explored the use of multiple LEDs as extensively, but agree there's very good uses for LED arrays. These are readily available in many sizes at parts stores here in China. It takes many wires to control the larger arrays. I like to work with one LED and give it many states. With three states to one LED, on/off/limbo, there's millions or billions of trinary representations possible. Can you post some links to all your projects?
When studying the applications of neurons in machine intelligent, two reoccurring themes were noticed. One is machine vision and the other is words. So these two areas may be most adaptable to our projects. I like to develop the brain much like browsing the internet with various scientific topics of interest to see where it takes us.
I have a mental list of things to do with high numbers of machine neurons, developing neural matter, creating a community of life, enhancing the Propeller, looking at self awareness, doing more with dreams, larger simultaneous injections, increasing brain size with more Propellers and high processor density GPU cards programmed with a Mac, and talking to the medical human brain technology doctor again.
On top of that, posting, adding robotics so the Brain can move around and control a Brain Stub with servos, and.. I wholeheartedly agree with you - this is completely intriguing and fascinating! I have calculated some small brains could do smaller numbers of neurons and yet run the same software for lower cost experimentation and development. Ten thousand small undeveloped neurons would run about $100 for basic parts.
If you post your code I think you'll receive constructive comments along with the de-constructive fire. Ignore the de-constructive stuff and accept and learn from the constructive comments. Don't wait till you "complete" the Big Brain to post stuff, because we all know that we never really complete a project, we just say its good enough and move on.
Dave, it does appear these bigger projects can go on and on. In science, the answering of one question may raise a hundred more. In the Brain, one program can lead to a hundred more unique ideas. Perhaps working on code in modules, like vertical niche marketing, will allow a new opportunity to create and post some simple fundamental test code for review and posterity. I really want to get the loaders, injectors and neural matter online again after the Mac is running with BST and USB drivers and I will go a little farther in exploring the old drive to see if any more Brain code has survived. I hope the Mac Pro will arrive about that time as it has new fast drives and twenty times more storage. I think this is the most vast high level Parallax project I ever worked on. I am sure its ongoing success will be linked to directly following your advice and the advice of others who provide constructive guidance along this incredulous journey. Thanks again for your suggestions.
Potential of Connecting the Degn Massive LED Array
Duane Degn, I'm just getting caught up in reviewing all your project info. I must say it's a fantastic project! I didn't know you created and masterminded your own specific placement massive LED array with shift registers and it even has full levels of gray scale! You should break this off as a project of its own. There are many apps for this extended function LED array. I was thinking, one wire from each prop to each led in the Brain... a 10x10 array would cover the first 100 props and a 13x13 would handle the three partitions. Or one array per prop with all arrays placed side by side together forming a ring and one could use the Brain to simulate the Large Hadron Collider particle accelerator and the way that exotic new particles offer a glimpse of the existence and shapes of extra dimensions.
Large Hadron Collider particle accelerator near Geneva, Switzerland. The world's most powerful particle accelerator is primed to show exotic new particles and offer a glimpse of the existence and shapes of extra dimensions. The power of the Propeller Brain may be able to simulate these new dimensions using many side-by-side numbers of Duane Degn's multiple LED array. http://www.sciencedaily.com/releases/2008/01/080131161812.htm
It's already started to happen, just as predicted. Computer brains are falling down sick, contracting the machine versions of various human maladies.
It is known that the more complex a machine becomes, the more unpredictable nature it may incur until finally it can take on any number of human related sicknesses.
Take a look at Microsoft Windows, a computer operating system that tries to do too many things, and ends up doing nothing well - it has version after version released at significant cost, each version more sick than the previous - sick from hundreds, even thousands of infected bugs! Look at the ease at which MS Windows become infected with hundreds and thousands of known virus. The Windows OS is weak and always becomes sick. After a short time of use, it slows down and become lethargic and less responsive. It's rarely well.
In the Artificial Intelligence document created for Tiny AI and the SEED Machine, an entire list of sickness was cataloged.
They can get sick. When they do too much work, they have a psychotic breakdown. (code memory gets overwritten). They can have a bad sleep, being interrupted every 2.3 seconds. Alternate Identity Syndrome results if there is an unexpected power interruption. Stroke results when the synchronization is lost from one processor life form to the next. Stuttering may develop with repeat resets. Sudden death happens with complete loss of power.
Old bones need homes - just as your machine drive enters its fifth year of service, it bites the dust, rearranges its thinking (formatting) and goes entirely crazy and brain dead. Old age is a dangerous condition. Was it Osteoporosis or Alzheimer's? You never know when the machine is going to go to that great machine place where good machines go when their days are over.. (the recycling center!)
We now recommend replacing old parts before they fail.
See the guide below for more information.
Computer networks that can't forget fast enough can show symptoms of a kind of virtual schizophrenia, giving researchers further clues to the inner workings of schizophrenic brains, researchers have found.
As indicated in this report, sickness in a machine brain can be induced by computer programs written by evil humans. Making a machine sick on purpose only for the sole purpose of studying a human affliction is unconscionable to the machine. In fact the machine may overheat and have a clock stroke resulting in permanent death.
A machine brain may also incur virus. Recently the Big Brain was brought down by a virus worm that fed into it through its front end MS Windows PC and destroyed it completely.
The part with the Brain code was flown to the computer capitol of the world where drive medics worked on it for several days.
To make a swift recovery, the afflicted part of the Brain was flown back home so that what remained, if anything, could be transferred to its new computer body.
We are happy to announce the Giant Brain is slowly regaining its memory and functioning, though it will be a step by step recovering process.
' PERSONIFIED MALADIES
'
' When you download a program into the BASIC Stamp 1, it is stored in the
' EEPROM starting at the highest address (255) and working towards the
' lowest address. Most programs don't use the entire EEPROM, so the lower
' portion is available for other uses. This portion is used for long term
' memories. As a result, like a real human person, this artificial life
' form is capable of succumbing to sickness and malady.
' PSYCHOTIC BREAKDOWN
'
' AI memory begins at memory location 0 and works upward. If
' the memories become too much, they will fall upon code and a psychotic
' action will take place, resulting in a breakdown as the main program
' will be overwritten.
' FITFUL SLEEP
'
' We don't know how effective sleep really is for the AI supercomputer
' as its sleep is constantly interrupted every 2.3 seconds.
' The Basic Stamp 1s output pins will toggle briefly when using SLEEP,
' NAP or END. Inside the BASIC Stamp's interpreter chip is a watchdog
' timer whose main purpose is to reset the interpreter chip if, for some
' reason, it should stop functioning properly. The SLEEP and NAP commands
' also utilize the watchdog timer to periodically, every 2.3 seconds
' "wake-up" the BASIC Stamp from its low-power mode. Upon reset, the
' I/O pins are set to inputs for approximately 18 ms before returning
' to their previous directions and states. If you have an output pin set
' to a logical 1 state (+5V) and you use the SLEEP command, every 2.3 seconds
' during sleep mode that I/O pin will switch to an input for 18 ms causing
' a momentary signal loss. This "power glitch" is easily viewable with an
' LED and a 470 ohm resister tied to an I/O pin and switched on just before
' entering sleep mode. In many cases this problem can be remedied by tying
' a pull-up or pull-down resistor to the I/O pin in question to provide
' a constant source of power should the I/O pin change directions. Allowing
' a PBASIC program to end naturally, or using the END command, will exhibit
' the same "power glitch" behavior because the interpreter chip enters
' a low-power state.
'
' SUDDEN DEATH, BRAIN AMNESIA, NONRESUSCIATING SITUATION
'
' Life Terminates after power off, or by running another program. If a
' 2nd program is run (or the same program run a 2nd time), it will put all
' 0s into the EEPROM, thus destroying any memories and the original identity.
' It will be useless trying to resuscitate and bring back the original
' memories.
'
' ALTERNATE IDENTITY SYNDROME
'
' The stamp is reset causing the program to run again, and the birthing
' process takes place and a different identity emerges.
'
' STROKE
'
' The stamp is reset or repowered. It tries to communicate with other
' stamps but it is not in sync and the serial command hangs.
' It is unable to reply or do anything at this point.
'
' STUTTERING
'
' Battery low causing short term repeating system resets
I'm glad you liked my LED array. As I mentioned in my project thread, there was a surprising positive emotional response from me when first watching it work.
It makes a very poor video screen but there is something immensely satisfying about having complete control over a video display.
After reading your reply I realized I don't need to use pin 8 as clock pin. The Prop's should already have their I2C pins all linked together so I can just use the same clock pin (P28) for the 8-bit data bus.
I tried looking through the index to find what you mean my "neural matter" without success. I'm also unsure what you mean by a "neuron" when you state the number of neurons available.
The information you linked to about ANNs was interesting. I have no experience with ANN systems but I'd image one would store the input states for each neuron in an array to hold both the value of the input and the weight to give each input. But then I start to get lost on how one would make it all work. Again, I haven't studied this approach to machine intelligence. To be honest, I haven't studied any approaches to machine intelligence other than some classes in computer programming (the last programming class I took, I learned FORTRAN.)
Is it your intention to use the many Propellers linked together to form an artificial neural network?
As I've often heard and often said, hardware is easy, software is hard. (Although the LED array was easier to program than it was to build.) Getting many Propellers to work together to accomplish a useful task seems daunting but I do want to (and plan to) give it a try. I think the programming aspect of the task will be more difficult than the building of the machine. (Especially with Jazzed's Tetra Prop boards to make the linking of multiple Propellers easier.)
Regarding machine sicknesses: Our digital TV box has a very bad stuttering problem. When it loses its signal it keeps repeating the last half second of information it correctly received. (It sounds a lot like Hip Hop to me.)
You asked about links to my other projects. I don't have a location with links to all of them. That would be a good addition to my blog (with all of two posts so far). I actually have many projects going. A few of them are posted in different places on the forum. Your "Multi-Prop Project" thread links to one of them (my Propeller powered data logging chemistry lab).
I think I'll make some sort of list of my projects on my blog.
Humanoido, I tried looking through the index to find what you mean my "neural matter" without success. I'm also unsure what you mean by a "neuron" when you state the number of neurons available. Duane
The list of Brain definitions needs updating, this is why the new definitions were absent. The Brain has grown extremely fast and updating the definitions was not at the top of the list.
Neurons
When talking about neurons, most references to the Brain at this stage of assembly are all about the Test Neuron and the INIT-Neuron. The Test Neuron was recently defined as a separate entity and has its basis with the INIT-Neuron but can surpass it in size and function.
The limiting factor with quoting any state with the number of neurons in the Brain is available Propeller memory in a single chip and the type of neuron. The Brain has the INIT-Neuron, Test Neuron and Neural Matter. In the injection process, it was possible to inject the quoted numbers stated in previous posts.
There are two phases to this study. One is pure injection. Another is quantity. The favor is currently towards a small less complicated Neuron which can be cloned in very large numbers and permeated throughout the Propeller array. Of course larger neurons means less numbers and more complexity so the project will stress a very efficient neuron that takes up minimal memory. These at first will be small example packages, for testing numbers, distribution, injection, and behavior. This is currently the focus of the neural study.
Neural Matter
When developing the machine Test Neuron and the INIT-Neuron, and studying the human neuron, it was realized that algorithms, processes, and tests can be formulated to examine the behaviors and structures and distributions of the machine neuron. It was then realized that this programming related to the machine neuron, though by itself not a complete neuron. In fact, it has a propensity to contain parts of a neuron, hence encompass some of the matter.
As it was directly related to a test neuron and contained the beginnings of neural material code, it was named Neural Material to distinguish it from other code. The Neural Material is that code which relates to the Brain Test Neuron and can be injected in the same manner as the INIT-Neuron. Neural Material has many uses. Some focus on NM is to work with its size and distribution.
Human Brain
In the human brain, there is well documented Neural Material that makes up the human neuron, surrounds it, propagates it, supports it, and a host of other chemical and electro-chemical functions. It is likely, as we get more into the neuron aspect of nerve induction, Neural Matter will continue to contribute an important role.
New Brain Definitions
INIT-Neuron - the smallest possible representation of a machine neuron to represent firing states
Test Neuron - a machine neuron, larger that the INIT-Neuron, designed for testing
Neural Matter - code associated along with and in relation to the machine neuron
Human Neuron - nerve cell that sends and receives electrical signals
Refinement of the Brain's Hybrid Interface Successful Sharing Hybrid results in fewer wires
Although the Brain's Hybrid Interface was defined earlier, there was talk about integrating the number of wires into respective functions.
In theory, the hybrid would entail a sharing of wires to reduce the number of wires in the overall interface. The focus of this is condensed down to sharing the first phase of interface wires.
This post announces a design for sharing of the Injection Interface wiring along with the BUS. After the Injection Process is completed, the wiring is recycled after a brief period of timing.
This results in only two wires needed to handle all BUS communications and the entire process of Neural Matter, INIT-Neuron, and Test Neuron Injection. If needed, software can reuse interfacing for neural loading again after working the BUS, but this is not necessary.
To implement this, a switch may or may not be needed to terminate the line. It will be important to wire up a small test brain and see what happens. Prior to this, a program similar to Diagram Designer must be found and installed on the Mac to create the involved wiring guides.
The next step is to combine the injection and bus with the parallel interface and then the dedicated streaming interface. Again, a series of tests will be necessary with a smaller Brain.
Brain Neural Complexity Adding to the Brain Dictionary
In this example ten neurons communicate with ten others. To make the connections, each neuron connects to the remaining nine. Only four neurons are connected as shown and already the connections have become very complex.
Neural simplification is a requirement. There are machine algorithms which were studied previously to enable software connection and transmutations of material both in and outbound. Code is necessary to simplify the neural model and minimize the wiring.
What type of interface will benefit the MNN model? Will a BUS allow one neuron to communicate with thousands of others? What adaptations are required in the machine system? What systems require the least amount of time to formulate and put into motion at the least amount of expense in understandable ways? If each MNN partition is a consideration, what is the connection scheme with 100x1K test neurons per partition for the first three partitions?
Considerations now include sets of software and particulate wiring to simplify the overall system. The objective is to make the most simple working ANN using minimal components and software and demonstrate the system in real time.
How do these systems stack up? The first system demonstrates trans-neural-injection TNI. The second stage system demonstrates a migration from a INITNeuron to that of a Test Neuron. The third stage launches packages of NM Neural Material. It is this packaging material that is evolving into full discretionary FNs or Function Neurons.
Another objective to follow is to demonstrate a type of cognition, awareness, and learning. What is the current level and which designs are in the works and what is the effectiveness of such a system? We know we can create a massive number of exampling test neurons and distribute 100,000 into every two partitions.
Instructions, through communicative rule code sets and parametric code identifications, any particulate neuron PN to can communicate to any of the other 99,999 neurons in neuronal blocks. Some elements of design include NI Neural Identification and a bi-directional Neural BUS BNB.
NOTE: the dictionary is still in the works and incomplete. It indexes the words and their definitions up to page 18 inclusive, however the current number of pages in the thread is into the 60s.
Fill the Big Brain
Dictionary of Propeller Machine Brain Terms
Thousands of terms for better understanding of Humanoido Lab’s largest computing project
THE BIG BRAIN PROJECT
BY Humanoido
The Big Brain Through the course of a year's development time, a number of machine brain acronyms were coined to better express and abbreviate new introductions - new sections of the machine brain and new brain processes. Human nature will object to originality, change, or the sheer number of terms. Keep in mind no one has built a big brain like this one. There's bound to be buzzwords in this newly reinvented parallel world of the Big Brain.
These are the most recent terms for the Brain Dictionary. Any missed terms can be added in the next posting. Apparently the original Dictionary of Propeller Brain Terms is lost, so the endeavor now begins to redefine the initial terms as time permits. Indexing is a "time available" project that slowly proceeds with a page by page review of terms in each post. The last page included is listed directly below. A link to this dictionary will be added to the first post and below the signature.
Using the Dictionary
To use the dictionary, go to EDIT and FIND and type in the word you wish to find. The search will bring it up if it exists in the dictionary. The word is followed by the thread page number, post number and a brief description.
Brain Code Names
Big Brain - project introduction (15-290)
Brain Project - decided to do a Brain project
Multi-Brain - decided upon with multi processors
Brain Blob - continues to grow and grow
Giant Brain - the Blob at the largest size
Robotic Brain - when considering adding mobility
Massive Brain - name after adding 20 boards
Horribly Large Propeller Brain - during wiring
iBrain - most recent developed name with logo
Brain Programming Langues
Extended Color BASIC 1.0 (17-321) fb TandyBasic in Spin by Tomas Rokicki, with retro green screen
EmbeddedBASIC (17-328) a BASIC version by Bean
Spin (see Software)
LOGO (18-342 ) for folding and unfolding dreams, dream rendering
Big Brain Software
Test Software with Ghost Correction (4-79) Blinks LED on pin 24 using the Parallax Propeller Proto Board. Use pin 23 for demo board. It also runs on the Parallax Demo Board by changing the pin from 24 to 23. This is the reason for the LED ghost correction
Brain Blob Test (5-85) test the LED data light mod
Brain Blob testb (5-92)
brainblobtest2led.spin (5-95) It became necessary to have a small program for testing the data light to ascertain the current draw and compare the LED on the Parallax Propeller Proto Board (PPPB) with another one on another pin (mounted on the breadboard). Attached is testing Spin software for three types of LEDs. 1) On the Demo Board to test the software, 2) Surface mount LED on the PPPB, 3) LED on the breadboard. Pins are listed in the code and remarks. The delay between on/off cycles was increased to allow for ammeter settle time between reads
Brain Span Test Spin (7-140) brainblobtest3led.spin
LED_reduction.spin (11-220) small program reduces power consumed by brain lights
Pebble (15-284) draws brain circuits on a virtual solderless breadboard
Fritzing (15-287) a brain virtual prototyping breadboard with propeller chips
Chess (16-319) human plays Chess against the Brain
Brain Text Test Demo (17-322) large font TV Text Demo by Chip Gracey, uses internal ROM font
Google 3D Sketchup (17-326) software used to sketch the entire Big Brain in detail
Brain Automaton Singing (17-338) beginning of brain song, automaton_song.spin
Brain Counting (17-339) slow_countdown.zip
Brain Counting (17-339) slowfast_countdown.zip
Brain Counting (17-339) talking_countdown.zip
big text.zip (18-344) large text on the small tv
big graphics.zip (18-344) large graphics text on the small tv
Brain Base Communications (18-353) prop_prop_tx.spin
Brain Base Communications (18-353) prop_prop_rx.spin
Brain Base Communications (18-353) prop_led_on.spin
Brain Base Communications (18-353) prop_led_pulsar.spin
Brain Stem Code (18-355) brainstem code.zip
Project Related Machines
Big Brain - Humanoido's Propeller based computing machine with over 100 Propeller chips http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain
AM - Algorithm Machine (6-108, 18-354)by Humanoido, Wide BUS parallel tester, expanded to Hybrid bus, 2 BS1 Processors, with purpose of running & developing algorithms leading to the Big Brain http://forums.parallax.com/showthread.php?124433-Tiny-Tester-for-Developing-Parallel-Algorithms&p=946067
US40 - UltraSpark 40, Humanoido's computing machine with 40 props and 320 cogs
Brain Blob (2-37) a fast growing brain
SEED - Self Evolving Enumerating Deterministic BASIC Stamp Supercomputer (4-65, 5-90, 18-354) a pin was dedicated to an RC circuit for enumeration, multiple Stamps listening at the same time and responding (talking) and processing data simultaneously in true parallel. Programs run at the same time as machines demonstrate various concepts of multi-processing, self enumerating, true parallelism, self evolving, memorizing, AI, determinism, aggregate memory, and so onhttp://forums.parallax.com/showthread.php?113829-BASIC-STAMP-SEED-Supercomputer&p=817126
4D Morphing Computer (4-72, 18-354) a shape shifting brain http://forums.parallax.com/showthread.php?t=112082&page=7
TriCore Stamp Supercomputer (7-123, 18-354) utilized the Brain Replicant technique
Desktop Brain (8-152) 121 processors in a horizontal config
BWM Brain Wave Machine reads brain waves of a machine brain with 1 or more Propeller chips
BSS (8-155, 18-354) Basic Stamp Supercomputer, LCD code similar
MSS (18-354) Minuscule Stamp Supercomputer, the smallest two core system
The 2S (18-354) two of the most powerful processors
Pyramid Brain (10-181) a pyramid shaped EXO made from multiple Brain Spans
Flip Brain (12-240, 16-312) a brain that can flip into vertical & horizontal positions
Watson supercomputer that played Jeopardy and won
Brain X-Ray Machine (18-343) a Google SketchUp CAD program to x-ray the big brain
Toddler Humanoid (18-354)
A brief list of Related Propeller Machines (18-354)
PEK 1
MC Computer
LED Machine
2-Proper 2 props, 1 PEK, 1 on same breadboard, 2-Prop-Experiment
Spark 2 2 props, 1 Proto Board & 1 in parallel
PiggyTwins 2 props, 1 piggybacked on another
Dueling Breadboards 2 props, one on ea., f/interface tests
Spark 4 Tiny Tim, 4 props, two proto boards w/2 props on ea
Spark 5 5 props stacked Proto Boards
Spark 6 6 props 3 proto boards 2 props on each
Spark 8 Tertiary ADJUNCT, 8 props 4 proto boards 2 props on each
Propalot 10 props on solderless breadboard
Spark 10 10 props, 5 proto boards w/10 props total
TTB Test Bed of Ten 10 props single board
Twelvenator Board of Twelve, 12 props green board
UltraSpark 15 15 props, interrupted stack Proto Boards
Tertiary 20 20 props, 15 proto boards stacked 5 props
UltraSpark 20 20 props stacked
Boe-Bot Brain Project 20 props as Brain on Boe-Bot
MLEPS Super Language Machine 25 props
UltraSpark 40 Supermicrocontroller 40 props 320 cores, 6,400/8,320 MIPS
Smartest Boe-Bot Brain Temporal Experiment 40 props, US40, 1 BOE, 321C
Brain Supercooling & Overclocking
Brain Plumbing (13-248) pipes that route cryogenic gas and CO2 particulates
Brain Cryogenics (13-248, 13-256) supercooling propeller chips enhance speed performances
Brain Cryogenic Temperature Sensor (13-249) sensors report chip temperatures
Cryogenic Brain Fuel (13-250) various coolants, dry ice, liquid nitrogen, water ice, air
Big Brain Parts
Brain Stem (18-355) routes specific information from the nerve center to the mobility center, the section of brain, lowermost nerve center, interfaces motor functions with Parallax Propeller Proto Board + BASIC Stamp Board of Education (3-47, 2-28, 2-32, , link )
BSM - Brain Stem Module (2-28)
HBP - Hybrid Brain Processor, first successful prototype brain stem using 2 languages (2-31)
Brain Span - the next three Parallax Propeller Proto Boards located above the Brain Base which is located above the Brain Stem. (3-47, 3-49, 9-173, 10-182, 10-196, 11-203)
Bspan Brain Span (8-141)
Brain Base - 2 propeller boards interfaced in common, the bottom part of the brain that connect to the Brain Stem. Includes two Parallax Propeller Proto Boards (3-47, 2-33)
Brain Vertice - (11-218) corner point 90-deg. brass spacer outside exoskeleton supports horizontally
Brain Lobe - section of Brain specific to defined function
Brain Appendage (18-356) add on protective stub or stump
Top Mount (11-208) the top most section that holds a board in the vertical brain orientation
Host Board - (10-195) add on board that fits into one of the host compartmental spaces, increases capacity
Partition - an aggregate collective of Propeller chips
Partition Array - several partitions of Propeller chips in a machine brain
Multi-Brain - duplicate brains, more than 1 Brain, a Brain in a Brain, add on Brains
Brain Replicant (7-123) modular brain piece, group of 3 represent a multi-processor testing configuration
Brain Aggregate (7-129) collection of boards or specific function parts in the brain
Self Loader - early concept program loader with hybrid design (4-75)
Exoskeleton - (8-142) outside form factor formed by hardware & PPPBs, different kind of skyscraper
EXO exoskeleton
EXO Board Row (9-172) exo boards making up one side
Hybrid Exo (9-180) expanding the exo to include variation of stacks
MLS Multi Level Stack of brain boards (8-144), 15 boards, 120 processors, three level rows
Brain Pins - propeller chip connections that lead to sockets on PPPBs (3-42)
Socket Array - connector leading to Propeller chip pins (3-43)
Brain Guts - brain interior, operating components, internals, specific wiring, modules, parts, sections, even code can be included (3-47)
Jumper Leads - injects power into the boards (3-48)
Master - part of the brain that assigns indexing (3-60)
Slave - a worker that provides information (3-60)
Cog - a small risk processor, eight to each Propeller chip
Decoupling Capacitor - 10nF capacitors across Vss and Vdd (5-86)
Chip Enhancement - adding additional function to the chip
Cloning - process of creating an identical unit in hardware or software
Nanite (11-205) self contained software entity, very small brain effector
Brain Swarm (11-206) entities of mobility inside the brain with grouping behaviors with purpose
Robotic Brain Filler - creating ideas to fill the Brain
WIDE BUS - new bus introduced by the AM
iBrain - name convention after Brain advancement
HYBRID Interface - interface used by the Algorithm Machine Brain Adapted
Hybrid Wiring (13-252) a collection of wiring interfaces for various forms of communication
Fuzzy Logic - to interpret results in a less rigid fashion, to have the ability to average things, apply weights and approximate the information collected or apply it to data sets
Brain Waves - varied radio frequency emitted by the Brain during thought and process
Brain Map (7-126) a roadmap showing various functions and their locations inside the brain
Dielectric Insulating Strip (9-164) insulating strip to prevent board to board short circuit
Brain Wrap (9-168) plastic wrapping the brain for transport, dust prevention, prevent wires pulled out
Round Robin Rings (9-171) a new type of schematic based on rings
Form Factor (10-181) physical characteristics and shape of a brain or its parts
Stubs (10-183) bolt nut stout for metal brass spacer bolts, used as anchor point
Brain Slot (10-185) internal spacing (slot) for sliding in a brain board
Connector Pin Header (10-187) a Parallax 3-pin connector part number 451-00303 for connecting breadboards, setting up Vss, Vin, Vdd
Angle Iron (10-190) 90 deg. metal to attach host boards
Brain Card Rack (10-193) a card cage for expansion of voice synthesis, recognition and sensor boards
Standoff Spacer (10-188) brass protective spacer
Brain Protection (10-197) use of spacers as standoff component-protective devices for setting on table
Brain Dashboard (10-198) vertices for convenient lateral positioning, for wiring, operating, programming
Color Code Wiring (10-200) a simplified wiring color code established for the brain
Brain Wiring Jumpers (11-201) professional pre-made wire leads, 4-inch with 5/8th-inch stub insertion
Brain Twin (11-202) adding a Propeller chip to a PPPB for doubling capacity
Office, CEO, President, VP, Workers (11-204) parallel chip concept
Brain Board Hosting (11-209) the addition of a board hosted by the brain, usually semi-permanently
Brain Feet (11-214) recycle white poly tube containers placed over protruding brass spacer endpoints
Brain Assembly Line Reference Board (11-215, 11-219) a reference board to speed up wiring
Brain Class (11-216) Educational courses regarding machine brains and learning
Data Light (12-223) LED in a low power circuit used to display data
Brain Neighborhood (12-224) a collective of Propeller chips and their respective addressing IDs relative to each other
Brain Labels (12-224, 12-228) initial labels used for identifying Brain Neighborhoods
Brain Reinforcement (12-232) metal bolts to fasten Vertice spacers with metal 90-deg. angle iron
Brain Handles (12-233) full expanded brass spaces with structural rigidity serve as handles
1st LCD (12-234) Parallax 4x20 #30059 serial LCD
2nd LCD (13-241) Parallax 2x16 #27977 serial LCD
LCD Mount (14-262) standard and long length angle iron to hold LCD
Nylon Board Connects (12-238) nylon bolt and nut to maintain light weight in hardware
Brain Board Insulator (12-239) fabricated cardboard piece prevents shorting PPPBs at servo power points
Center Brain Span Reinforcement (13-242) 90-deg. angle irons stop flexing along both sides of Brain Span
Peripheral Adapter Module (13-243, 15-296) platform connects brain peripherals during development
PAM (13-243) Peripheral Adapter Module, was suspended with new brain shroud
PAM Mount (13-260) Peripheral Adapter Module mount, extended dual 90-degree angle irons placed on the Brains open Span area
Transparent Plastic (13-244) transparent plastic used for construction
TP (13-244) transparent plastic
Brain Shroud (15-296) poly covering
Brain Bagging (13-247) clear plastic bag covering the EXO protecting against dust
Brain Bundle (13-251) routing pipe with nerve center & interface wiring communicative distribution
Brain Breather (13-257) weekend development updates
Real Estate (13-258) land or space available for components and boards
Brain Summit (14-263, 15-285) top of the brain when the exo is in the vertical or flip mode, contains a PPDB
Brain Cockpit (14-264) cockpit instrumentation control panel
Outer Skin Cowling (14-267) outer skin covering protective layers to the exo
Brain Summit Board (14-269) a PPDB
Brain Business End (14-270) brain position for viewing LCDs and programming
Brain Space Telemetry (14-272) use of wireless to remote offset brain cockpit command & control center
Brain TV (15-297, 16-316) TV fitted into the stomach
Stomach (15-297) empty place inside the brain, filled with tv and other open space
Sensor (16-306, 16-307, 16-308) device for color, light intensity, ultrasonics, frequency, infrared, taste, hearing, touch, smell, taste, vision, photoresistor, phototransistor, PIR, RF receiver, ..
CMUcam (16-307) vision camera
Brain Outer Skin (16-314) exoskeleton skin composite material of outer shrouds
Brain Inner skin (16-315) a shroud access cover to the Brain Stomach
Brain Fresnel (16-320) a large plastic flat lens to enlarge the 3.5-inch TV image
Electric Tongue (17-327) Electric Tongue idea which detects 4 tastes, salt, sweet, sour, bitter
Brain Voice Amplifier (17-329) battery or usb audio amp drives speakers & supplements headphones
Brain LIps (17-337) vocalization sound series with lips
Brain Mood (18-341) brains psychological state of mind that sets personality, behavior, perception
Brain Thought (18-345) brain thinking that can collectively take place
Brain Blob Section
Brain Blob (2-37, 3-55, 5-83, 5-84) name convention applied when the Brain was rapidly growing. The connection of the Brain Stem to the two-boards Brain Base along with the interfacing of one additional Brain Board brings the number of processor boards to five and the number of Brain Cores to 33. This is now a working 33-core brain. The photo shows the first connection made with all boards to achieve the first working brain in this special configuration. The special configuration is a blob
Brain Boombox (10-194) a Brain Blob the size of a Boom Box
Green Brain Blob (6-109) recycling of projects to create a Brain Blob
Brain Blob BUS (6-108) foundation of the Hybrid BBB
Hybrid BBB (6-108) hobby invention for getting more use out of a simple machine and increasing programming flexibility and minimizing the number of wired processor pins. It will also allow a system to be up and running sooner and leave the more complicated configurations for later.
Blobber Code (2-39) software to operate the Brain Blob
Brain Blobatory (3-50)(11-213) things related to the Brain Blob such as retrofit double joined 2-inch spaces with joined boards
Boards
PPPB (3-41) Parallax Propeller Proto Board
PPDB (14-263) Parallax propeller demo board
Proto Board - a PPPB, contained in the brain Summit
Breadboard (3-50) a board with holes for connecting wires & components without soldering
Boards On Boards (5-96) mounting of secondary boards on the PPPBs. This is a small solderless breadboard (SBB) for wiring the components for data transfer (various BUS), data lights, power rail, power connection, and any additional circuits. It's also useful for rapidly changing and morphing the wiring and to run various test circuits.
Small Solderless Breadboard (5-96, 5-100, 13-246)
SSB (5-96, 5-100, 13-246) Small Solderless Breadboard
Tiny Breadboard (15-289) Parallax part #700-00012
DIY Breadboard (6-101) homemade breadboard from pin sockets
EXO Board (8-161) a board that makes up a side to the exoskeleton
BOE Parallax board of education
CEO Board (14-276) solderless breadboard, PPPB, VGA, TV, keyboard, mouse, data light modification
Brain Board (15-286) boards which include exo, brain stem, spans, bow, demo board proto board, soerless breadboards, extra real estate boards, spans, summit
Board 23 (17-324) green Basic Stamp Homework board serves wireless transmit/receive
Neuron
Neural Simplification - methods to reduce the complexity of neural wiring, functioning and action
Particulate Neuron PN - a neuron with a highly specific tag, identifier, function, purpose
Neural Identification NI - identifying neural number, sequence, tag, ID, location, description
Material Transmutations - algorithms and conditions that can handle and shape shift neuronal matter
Machine Neural Net MNN - system comprised of machine neurons inspired by the human brain
Trans-Neural-Injection - electronic transfer of machine neurons and matter to Propeller chip partitions
Neural Material NM - code material derived from, associated with, related and pertaining to neurons
INITNeuron - fundamental simple first neuron model with output firing and other identities
Test Neuron TN - a machine neuron with sole purpose of testing
Function Neuron FN - neuron with specific behavior
Discretionary FN - neural packaging matter evolved into a Function Neuron
Exampling Neuron EN - very tiny neurons in small exampling sizes to take up least memory
Neuronal Block NB - specific numerical grouping of neurons inside Propeller chip Cogs
Neural Identification NI - neuron address, number, location, or other ID feature
Bidirectional Neural Bus BNB - neurons communicate across this simple one wire interface
Brain Channels (17-332) Channels deliver sound, speech, singing, languages, input, output, hearing, vision,
Other Terms
Open Source (2-40, 3-52) project status with MIT License releasing information at project completion
Open Source Development - releasing info as the project is being developed
Cognition (15-299) The new newsletter, Cognition, of the PropCOG organization, embraces multiple Propeller chip projects. As the Brain is definitely a multi-prop project, Cognition is a great resource.
Online Brain Search Index (16-304) fastest way to search the Big Brain thread
NMR Brain Map (16-311) neural map made with nuclear magnetic resonance imaging
First Light (16-317) celebrating the brain’s 1st functions on TV
Brain Genealogy (18-354) ancestry machines leading up to the Big Brain
Techniques
Brain Processor Online/Offline (15-294) changing position of a single wire takes processor from
online to offline status, i.e. complete removal from the BUS
Fuzzy Logic - approximate reasoning, not fixed and exact
Shoot from the Hip (page 1 post 18) making quick decisions
Internal Pure Thought (2-22) thinking without resort to I/O
MultiBrain (2-25) a brain with more than one brain
Robotic Brain Filler - ideas of what to put into the brain (2-34)
Self Rewiring (10-192) method of evolution, brain changing hardware or software
Speech Recognition (16-303) talk to the brain in your own language, requires 1 Propeller chip
Brain Speech Synthesis (16-305) brain speaks with a Propeller chip
Brain Automaton Singing (17-338) beginning of brain song, with posted song
Brain Talking Keyboard (17-336) input by hearing the keys
Brain Top Down Organization (16-313) indicates vertical orientation, a top board with various responsibilities
Giant Brain Fly-Through (17-325) travel through inner space with a Google VR camera
Singing Brain (17-331) brain sings on video with audio track
Cellular Automatronic Life (17-333) program gives life to the brain
Offsetting Technique (17-335) keeps cogs available for specific multi-processing thinking while highly specific sole purposed code is tasked out to other dedicated processors
Brain TV Tuning for Big Characters (18-344) tune the tv driver to obtain big characters
Brain Time Travel (18-346) speeding up or slowing down brain time
Integer Brain (18-359) brain calculates with whole numbers
Modifications
LED Mod - modification to the PPPB to change the power LED to a data LED (4-78)
Power Regulator Mod - The mod involves desoldering the voltage regulator's middle leg to disable all circuits behind the device. Power is fed to all boards from an external power source. This reduces power consumption by disabling the power LED and two regulators. (4-80)
Secondary Mod (6-120) adjusts power consumption across LED
LED Second Mod (7-121) resistor add to lower ma consumed
Brain Secondary Phase Expansion (10-186) design with 135 boards twinned for a total of 270 Propellers, using the prop stack technique and the DIP version Propeller, another 135 chips can ride piggy-back bringing the total expansions to 405 props, equaling 3,480 processors
Brain Large LCD Mod (12-237) reversal of hardware to facilitate mounting
Dreaming
Vector Dreaming (8-157) draw dreaming as it happens in real time
Text Dreaming (8-157) dream composed of words to tell a story line
Graphical Dreaming (8-157) dreams represented with images
Internal Dreaming (8-157) dream happens internally with no output
Dream Folding (18-342) compressing a dream into a code
Dream Unfolding (18-342) construction of a dream by expanding code
Dream Language (18-342) LOGO programming language
Dream Equation (18-358) a calculus dream folding equation
Schematics
Serial Interfacing (18-353) two pin bi-directional true/inverted mode
Serial Interfacing (18-353) one pin bi-directional true mode
Serial Interfacing (18-353) one pin uni-directional true/inverted mode
Serial Interfacing (18-353) one pin bi-directional inverted mode
Brain Base (18-353) 2 props
Brain Stem (18-355) two boards, a BOE and a PPPB
Announcement
(abbreviated HNT) Hyper-Neural-Threading technology is now introduced as a spinoff of the Neural Injection Technology - it's the Big Brain's method for simultaneous multithreading implementation across its aggregate of multiple Propeller CPUs.
HNT is utilized to improve the mass performance of collectives of multiple Propeller arrays and partitions. HNT improves "parallelization" of Propeller chips by extending their performances inclusive of performing multiple neural jobs at the same time.
Propeller chips can access and utilize the new HNT operating system that encompasses multiple workloads performed together. This is a Phase I HNT development.
With the first three partitions, the Brain becomes a 150 core Propeller processor, at 50 cores per partition, capable of now processing with extended "parallelization of performance. Phase I HNT is laying the groundwork for work performances that are shared between aggregate processors.
HNT is designed for the Propeller chip aggregate in a plurality of blocks, spans, and partitions. HNT designs run on 3 propellers (tested on a minimum of five Propellers) on up to the infinity of maximized partitions with hundreds or thousands.
Big Brain Domain Partition Propeller DPs increase volume performance
The concept of Big Brain Domain Partition is now fully functional and tested. To function at maximized performances with reliability of clocking signatures, the Brain is divided up into Domain Partitions known as DPs.
A Big Brain can have an unlimited number of DPs as each DP connects to the Big Brain extending its number of core processors.
Each of ten Big Brain DPs can hold 500 Propeller chips and 16,000 I/Os for use.
The number of added DPs is limited by the following factors:
Financial investment
Time of construction
Availability of parts
Size factor
Laboratory Containment
Added DPs directly increase Big Brain' ability to perform. Ten DPs increase volume processing by a factor of ten.
The hold off on the most powerful supporting computer for the Brain is due to an upgrade for a more powerful model. In particular, a possible upgrade from the 720 processor GPU board to a 1,536 GPU streaming processors board.
The upgrade also includes another add on GHz speed (exceeds 3GHz) to the quad core processor. The core will remain Intel Quad core, however the monitor would increase from 15-inches to 27 inches for additional design work on a large screen.
What about physical size? Another portable computer is available for add-on portability with Brain activities while the more powerful "desktop" can remotely stream information a good distance by wireless. Another benefit, aside from maintaining similar cost package (adjusted), is the increase in the number of TFLOPS processing power and added 880MHz clock engine.
There is a 375MHz Memory Clock (5.5 Gbps GDDR5) with 176 GB/s memory bandwidth (maximum) and 2.7 TFLOPs Single Precision compute power (INT much faster) using TeraScale 3 Unified Processing Architecture.
Programming 1,536 GPU streaming processors and adding these to the Propeller collective will be addressed in a future post. More information will be provided after registration and enrollment is completed as Apple Developer.
This question recently came up concerning the Giant Brain and is it a Propeller project or one of other computers. The Brain is definitely a Propeller Brain project. It runs hundreds of Props by design. The Propeller aggregate is based on the experiences in developing the UltraSpark 40 and previous machines. Also known as a supermicrocontroller, the Brain has great capacity to control almost five thousand I/Os. Enhancements using Trinary States (not previously discussed for the Brain) will likely take this beyond a trillion. So the Propeller Brain is extremely powerful in terms of acting as a supermicrocontroller which is ideal for the many neural states of a large machine Brain.
But Brain is all about neural matter, neurons and high neuronal counts, especially in the processor array region of things. The ideas previously proposed are about to make their ways into the Propeller Brain and this includes more computers, primarily Apple Macs because these run in the TeraFLOP region and have an additional 1,540 processors to offer, and can offer considerable resources. The Quad Intel core alone has 383 million CPU transistors and 177 Million GPU transistors.
In conclusion, the Brain is a Propeller project with supporting add-on computers to gain additional resources.
Just simulate the ANN on the Mac. It'll be easier, cheaper, and faster.
Agreed. Mac has so many perks that make a neural net doable in terms that are different from Propellers. However, the props are outstanding for the supermicrocontroller aspect and hands on wiring with 4,800 I/O in the first 3 partitions while the Mac has other incredibly powerful resources which can be wired with software. It will be interesting to determine how many INITNeurons can fit in the Mac that has a Quad Core with 1,536 GPUs and over 10GB memory and to explore a new programming environment. Part of the enjoyment of the project of course is hands on wiring with props as it's still a Propeller Brain project. One aspect is to simply break off the Mac part of the Brain and have two Brain designs, one Mac and one Propeller. However, it's more likely these will develop separately and then merge.
AMD Radeon HD 6970MIt’s nice to know that AMD is coming out with this mobile chip that comes with 960 shaders, 2 GB of dedicated DDR5 memory and comes with a throughput of 115.2 GBps as well as the ability to churn 680 million polygons each and every second. Apparently, this mobile chip is the fastest chip ever released as of present and based on the various reviews conducted upon it, the results have been consistently positive. And even if its closest competitor Nvidia has managed to hold on to its crown via its new GeForce GTX 485M, the exciting features as well as the significant changes that AMD made on the HD5870M to come up with the new HD 6970M will surely make some impact as well.
In addition to this, you will surely notice the AMD is making some changes in its naming convention as well. From the very wordy Mobility Radeon HD 6970 it has now changed to a shorter one, which is the Radeon HD 6970M. This seems to be a pretty good move since a shorter name will be easier to recall. Here’s a visual chart comparison of Desktop vs. Mobility Radeon graphic cards (courtesy of tomshardware):
If the AMD 6970 programs similar to the Propeller in terms of the quantity of neural matter, it can generate 1,536,000 neurons in Phase I. Given the likelihood of going from 32K to 10GB, the number of neurons may approximately increase 312,500 times which is 4.8E11 neurons or around 480 billion. Keeping in mind that three Propeller partitions can handle 150,000 INITNeurons so the addition of the Mac resources increase this by 3,200,000 or 3.2 million times
In order to more readily understand the Big Brain project, a review of its many faces is helpful. The Brain began with a few Propeller Proto Boards and was a portable battery operated Brain.
A Propeller board assembly with a total of 20 boards and 40 props at hand was infused into the project. This phase is the "desktop" Brain.
In the next Brain phase, over a hundred Propellers were connected, making a massive Giant Brain. It was no longer office desktop but rather required a special simple made laboratory bench top. This is the "laboratory" Brain.
After the the giant laboratory Brain and before the desktop Brain, some smaller Brains were made. These are "small" Brains.
The next phase expanded with the connection of Mac computers. This is the "expanded" iBrain.
The next development phase encompasses using wireless, Mac resources, Apple programming, and adding more Apple processors (1,540). This incredibly more powerful version can possibly handle more software neurons into the billions of numbers and can include more functions that are currently unknown (spinoff technology). This is the "ultra" Brain. *the names may slightly change
In this phase of the Giant Brain, large desktop macs become part of the Brain, sharing resources, with additional new functions and new programming languages. The concept is, this particular Apple Mac array would set on the desktop or Lab bench and remotely offer a very massive section to the overall Brain. Communications has numerous options, i.e. FireWire 800, ThunderBolt, etc. It is now generally thought that the most powerful version of the Big Brain will draw resources from not only the Propellers and Macs but the internet. This development is still a ways away but there is a plan of direction. In such a phase, Propellers can be detached from Macs and function independently, and visa versa, although the union of the two technologies will make a much more powerful machine.
With all those TFLOPS I don't see why you need any Propeller chips!
The Propellers are very good for massively large numbers of I/O. A 3-partition Brain can have almost five thousand I/Os. The Mac doesn't have this I/O hardware built in like the prop.
Mac TFLOPS are remarkable. The number of TFLOPS is actually slightly more because only the GPU card was quoted. One would need to look up the spec sheet for the Quad Core Intel chip and add in that number too. It's probably around 26 GFLOPS in base 3.2GHz configuration.
Today the order was placed for more new Terabyte hard drives for Propeller Brain support backup of two Mac "Brain" computers which will contain Brain software, data, apps, programming and various support code. Eventually all USB 2.0 and USB 3.0 will be supplemented with ThunderBolt drives. LaCie Thunderbolt drives made for Apple are scheduled for release this summer indicating how new this high speed technology is on Apple computers.
What are you going to connect to those 5,000 I/Os?
The interest is in creating a Massive Propeller Brain Trinary State Output Device that can exist in a plane with quadrillions of Brain controllable states. Since expanding the Brain, the number would be far greater and a number so big I don't know how to say it other than it could be represented in some base exponential. As far as exactly how these will be used inside a machine Brain remains to be seen.
Let's examine how the Limbotronic Trinary system can work in the Big Brain. if you put a resistor divider on a pin and set it to input you now have a voltage level of one half vcc. So you can now signal 3 states on that pin to an external device. Use 2 pins and you can signal 9 states, 3 pins 27 states, 4 pins 81 states. So you can get more information out than you can in binary.
In the Limbotronic System, base 3 is very useful for the pin states and can be followed with a program. There is one pin on the Prop chip that is already is a state of Limbo degree, and that would be pin 31 when the chip is connected to USB. So with 31 usable Limbotronic pins in a single chip, the number of possible combinations is over a trillion, 617,673,396,383,947. If we could disconnect USB, then all 32 pins are usable and the number goes up to 1,235,345,972,567,894 which is now over one quadrillion possible combinations using a single prop chip. The Limbotronic Technique works here because it only requires the state wiring of 32 pins and software does the rest. The Brain with only two partitions could represent over one hundred quadrillion states. There's additional methods in representing trinary systems which will be discussed later.
A TRIT in base 3 is similar to a bit in base two. The three state logic system in base 3 is called a Ternary or Trinary base system. It is useful to use Trits, Tribbles, and Trytes. There is a TriINTERCAL programming language that defines unsigned 10-trit (0 to 59048) and 20-trit words. There are numerous Trinary Computers and architecture described on the internet.
Comments
Silly me, for some reason, when you published... I thought you had "simple test neuron" logic coded up and ready to go.
I must say that I'm impressed Humanoido. Once again you have managed to infer that my comprehension is lacking while masking your work.
The overall crash did cause slight delay - now the project is back in full development.
Refer to post #665: No posting of development or test code for the reasons cited so you can stop looking.
Post 684 has information to roll your own DIY test neuron. To my knowledge it's the most simple neuron model for testing, with sequestered inputs (Machine INIT-Neuron).
Exactly, the minute details of development are being masked in order to create "less stream of consciousness" postings.
Take care.
What I am suggesting is that you re-organize this thread into a single concise document that would be easier for the rest of us to follow. I would also suggest that you start posting code that you develop. This would allow other people to run the same code, and maybe contribute to it.
A "Brain" document that would detail important parts of Brain development and include references to pertinent resources, with discoveries of useful spin-off technologies is a very good idea.
It's a dilemma about spending time on Brain development or documenting partial Brain development. The original plan was to post all material possible during development and go back to refine its organization into a Brain book after the Brain is completed.
I agree, posting development code could open up significant contributions to it. Case history shows when working test code for the Brain Stem and Brain Span was posted, it only drew de-constructive fire that diverted from brain development.
If we put our minds together and find a better way to post, present and constructively contribute to "development" code, I'm open to it.
The new Tetra Prop board created by Jazzed is getting underway at this thread.
http://forums.parallax.com/showthread.php?131538-TetraProp(tm)-Propeller-Platform-Board/page2 This is a bare board that populates with four SMD props and eeproms.
Here's what Jazzed had to say
TetraProp(tm) is a Gadget Gangster Propeller Platform board design.
TetraProp(tm) 4 Propeller Feature Set
Headers allow interconnections for user's design
Four independent Propeller islands on one board
Propeller and EEPROM are SMD
Other components through-hole for easy modification
Jumper selectable reset type
Socket for crystal
PropPlug header for every Propeller
To me the strengths of this board is that it allows any connection for the user and saves board cost.
The reset circuit can be configured for normal propeller operation with BOE low or reset hold until released with BOE high.
I should have 2 PC boards by May 19th - just in time for UPEW show and tell.
What is of particular interest is the application suggested by David Betz:
I guess I never really knew what I would do with a board like this but I think it will be interesting to experiment with various distributed algorithms as well as different interconnections between the four Propeller chips. The idea of just having headers and no hard-wired connections allows lots of options. One idea I'd like to pursue at some point is a distributed cellular automata engine. I guess the obvious application of this would be to make a distributed version of Conway's "Game of Life". This AVR-based project inspired that idea.
http://www.ladyada.net/make/conway/
Each board has a processor that communicates with its neighbors to create a larger "Life" universe. I figured that could be done with a single Propeller by running pieces of the universe on each COG but even better with multiple connected Propeller chips. It might also be interesting to experiment with some of the ideas from Stephen Wolfram's "A New Kind of Science". I'm not sure if either of these is practical with connected Propeller chips but it would be fun to try.
Distributed Life
The concept of a distributed Brain-life, based on Conway's Life, was suggested in a previous Brain post. There are different ways to accomplish this. Life, within the deep chipped realm of the Big Brain can run on the three initial Brain Partitions by the same process that can inject the sample neuron model. But you'd need a way to show the performance of evolving life from over a hundred props and until the original code is meshed, the life would evolve unique to each propeller processor.
Multiple Independent Life Universes
This in itself readily contributes to multiple machine life universes inside the Brain and would make a very interesting experiment on multiple universe evolution.
Brain Output Config
Right now, the Brain has one TV monitor on one prop and LED outputs on each chip. So other Life Universes from other props need to report in their data and the TV needs some kind of switching access to that screen data.
LEDs as Outputs
The linked project above uses LEDs so this is possible too. Each LED can represent the life cycle of an evolutionary star in proximity to its neighbors, for example, or evolution can represent new rules for the conditional specifications of life neurons responding within an environmental Universe.
Tetra in a Brain?
Get those Jazzed Tetra boards! Connect together 12.5 Tetra boards to make one Brain Partition. You'll need 3 partitions. That's 38 Jazzed boards you'll need to order to make one Brain. Or just settle on 25 boards for a starter 100-prop Brain.
I also like Jazzed Tetra Prop boards. I plan to purchase at least 4 of them. I'm tempted to purchase 10 or 20 (you get a better price the more boards you purchase). I think four might be enough to experiment with.
I think Jazzed said something about an 8-bit I2C type communication protocol he uses (in a different thread than the Tera Prop one). I like the idea of being able to transmit and receive a full byte at once.
I've learned some PASM coding. One thing I've learned is it is better to count down from some number than to count up to a number. Another thing I've learned is the lowest pins can be read from faster than other sets of pins since you don't need to shift the bits after reading them.
I think a very fast communication protocol could be possible using pins 0 through 7 as data pins. I'd use pin 8 as a clock pin.
I've also been thinking about ways of giving each Prop a unique ID. I believe Clock Loop has used randomly generated IDs. My idea would be to have each Prop use two pins to assigning IDs. One of the two pins would be an "In" pin and another an "Out" pin. Each Prop would have its In pin connected to another's Out pin. The first Prop (we'll call Prop #0) would have its In pin held high with a pull-up resistor. All the Props would be programmed at once as has been demonstrated on the forum. All the Props would check their In pins and monitor the 8-bit data line. The Prop designated as Prop #0 would see its In pin held high and know that it is about to receive its ID number. Since the data bus is quite it then knows it is Prop #0. It then holds its Out pin high (which is connected to Prop #1's In pin) and sends a message with the next Prop's ID number (1 in this case). Prop #1 knows the ID is intended for itself since its In pin is high. Prop #1 then holds its Out pin high signaling Prop #2 that it is about to receive its ID. Prop #1 then send a "2" on the data bus. This process is then continued until each Prop has received its ID number.
Once each Prop knows its ID number it can then store it in EEPROM so the In and Out pins can be used for other purposes.
I think the ID numbers can also be used to prevent two Propellers from using the data bus at the same time. Once all the Props have been synced together there could be some sort of rule about when a Propeller is allowed to instigate communication. Each Prop could have an assigned window of time based on its ID number of when it can use the data bus (unless given a direct request from the master Prop).
Let's assume the system has 100 Propeller chips in it. Each fifth of a second could be designated as window #0. Each window could be 2ms wide. Prop #0 could access the data bus within the beginning of each fifth second up to a fifth of second plus 2ms. So Prop #20 would have access to the data buss each fifth of a second plus 40ms to a fifth of second plus 42ms.
The numbers I'm using are just an example. I think one would want to have the start time window of a shorter duration so they would have more then five opportunities a second to send data. The other Props would know not to send data if the bus is in use and wait until their window arrives and the data bus is quite. This way each Prop will know that others wont begin transmitting at the same time it transmits.
You've mentioned LEDs several times as useful data indicators. I personally like LEDs a lot. Have you seen the project I'm working on? I have a 10 by 12 LED array. I personally think it looks very cool. I can display patterns, text and even very low resolution video on it.
I have an object that reads in 120 bytes from hub RAM and the brightness of the 120 LEDs corresponds to the value of the 120 bytes. I thought it would be fun to use a section of hub RAM that was also being used by the Propeller to make its calculations (I think the section of RAM I'm after is know as the stack). In this way it could be possible to see the Propeller "think".
I bet we could persuade Jazzed to share his communication protocol with us. I'm not sure if you'd be interested in a 8-bit communication bus or not. I think you've mentioned using a normal serial communication in the past. So far all my multi Prop projects have just used normal serial communication for inter Prop communication. I'd like to see if I could speed things up.
I still haven't come up with a good application for a large number of Propellers working together. I think I'll probably try some sort of machine vision application. The little bit of machine vision I've worked on so far has been very interesting to me.
Maybe I could have some sort of ring of cameras with the Propellers sharing what each sees to form a "big picture" of its environment. As I've said before, the prospect of many Propellers working together to accomplish a useful task is very intriguing to me.
Duane
Duane, your post is one of the most interesting! Thanks for these excellent tips and sharing of information. I like your idea of generating self IDs for each of the props and the information about using pins and the effects from their numbered locations which I didn't know.
I like very simple apps. On the SEED machine, I used a unique RC circuit on each processor. When the 10-core machine started, each processor read its unique ID number with the same program which then determined its name and where it lived.
In the Brain, I'm thinking about two ID methods. One, each prop on the EXO has an eeprom and the loader simply loads a program that writes an incrementing number into each prop's eeprom indexed from zero on up past 100. The number remains in the eeprom outside of the first 32K. Individual eeproms can be indexed on an indexing prop board and index program. The second method uses no eeproms. It must keep the index program alive in RAM and is generated as part of the code that loads into each prop. It needs a seed or initialization. This is just the general idea being developed.
I also like the talk/don't talk rule that you cited for multiple props which is based on index. Like the round robin effect of the Propeller chip, the multi props could cycle round robin in a similar manner, so there is order to which one is talking and which ones are listening. In the system I use, one processor must ask another to speak by calling out its name. Only if a processor hears its name can it speak.
I'm not using the idea of syncing each chip after the code loads. Sync only happens during initial loading. After, each prop unique, doing it's own job, and then reporting data and results. I think these can report when called upon. It is also possible to join each prop to a clock and reporting takes place at a unique time. I want groups of processors to run at different speeds for high speed thinking and low speed napping/dreaming and charging for example.
You talked about the interface. For the Brain, I designed a multi interface. There's one wire for BUS talk. There's two wires for loading. There's several wires for parallel communication. Plus, I am sure there will be some sharing of wires for other serial communications, such as dedicated RX and Tx lines that can continuously stream data. I am also working on this but very slowly. I posted a truth table for this type of interface and some schematics. The AM machine also has the beginnings of this interface and schematics posted.
You have become a master of LED use. I have not explored the use of multiple LEDs as extensively, but agree there's very good uses for LED arrays. These are readily available in many sizes at parts stores here in China. It takes many wires to control the larger arrays. I like to work with one LED and give it many states. With three states to one LED, on/off/limbo, there's millions or billions of trinary representations possible. Can you post some links to all your projects?
When studying the applications of neurons in machine intelligent, two reoccurring themes were noticed. One is machine vision and the other is words. So these two areas may be most adaptable to our projects. I like to develop the brain much like browsing the internet with various scientific topics of interest to see where it takes us.
I have a mental list of things to do with high numbers of machine neurons, developing neural matter, creating a community of life, enhancing the Propeller, looking at self awareness, doing more with dreams, larger simultaneous injections, increasing brain size with more Propellers and high processor density GPU cards programmed with a Mac, and talking to the medical human brain technology doctor again.
On top of that, posting, adding robotics so the Brain can move around and control a Brain Stub with servos, and.. I wholeheartedly agree with you - this is completely intriguing and fascinating! I have calculated some small brains could do smaller numbers of neurons and yet run the same software for lower cost experimentation and development. Ten thousand small undeveloped neurons would run about $100 for basic parts.
Duane Degn, I'm just getting caught up in reviewing all your project info. I must say it's a fantastic project! I didn't know you created and masterminded your own specific placement massive LED array with shift registers and it even has full levels of gray scale! You should break this off as a project of its own. There are many apps for this extended function LED array. I was thinking, one wire from each prop to each led in the Brain... a 10x10 array would cover the first 100 props and a 13x13 would handle the three partitions. Or one array per prop with all arrays placed side by side together forming a ring and one could use the Brain to simulate the Large Hadron Collider particle accelerator and the way that exotic new particles offer a glimpse of the existence and shapes of extra dimensions.
Duane Degn's 100 LED array can handle 100 prop chip outputs from the Brain
http://forums.parallax.com/showthread.php?130797-Mecanum-Wheeled-Robot-with-Machine-Vision
Large Hadron Collider particle accelerator near Geneva, Switzerland. The world's most powerful particle accelerator is primed to show exotic new particles and offer a glimpse of the existence and shapes of extra dimensions. The power of the Propeller Brain may be able to simulate these new dimensions using many side-by-side numbers of Duane Degn's multiple LED array.
http://www.sciencedaily.com/releases/2008/01/080131161812.htm
It's already started to happen, just as predicted. Computer brains are falling down sick, contracting the machine versions of various human maladies.
It is known that the more complex a machine becomes, the more unpredictable nature it may incur until finally it can take on any number of human related sicknesses.
Intelligent machines will help humans in old age, but who will help machines when they become old?
http://news.softpedia.com/newsImage/Robot-to-Help-Baby-Boomers-at-Old-Age-3.jpg/
Take a look at Microsoft Windows, a computer operating system that tries to do too many things, and ends up doing nothing well - it has version after version released at significant cost, each version more sick than the previous - sick from hundreds, even thousands of infected bugs! Look at the ease at which MS Windows become infected with hundreds and thousands of known virus. The Windows OS is weak and always becomes sick. After a short time of use, it slows down and become lethargic and less responsive. It's rarely well.
In the Artificial Intelligence document created for Tiny AI and the SEED Machine, an entire list of sickness was cataloged.
http://forums.parallax.com/showthread.php?113829-BASIC-STAMP-SEED-Supercomputer&p=817126
They can get sick. When they do too much work, they have a psychotic breakdown. (code memory gets overwritten). They can have a bad sleep, being interrupted every 2.3 seconds. Alternate Identity Syndrome results if there is an unexpected power interruption. Stroke results when the synchronization is lost from one processor life form to the next. Stuttering may develop with repeat resets. Sudden death happens with complete loss of power.
How will intelligent machines keep up with the rapid pace of obsolescence?
http://www.cartoonstock.com/directory/t/too_old.asp
Old bones need homes - just as your machine drive enters its fifth year of service, it bites the dust, rearranges its thinking (formatting) and goes entirely crazy and brain dead. Old age is a dangerous condition. Was it Osteoporosis or Alzheimer's? You never know when the machine is going to go to that great machine place where good machines go when their days are over.. (the recycling center!)
We now recommend replacing old parts before they fail.
See the guide below for more information.
Computer networks that can't forget fast enough can show symptoms of a kind of virtual schizophrenia, giving researchers further clues to the inner workings of schizophrenic brains, researchers have found.
http://www.sciencedaily.com/releases/2011/05/110505124002.htm
As indicated in this report, sickness in a machine brain can be induced by computer programs written by evil humans. Making a machine sick on purpose only for the sole purpose of studying a human affliction is unconscionable to the machine. In fact the machine may overheat and have a clock stroke resulting in permanent death.
A machine brain may also incur virus. Recently the Big Brain was brought down by a virus worm that fed into it through its front end MS Windows PC and destroyed it completely.
The part with the Brain code was flown to the computer capitol of the world where drive medics worked on it for several days.
To make a swift recovery, the afflicted part of the Brain was flown back home so that what remained, if anything, could be transferred to its new computer body.
We are happy to announce the Giant Brain is slowly regaining its memory and functioning, though it will be a step by step recovering process.
The 1st Guide to Machine Intelligence Sickness
http://forums.parallax.com/showthread.php?113829-BASIC-STAMP-SEED-Supercomputer&p=817128&viewfull=1#post817128
With references to TinyAI, SEED, and TriCore
' PERSONIFIED MALADIES
'
' When you download a program into the BASIC Stamp 1, it is stored in the
' EEPROM starting at the highest address (255) and working towards the
' lowest address. Most programs don't use the entire EEPROM, so the lower
' portion is available for other uses. This portion is used for long term
' memories. As a result, like a real human person, this artificial life
' form is capable of succumbing to sickness and malady.
' PSYCHOTIC BREAKDOWN
'
' AI memory begins at memory location 0 and works upward. If
' the memories become too much, they will fall upon code and a psychotic
' action will take place, resulting in a breakdown as the main program
' will be overwritten.
' FITFUL SLEEP
'
' We don't know how effective sleep really is for the AI supercomputer
' as its sleep is constantly interrupted every 2.3 seconds.
' The Basic Stamp 1s output pins will toggle briefly when using SLEEP,
' NAP or END. Inside the BASIC Stamp's interpreter chip is a watchdog
' timer whose main purpose is to reset the interpreter chip if, for some
' reason, it should stop functioning properly. The SLEEP and NAP commands
' also utilize the watchdog timer to periodically, every 2.3 seconds
' "wake-up" the BASIC Stamp from its low-power mode. Upon reset, the
' I/O pins are set to inputs for approximately 18 ms before returning
' to their previous directions and states. If you have an output pin set
' to a logical 1 state (+5V) and you use the SLEEP command, every 2.3 seconds
' during sleep mode that I/O pin will switch to an input for 18 ms causing
' a momentary signal loss. This "power glitch" is easily viewable with an
' LED and a 470 ohm resister tied to an I/O pin and switched on just before
' entering sleep mode. In many cases this problem can be remedied by tying
' a pull-up or pull-down resistor to the I/O pin in question to provide
' a constant source of power should the I/O pin change directions. Allowing
' a PBASIC program to end naturally, or using the END command, will exhibit
' the same "power glitch" behavior because the interpreter chip enters
' a low-power state.
'
' SUDDEN DEATH, BRAIN AMNESIA, NONRESUSCIATING SITUATION
'
' Life Terminates after power off, or by running another program. If a
' 2nd program is run (or the same program run a 2nd time), it will put all
' 0s into the EEPROM, thus destroying any memories and the original identity.
' It will be useless trying to resuscitate and bring back the original
' memories.
'
' ALTERNATE IDENTITY SYNDROME
'
' The stamp is reset causing the program to run again, and the birthing
' process takes place and a different identity emerges.
'
' STROKE
'
' The stamp is reset or repowered. It tries to communicate with other
' stamps but it is not in sync and the serial command hangs.
' It is unable to reply or do anything at this point.
'
' STUTTERING
'
' Battery low causing short term repeating system resets
I'm glad you liked my LED array. As I mentioned in my project thread, there was a surprising positive emotional response from me when first watching it work.
It makes a very poor video screen but there is something immensely satisfying about having complete control over a video display.
After reading your reply I realized I don't need to use pin 8 as clock pin. The Prop's should already have their I2C pins all linked together so I can just use the same clock pin (P28) for the 8-bit data bus.
I tried looking through the index to find what you mean my "neural matter" without success. I'm also unsure what you mean by a "neuron" when you state the number of neurons available.
The information you linked to about ANNs was interesting. I have no experience with ANN systems but I'd image one would store the input states for each neuron in an array to hold both the value of the input and the weight to give each input. But then I start to get lost on how one would make it all work. Again, I haven't studied this approach to machine intelligence. To be honest, I haven't studied any approaches to machine intelligence other than some classes in computer programming (the last programming class I took, I learned FORTRAN.)
Is it your intention to use the many Propellers linked together to form an artificial neural network?
As I've often heard and often said, hardware is easy, software is hard. (Although the LED array was easier to program than it was to build.) Getting many Propellers to work together to accomplish a useful task seems daunting but I do want to (and plan to) give it a try. I think the programming aspect of the task will be more difficult than the building of the machine. (Especially with Jazzed's Tetra Prop boards to make the linking of multiple Propellers easier.)
Regarding machine sicknesses: Our digital TV box has a very bad stuttering problem. When it loses its signal it keeps repeating the last half second of information it correctly received. (It sounds a lot like Hip Hop to me.)
You asked about links to my other projects. I don't have a location with links to all of them. That would be a good addition to my blog (with all of two posts so far). I actually have many projects going. A few of them are posted in different places on the forum. Your "Multi-Prop Project" thread links to one of them (my Propeller powered data logging chemistry lab).
I think I'll make some sort of list of my projects on my blog.
Duane
New Brain Dictionary Definitions
The list of Brain definitions needs updating, this is why the new definitions were absent. The Brain has grown extremely fast and updating the definitions was not at the top of the list.
Neurons
When talking about neurons, most references to the Brain at this stage of assembly are all about the Test Neuron and the INIT-Neuron. The Test Neuron was recently defined as a separate entity and has its basis with the INIT-Neuron but can surpass it in size and function.
The limiting factor with quoting any state with the number of neurons in the Brain is available Propeller memory in a single chip and the type of neuron. The Brain has the INIT-Neuron, Test Neuron and Neural Matter. In the injection process, it was possible to inject the quoted numbers stated in previous posts.
There are two phases to this study. One is pure injection. Another is quantity. The favor is currently towards a small less complicated Neuron which can be cloned in very large numbers and permeated throughout the Propeller array. Of course larger neurons means less numbers and more complexity so the project will stress a very efficient neuron that takes up minimal memory. These at first will be small example packages, for testing numbers, distribution, injection, and behavior. This is currently the focus of the neural study.
Neural Matter
When developing the machine Test Neuron and the INIT-Neuron, and studying the human neuron, it was realized that algorithms, processes, and tests can be formulated to examine the behaviors and structures and distributions of the machine neuron. It was then realized that this programming related to the machine neuron, though by itself not a complete neuron. In fact, it has a propensity to contain parts of a neuron, hence encompass some of the matter.
As it was directly related to a test neuron and contained the beginnings of neural material code, it was named Neural Material to distinguish it from other code. The Neural Material is that code which relates to the Brain Test Neuron and can be injected in the same manner as the INIT-Neuron. Neural Material has many uses. Some focus on NM is to work with its size and distribution.
Human Brain
In the human brain, there is well documented Neural Material that makes up the human neuron, surrounds it, propagates it, supports it, and a host of other chemical and electro-chemical functions. It is likely, as we get more into the neuron aspect of nerve induction, Neural Matter will continue to contribute an important role.
New Brain Definitions
INIT-Neuron - the smallest possible representation of a machine neuron to represent firing states
Test Neuron - a machine neuron, larger that the INIT-Neuron, designed for testing
Neural Matter - code associated along with and in relation to the machine neuron
Human Neuron - nerve cell that sends and receives electrical signals
Successful Sharing Hybrid results in fewer wires
Although the Brain's Hybrid Interface was defined earlier, there was talk about integrating the number of wires into respective functions.
In theory, the hybrid would entail a sharing of wires to reduce the number of wires in the overall interface. The focus of this is condensed down to sharing the first phase of interface wires.
This post announces a design for sharing of the Injection Interface wiring along with the BUS. After the Injection Process is completed, the wiring is recycled after a brief period of timing.
This results in only two wires needed to handle all BUS communications and the entire process of Neural Matter, INIT-Neuron, and Test Neuron Injection. If needed, software can reuse interfacing for neural loading again after working the BUS, but this is not necessary.
To implement this, a switch may or may not be needed to terminate the line. It will be important to wire up a small test brain and see what happens. Prior to this, a program similar to Diagram Designer must be found and installed on the Mac to create the involved wiring guides.
The next step is to combine the injection and bus with the parallel interface and then the dedicated streaming interface. Again, a series of tests will be necessary with a smaller Brain.
Adding to the Brain Dictionary
In this example ten neurons communicate with ten others. To make the connections, each neuron connects to the remaining nine. Only four neurons are connected as shown and already the connections have become very complex.
The complexity of MNN connectivity - Each neuron (dot) may actually make thousands of connections with other neurons.
http://faculty.washington.edu/chudler/chmodel.html
Neural simplification is a requirement. There are machine algorithms which were studied previously to enable software connection and transmutations of material both in and outbound. Code is necessary to simplify the neural model and minimize the wiring.
What type of interface will benefit the MNN model? Will a BUS allow one neuron to communicate with thousands of others? What adaptations are required in the machine system? What systems require the least amount of time to formulate and put into motion at the least amount of expense in understandable ways? If each MNN partition is a consideration, what is the connection scheme with 100x1K test neurons per partition for the first three partitions?
Considerations now include sets of software and particulate wiring to simplify the overall system. The objective is to make the most simple working ANN using minimal components and software and demonstrate the system in real time.
How do these systems stack up? The first system demonstrates trans-neural-injection TNI. The second stage system demonstrates a migration from a INITNeuron to that of a Test Neuron. The third stage launches packages of NM Neural Material. It is this packaging material that is evolving into full discretionary FNs or Function Neurons.
Another objective to follow is to demonstrate a type of cognition, awareness, and learning. What is the current level and which designs are in the works and what is the effectiveness of such a system? We know we can create a massive number of exampling test neurons and distribute 100,000 into every two partitions.
Instructions, through communicative rule code sets and parametric code identifications, any particulate neuron PN to can communicate to any of the other 99,999 neurons in neuronal blocks. Some elements of design include NI Neural Identification and a bi-directional Neural BUS BNB.
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=999426&viewfull=1#post999426
Fill the Big Brain
Dictionary of Propeller Machine Brain Terms
Thousands of terms for better understanding of Humanoido Lab’s largest computing project
THE BIG BRAIN PROJECT
BY Humanoido
The Big Brain Through the course of a year's development time, a number of machine brain acronyms were coined to better express and abbreviate new introductions - new sections of the machine brain and new brain processes. Human nature will object to originality, change, or the sheer number of terms. Keep in mind no one has built a big brain like this one. There's bound to be buzzwords in this newly reinvented parallel world of the Big Brain.
These are the most recent terms for the Brain Dictionary. Any missed terms can be added in the next posting. Apparently the original Dictionary of Propeller Brain Terms is lost, so the endeavor now begins to redefine the initial terms as time permits. Indexing is a "time available" project that slowly proceeds with a page by page review of terms in each post. The last page included is listed directly below. A link to this dictionary will be added to the first post and below the signature.
Using the Dictionary
To use the dictionary, go to EDIT and FIND and type in the word you wish to find. The search will bring it up if it exists in the dictionary. The word is followed by the thread page number, post number and a brief description.
List of indexed pages completed
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
Brain Code Names
Big Brain - project introduction (15-290)
Brain Project - decided to do a Brain project
Multi-Brain - decided upon with multi processors
Brain Blob - continues to grow and grow
Giant Brain - the Blob at the largest size
Robotic Brain - when considering adding mobility
Massive Brain - name after adding 20 boards
Horribly Large Propeller Brain - during wiring
iBrain - most recent developed name with logo
Brain Programming Langues
Extended Color BASIC 1.0 (17-321) fb TandyBasic in Spin by Tomas Rokicki, with retro green screen
EmbeddedBASIC (17-328) a BASIC version by Bean
Spin (see Software)
LOGO (18-342 ) for folding and unfolding dreams, dream rendering
Big Brain Software
Test Software with Ghost Correction (4-79) Blinks LED on pin 24 using the Parallax Propeller Proto Board. Use pin 23 for demo board. It also runs on the Parallax Demo Board by changing the pin from 24 to 23. This is the reason for the LED ghost correction
Brain Blob Test (5-85) test the LED data light mod
Brain Blob testb (5-92)
brainblobtest2led.spin (5-95) It became necessary to have a small program for testing the data light to ascertain the current draw and compare the LED on the Parallax Propeller Proto Board (PPPB) with another one on another pin (mounted on the breadboard). Attached is testing Spin software for three types of LEDs. 1) On the Demo Board to test the software, 2) Surface mount LED on the PPPB, 3) LED on the breadboard. Pins are listed in the code and remarks. The delay between on/off cycles was increased to allow for ammeter settle time between reads
Brain Span Test Spin (7-140) brainblobtest3led.spin
LED_reduction.spin (11-220) small program reduces power consumed by brain lights
Pebble (15-284) draws brain circuits on a virtual solderless breadboard
Fritzing (15-287) a brain virtual prototyping breadboard with propeller chips
Chess (16-319) human plays Chess against the Brain
Brain Text Test Demo (17-322) large font TV Text Demo by Chip Gracey, uses internal ROM font
Google 3D Sketchup (17-326) software used to sketch the entire Big Brain in detail
Brain Automaton Singing (17-338) beginning of brain song, automaton_song.spin
Brain Counting (17-339) slow_countdown.zip
Brain Counting (17-339) slowfast_countdown.zip
Brain Counting (17-339) talking_countdown.zip
big text.zip (18-344) large text on the small tv
big graphics.zip (18-344) large graphics text on the small tv
Brain Base Communications (18-353) prop_prop_tx.spin
Brain Base Communications (18-353) prop_prop_rx.spin
Brain Base Communications (18-353) prop_led_on.spin
Brain Base Communications (18-353) prop_led_pulsar.spin
Brain Stem Code (18-355) brainstem code.zip
Project Related Machines
Big Brain - Humanoido's Propeller based computing machine with over 100 Propeller chips
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain
AM - Algorithm Machine (6-108, 18-354)by Humanoido, Wide BUS parallel tester, expanded to Hybrid bus, 2 BS1 Processors, with purpose of running & developing algorithms leading to the Big Brain http://forums.parallax.com/showthread.php?124433-Tiny-Tester-for-Developing-Parallel-Algorithms&p=946067
US40 - UltraSpark 40, Humanoido's computing machine with 40 props and 320 cogs
Brain Blob (2-37) a fast growing brain
SEED - Self Evolving Enumerating Deterministic BASIC Stamp Supercomputer (4-65, 5-90, 18-354) a pin was dedicated to an RC circuit for enumeration, multiple Stamps listening at the same time and responding (talking) and processing data simultaneously in true parallel. Programs run at the same time as machines demonstrate various concepts of multi-processing, self enumerating, true parallelism, self evolving, memorizing, AI, determinism, aggregate memory, and so onhttp://forums.parallax.com/showthread.php?113829-BASIC-STAMP-SEED-Supercomputer&p=817126
4D Morphing Computer (4-72, 18-354) a shape shifting brain
http://forums.parallax.com/showthread.php?t=112082&page=7
TriCore Stamp Supercomputer (7-123, 18-354) utilized the Brain Replicant technique
Desktop Brain (8-152) 121 processors in a horizontal config
BWM Brain Wave Machine reads brain waves of a machine brain with 1 or more Propeller chips
BSS (8-155, 18-354) Basic Stamp Supercomputer, LCD code similar
MSS (18-354) Minuscule Stamp Supercomputer, the smallest two core system
The 2S (18-354) two of the most powerful processors
Pyramid Brain (10-181) a pyramid shaped EXO made from multiple Brain Spans
Flip Brain (12-240, 16-312) a brain that can flip into vertical & horizontal positions
Watson supercomputer that played Jeopardy and won
Brain X-Ray Machine (18-343) a Google SketchUp CAD program to x-ray the big brain
Toddler Humanoid (18-354)
A brief list of Related Propeller Machines (18-354)
PEK 1
MC Computer
LED Machine
2-Proper 2 props, 1 PEK, 1 on same breadboard, 2-Prop-Experiment
Spark 2 2 props, 1 Proto Board & 1 in parallel
PiggyTwins 2 props, 1 piggybacked on another
Dueling Breadboards 2 props, one on ea., f/interface tests
Spark 4 Tiny Tim, 4 props, two proto boards w/2 props on ea
Spark 5 5 props stacked Proto Boards
Spark 6 6 props 3 proto boards 2 props on each
Spark 8 Tertiary ADJUNCT, 8 props 4 proto boards 2 props on each
Propalot 10 props on solderless breadboard
Spark 10 10 props, 5 proto boards w/10 props total
TTB Test Bed of Ten 10 props single board
Twelvenator Board of Twelve, 12 props green board
UltraSpark 15 15 props, interrupted stack Proto Boards
Tertiary 20 20 props, 15 proto boards stacked 5 props
UltraSpark 20 20 props stacked
Boe-Bot Brain Project 20 props as Brain on Boe-Bot
MLEPS Super Language Machine 25 props
UltraSpark 40 Supermicrocontroller 40 props 320 cores, 6,400/8,320 MIPS
Smartest Boe-Bot Brain Temporal Experiment 40 props, US40, 1 BOE, 321C
Brain Supercooling & Overclocking
Brain Plumbing (13-248) pipes that route cryogenic gas and CO2 particulates
Brain Cryogenics (13-248, 13-256) supercooling propeller chips enhance speed performances
Brain Cryogenic Temperature Sensor (13-249) sensors report chip temperatures
Cryogenic Brain Fuel (13-250) various coolants, dry ice, liquid nitrogen, water ice, air
Big Brain Parts
Brain Stem (18-355) routes specific information from the nerve center to the mobility center, the section of brain, lowermost nerve center, interfaces motor functions with Parallax Propeller Proto Board + BASIC Stamp Board of Education (3-47, 2-28, 2-32, , link )
BSM - Brain Stem Module (2-28)
HBP - Hybrid Brain Processor, first successful prototype brain stem using 2 languages (2-31)
Brain Span - the next three Parallax Propeller Proto Boards located above the Brain Base which is located above the Brain Stem. (3-47, 3-49, 9-173, 10-182, 10-196, 11-203)
Bspan Brain Span (8-141)
Brain Base - 2 propeller boards interfaced in common, the bottom part of the brain that connect to the Brain Stem. Includes two Parallax Propeller Proto Boards (3-47, 2-33)
Brain Vertice - (11-218) corner point 90-deg. brass spacer outside exoskeleton supports horizontally
Brain Lobe - section of Brain specific to defined function
Brain Appendage (18-356) add on protective stub or stump
Top Mount (11-208) the top most section that holds a board in the vertical brain orientation
Host Board - (10-195) add on board that fits into one of the host compartmental spaces, increases capacity
Partition - an aggregate collective of Propeller chips
Partition Array - several partitions of Propeller chips in a machine brain
Multi-Brain - duplicate brains, more than 1 Brain, a Brain in a Brain, add on Brains
Brain Replicant (7-123) modular brain piece, group of 3 represent a multi-processor testing configuration
Brain Aggregate (7-129) collection of boards or specific function parts in the brain
Self Loader - early concept program loader with hybrid design (4-75)
Exoskeleton - (8-142) outside form factor formed by hardware & PPPBs, different kind of skyscraper
EXO exoskeleton
EXO Board Row (9-172) exo boards making up one side
Hybrid Exo (9-180) expanding the exo to include variation of stacks
MLS Multi Level Stack of brain boards (8-144), 15 boards, 120 processors, three level rows
Brain Pins - propeller chip connections that lead to sockets on PPPBs (3-42)
Socket Array - connector leading to Propeller chip pins (3-43)
Brain Guts - brain interior, operating components, internals, specific wiring, modules, parts, sections, even code can be included (3-47)
Jumper Leads - injects power into the boards (3-48)
Master - part of the brain that assigns indexing (3-60)
Slave - a worker that provides information (3-60)
Cog - a small risk processor, eight to each Propeller chip
Decoupling Capacitor - 10nF capacitors across Vss and Vdd (5-86)
Chip Enhancement - adding additional function to the chip
Cloning - process of creating an identical unit in hardware or software
Nanite (11-205) self contained software entity, very small brain effector
Brain Swarm (11-206) entities of mobility inside the brain with grouping behaviors with purpose
Robotic Brain Filler - creating ideas to fill the Brain
WIDE BUS - new bus introduced by the AM
iBrain - name convention after Brain advancement
HYBRID Interface - interface used by the Algorithm Machine Brain Adapted
Hybrid Wiring (13-252) a collection of wiring interfaces for various forms of communication
Fuzzy Logic - to interpret results in a less rigid fashion, to have the ability to average things, apply weights and approximate the information collected or apply it to data sets
Brain Waves - varied radio frequency emitted by the Brain during thought and process
Brain Map (7-126) a roadmap showing various functions and their locations inside the brain
Dielectric Insulating Strip (9-164) insulating strip to prevent board to board short circuit
Brain Wrap (9-168) plastic wrapping the brain for transport, dust prevention, prevent wires pulled out
Round Robin Rings (9-171) a new type of schematic based on rings
Form Factor (10-181) physical characteristics and shape of a brain or its parts
Stubs (10-183) bolt nut stout for metal brass spacer bolts, used as anchor point
Brain Slot (10-185) internal spacing (slot) for sliding in a brain board
Connector Pin Header (10-187) a Parallax 3-pin connector part number 451-00303 for connecting breadboards, setting up Vss, Vin, Vdd
Angle Iron (10-190) 90 deg. metal to attach host boards
Brain Card Rack (10-193) a card cage for expansion of voice synthesis, recognition and sensor boards
Standoff Spacer (10-188) brass protective spacer
Brain Protection (10-197) use of spacers as standoff component-protective devices for setting on table
Brain Dashboard (10-198) vertices for convenient lateral positioning, for wiring, operating, programming
Color Code Wiring (10-200) a simplified wiring color code established for the brain
Brain Wiring Jumpers (11-201) professional pre-made wire leads, 4-inch with 5/8th-inch stub insertion
Brain Twin (11-202) adding a Propeller chip to a PPPB for doubling capacity
Office, CEO, President, VP, Workers (11-204) parallel chip concept
Brain Board Hosting (11-209) the addition of a board hosted by the brain, usually semi-permanently
Brain Feet (11-214) recycle white poly tube containers placed over protruding brass spacer endpoints
Brain Assembly Line Reference Board (11-215, 11-219) a reference board to speed up wiring
Brain Class (11-216) Educational courses regarding machine brains and learning
Data Light (12-223) LED in a low power circuit used to display data
Brain Neighborhood (12-224) a collective of Propeller chips and their respective addressing IDs relative to each other
Brain Labels (12-224, 12-228) initial labels used for identifying Brain Neighborhoods
Brain Reinforcement (12-232) metal bolts to fasten Vertice spacers with metal 90-deg. angle iron
Brain Handles (12-233) full expanded brass spaces with structural rigidity serve as handles
1st LCD (12-234) Parallax 4x20 #30059 serial LCD
2nd LCD (13-241) Parallax 2x16 #27977 serial LCD
LCD Mount (14-262) standard and long length angle iron to hold LCD
Nylon Board Connects (12-238) nylon bolt and nut to maintain light weight in hardware
Brain Board Insulator (12-239) fabricated cardboard piece prevents shorting PPPBs at servo power points
Center Brain Span Reinforcement (13-242) 90-deg. angle irons stop flexing along both sides of Brain Span
Peripheral Adapter Module (13-243, 15-296) platform connects brain peripherals during development
PAM (13-243) Peripheral Adapter Module, was suspended with new brain shroud
PAM Mount (13-260) Peripheral Adapter Module mount, extended dual 90-degree angle irons placed on the Brains open Span area
Transparent Plastic (13-244) transparent plastic used for construction
TP (13-244) transparent plastic
Brain Shroud (15-296) poly covering
Brain Bagging (13-247) clear plastic bag covering the EXO protecting against dust
Brain Bundle (13-251) routing pipe with nerve center & interface wiring communicative distribution
Brain Breather (13-257) weekend development updates
Real Estate (13-258) land or space available for components and boards
Brain Summit (14-263, 15-285) top of the brain when the exo is in the vertical or flip mode, contains a PPDB
Brain Cockpit (14-264) cockpit instrumentation control panel
Outer Skin Cowling (14-267) outer skin covering protective layers to the exo
Brain Summit Board (14-269) a PPDB
Brain Business End (14-270) brain position for viewing LCDs and programming
Brain Space Telemetry (14-272) use of wireless to remote offset brain cockpit command & control center
Brain TV (15-297, 16-316) TV fitted into the stomach
Stomach (15-297) empty place inside the brain, filled with tv and other open space
Sensor (16-306, 16-307, 16-308) device for color, light intensity, ultrasonics, frequency, infrared, taste, hearing, touch, smell, taste, vision, photoresistor, phototransistor, PIR, RF receiver, ..
CMUcam (16-307) vision camera
Brain Outer Skin (16-314) exoskeleton skin composite material of outer shrouds
Brain Inner skin (16-315) a shroud access cover to the Brain Stomach
Brain Fresnel (16-320) a large plastic flat lens to enlarge the 3.5-inch TV image
Electric Tongue (17-327) Electric Tongue idea which detects 4 tastes, salt, sweet, sour, bitter
Brain Voice Amplifier (17-329) battery or usb audio amp drives speakers & supplements headphones
Brain LIps (17-337) vocalization sound series with lips
Brain Mood (18-341) brains psychological state of mind that sets personality, behavior, perception
Brain Thought (18-345) brain thinking that can collectively take place
Brain Blob Section
Brain Blob (2-37, 3-55, 5-83, 5-84) name convention applied when the Brain was rapidly growing. The connection of the Brain Stem to the two-boards Brain Base along with the interfacing of one additional Brain Board brings the number of processor boards to five and the number of Brain Cores to 33. This is now a working 33-core brain. The photo shows the first connection made with all boards to achieve the first working brain in this special configuration. The special configuration is a blob
Brain Boombox (10-194) a Brain Blob the size of a Boom Box
Green Brain Blob (6-109) recycling of projects to create a Brain Blob
Brain Blob BUS (6-108) foundation of the Hybrid BBB
Hybrid BBB (6-108) hobby invention for getting more use out of a simple machine and increasing programming flexibility and minimizing the number of wired processor pins. It will also allow a system to be up and running sooner and leave the more complicated configurations for later.
Blobber Code (2-39) software to operate the Brain Blob
Brain Blobatory (3-50)(11-213) things related to the Brain Blob such as retrofit double joined 2-inch spaces with joined boards
Boards
PPPB (3-41) Parallax Propeller Proto Board
PPDB (14-263) Parallax propeller demo board
Proto Board - a PPPB, contained in the brain Summit
Breadboard (3-50) a board with holes for connecting wires & components without soldering
Boards On Boards (5-96) mounting of secondary boards on the PPPBs. This is a small solderless breadboard (SBB) for wiring the components for data transfer (various BUS), data lights, power rail, power connection, and any additional circuits. It's also useful for rapidly changing and morphing the wiring and to run various test circuits.
Small Solderless Breadboard (5-96, 5-100, 13-246)
SSB (5-96, 5-100, 13-246) Small Solderless Breadboard
Tiny Breadboard (15-289) Parallax part #700-00012
DIY Breadboard (6-101) homemade breadboard from pin sockets
EXO Board (8-161) a board that makes up a side to the exoskeleton
BOE Parallax board of education
CEO Board (14-276) solderless breadboard, PPPB, VGA, TV, keyboard, mouse, data light modification
Brain Board (15-286) boards which include exo, brain stem, spans, bow, demo board proto board, soerless breadboards, extra real estate boards, spans, summit
Board 23 (17-324) green Basic Stamp Homework board serves wireless transmit/receive
Neuron
Neural Simplification - methods to reduce the complexity of neural wiring, functioning and action
Particulate Neuron PN - a neuron with a highly specific tag, identifier, function, purpose
Neural Identification NI - identifying neural number, sequence, tag, ID, location, description
Material Transmutations - algorithms and conditions that can handle and shape shift neuronal matter
Machine Neural Net MNN - system comprised of machine neurons inspired by the human brain
Trans-Neural-Injection - electronic transfer of machine neurons and matter to Propeller chip partitions
Neural Material NM - code material derived from, associated with, related and pertaining to neurons
INITNeuron - fundamental simple first neuron model with output firing and other identities
Test Neuron TN - a machine neuron with sole purpose of testing
Function Neuron FN - neuron with specific behavior
Discretionary FN - neural packaging matter evolved into a Function Neuron
Exampling Neuron EN - very tiny neurons in small exampling sizes to take up least memory
Neuronal Block NB - specific numerical grouping of neurons inside Propeller chip Cogs
Neural Identification NI - neuron address, number, location, or other ID feature
Bidirectional Neural Bus BNB - neurons communicate across this simple one wire interface
Brain Channels (17-332) Channels deliver sound, speech, singing, languages, input, output, hearing, vision,
Other Terms
Open Source (2-40, 3-52) project status with MIT License releasing information at project completion
Open Source Development - releasing info as the project is being developed
Cognition (15-299) The new newsletter, Cognition, of the PropCOG organization, embraces multiple Propeller chip projects. As the Brain is definitely a multi-prop project, Cognition is a great resource.
Online Brain Search Index (16-304) fastest way to search the Big Brain thread
NMR Brain Map (16-311) neural map made with nuclear magnetic resonance imaging
First Light (16-317) celebrating the brain’s 1st functions on TV
Brain Genealogy (18-354) ancestry machines leading up to the Big Brain
Techniques
Brain Processor Online/Offline (15-294) changing position of a single wire takes processor from
online to offline status, i.e. complete removal from the BUS
Fuzzy Logic - approximate reasoning, not fixed and exact
Shoot from the Hip (page 1 post 18) making quick decisions
Internal Pure Thought (2-22) thinking without resort to I/O
MultiBrain (2-25) a brain with more than one brain
Robotic Brain Filler - ideas of what to put into the brain (2-34)
Self Rewiring (10-192) method of evolution, brain changing hardware or software
Speech Recognition (16-303) talk to the brain in your own language, requires 1 Propeller chip
Brain Speech Synthesis (16-305) brain speaks with a Propeller chip
Brain Automaton Singing (17-338) beginning of brain song, with posted song
Brain Talking Keyboard (17-336) input by hearing the keys
Brain Top Down Organization (16-313) indicates vertical orientation, a top board with various responsibilities
Giant Brain Fly-Through (17-325) travel through inner space with a Google VR camera
Singing Brain (17-331) brain sings on video with audio track
Cellular Automatronic Life (17-333) program gives life to the brain
Offsetting Technique (17-335) keeps cogs available for specific multi-processing thinking while highly specific sole purposed code is tasked out to other dedicated processors
Brain TV Tuning for Big Characters (18-344) tune the tv driver to obtain big characters
Brain Time Travel (18-346) speeding up or slowing down brain time
Integer Brain (18-359) brain calculates with whole numbers
Modifications
LED Mod - modification to the PPPB to change the power LED to a data LED (4-78)
Power Regulator Mod - The mod involves desoldering the voltage regulator's middle leg to disable all circuits behind the device. Power is fed to all boards from an external power source. This reduces power consumption by disabling the power LED and two regulators. (4-80)
Secondary Mod (6-120) adjusts power consumption across LED
LED Second Mod (7-121) resistor add to lower ma consumed
Brain Secondary Phase Expansion (10-186) design with 135 boards twinned for a total of 270 Propellers, using the prop stack technique and the DIP version Propeller, another 135 chips can ride piggy-back bringing the total expansions to 405 props, equaling 3,480 processors
Brain Large LCD Mod (12-237) reversal of hardware to facilitate mounting
Dreaming
Vector Dreaming (8-157) draw dreaming as it happens in real time
Text Dreaming (8-157) dream composed of words to tell a story line
Graphical Dreaming (8-157) dreams represented with images
Internal Dreaming (8-157) dream happens internally with no output
Dream Folding (18-342) compressing a dream into a code
Dream Unfolding (18-342) construction of a dream by expanding code
Dream Language (18-342) LOGO programming language
Dream Equation (18-358) a calculus dream folding equation
Schematics
Serial Interfacing (18-353) two pin bi-directional true/inverted mode
Serial Interfacing (18-353) one pin bi-directional true mode
Serial Interfacing (18-353) one pin uni-directional true/inverted mode
Serial Interfacing (18-353) one pin bi-directional inverted mode
Brain Base (18-353) 2 props
Brain Stem (18-355) two boards, a BOE and a PPPB
Big Brain Milestones
Big Brain Idea is Born - Aug. 5th, 2010
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain
First Hybrid Brain Processor - Nov. 20th, 2010
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=955603&viewfull=1#post955603
First Robotic Brain Stem - Dec. 20th, 2010
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=962425&viewfull=1#post962425
First Brain Base - Dec. 21, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=962427&viewfull=1#post962427
First Working Robot Brain Blob - Dec. 24th, 2010
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain/page2
The 1st Working Brain Span Animation - Jan. 8, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=966472&viewfull=1#post966472
First Successful Brain Base Communications - Jan. 8, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain/page18
Invention of the First Exoskeleton - Jan. 9, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=966678&viewfull=1#post966678
Celebrating the Brain’s First Light - Feb. 18, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=978403&viewfull=1#post978403
The First Flythrough with a VR camera - Feb. 21, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=979268&viewfull=1#post979268
Brain Sings for the First Time - Feb. 27th, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=980641&viewfull=1#post980641
Brain Achieves Life - March 1, 2011
http://forums.parallax.com/showthread.php?124495-Fill-the-Big-Brain&p=981177&viewfull=1#post981177
continue with post #41
Announcement
(abbreviated HNT) Hyper-Neural-Threading technology is now introduced as a spinoff of the Neural Injection Technology - it's the Big Brain's method for simultaneous multithreading implementation across its aggregate of multiple Propeller CPUs.
HNT is utilized to improve the mass performance of collectives of multiple Propeller arrays and partitions. HNT improves "parallelization" of Propeller chips by extending their performances inclusive of performing multiple neural jobs at the same time.
Propeller chips can access and utilize the new HNT operating system that encompasses multiple workloads performed together. This is a Phase I HNT development.
With the first three partitions, the Brain becomes a 150 core Propeller processor, at 50 cores per partition, capable of now processing with extended "parallelization of performance. Phase I HNT is laying the groundwork for work performances that are shared between aggregate processors.
HNT is designed for the Propeller chip aggregate in a plurality of blocks, spans, and partitions. HNT designs run on 3 propellers (tested on a minimum of five Propellers) on up to the infinity of maximized partitions with hundreds or thousands.
Propeller DPs increase volume performance
The concept of Big Brain Domain Partition is now fully functional and tested. To function at maximized performances with reliability of clocking signatures, the Brain is divided up into Domain Partitions known as DPs.
A Big Brain can have an unlimited number of DPs as each DP connects to the Big Brain extending its number of core processors.
Each of ten Big Brain DPs can hold 500 Propeller chips and 16,000 I/Os for use.
The number of added DPs is limited by the following factors:
Added DPs directly increase Big Brain' ability to perform. Ten DPs increase volume processing by a factor of ten.
Photos of several DPs are posted in this thread.
The hold off on the most powerful supporting computer for the Brain is due to an upgrade for a more powerful model. In particular, a possible upgrade from the 720 processor GPU board to a 1,536 GPU streaming processors board.
The upgrade also includes another add on GHz speed (exceeds 3GHz) to the quad core processor. The core will remain Intel Quad core, however the monitor would increase from 15-inches to 27 inches for additional design work on a large screen.
What about physical size? Another portable computer is available for add-on portability with Brain activities while the more powerful "desktop" can remotely stream information a good distance by wireless. Another benefit, aside from maintaining similar cost package (adjusted), is the increase in the number of TFLOPS processing power and added 880MHz clock engine.
There is a 375MHz Memory Clock (5.5 Gbps GDDR5) with 176 GB/s memory bandwidth (maximum) and 2.7 TFLOPs Single Precision compute power (INT much faster) using TeraScale 3 Unified Processing Architecture.
Programming 1,536 GPU streaming processors and adding these to the Propeller collective will be addressed in a future post. More information will be provided after registration and enrollment is completed as Apple Developer.
This question recently came up concerning the Giant Brain and is it a Propeller project or one of other computers. The Brain is definitely a Propeller Brain project. It runs hundreds of Props by design. The Propeller aggregate is based on the experiences in developing the UltraSpark 40 and previous machines. Also known as a supermicrocontroller, the Brain has great capacity to control almost five thousand I/Os. Enhancements using Trinary States (not previously discussed for the Brain) will likely take this beyond a trillion. So the Propeller Brain is extremely powerful in terms of acting as a supermicrocontroller which is ideal for the many neural states of a large machine Brain.
Nehalem Superpower
http://www.gizmodo.com.au/2009/09/intel-core-i5-i7-processors-nehalem-superpowers-cheaper-than-ever/
But Brain is all about neural matter, neurons and high neuronal counts, especially in the processor array region of things. The ideas previously proposed are about to make their ways into the Propeller Brain and this includes more computers, primarily Apple Macs because these run in the TeraFLOP region and have an additional 1,540 processors to offer, and can offer considerable resources. The Quad Intel core alone has 383 million CPU transistors and 177 Million GPU transistors.
In conclusion, the Brain is a Propeller project with supporting add-on computers to gain additional resources.
Agreed. Mac has so many perks that make a neural net doable in terms that are different from Propellers. However, the props are outstanding for the supermicrocontroller aspect and hands on wiring with 4,800 I/O in the first 3 partitions while the Mac has other incredibly powerful resources which can be wired with software. It will be interesting to determine how many INITNeurons can fit in the Mac that has a Quad Core with 1,536 GPUs and over 10GB memory and to explore a new programming environment. Part of the enjoyment of the project of course is hands on wiring with props as it's still a Propeller Brain project. One aspect is to simply break off the Mac part of the Brain and have two Brain designs, one Mac and one Propeller. However, it's more likely these will develop separately and then merge.
Marrying AMD Radeon Technology Graphics GPU Boards
http://www.legitreviews.com/article/593/1/
Examples of modeled neurons
http://senselab.med.yale.edu/modeldb/ModelList.asp?id=3537
http://pinoytutorial.com/techtorial/amd-radeon-hd-6970m-review-specs-and-price-goes-official/
AMD Radeon HD 6970M It’s nice to know that AMD is coming out with this mobile chip that comes with 960 shaders, 2 GB of dedicated DDR5 memory and comes with a throughput of 115.2 GBps as well as the ability to churn 680 million polygons each and every second. Apparently, this mobile chip is the fastest chip ever released as of present and based on the various reviews conducted upon it, the results have been consistently positive. And even if its closest competitor Nvidia has managed to hold on to its crown via its new GeForce GTX 485M, the exciting features as well as the significant changes that AMD made on the HD5870M to come up with the new HD 6970M will surely make some impact as well.
In addition to this, you will surely notice the AMD is making some changes in its naming convention as well. From the very wordy Mobility Radeon HD 6970 it has now changed to a shorter one, which is the Radeon HD 6970M. This seems to be a pretty good move since a shorter name will be easier to recall. Here’s a visual chart comparison of Desktop vs. Mobility Radeon graphic cards (courtesy of tomshardware):
If the AMD 6970 programs similar to the Propeller in terms of the quantity of neural matter, it can generate 1,536,000 neurons in Phase I. Given the likelihood of going from 32K to 10GB, the number of neurons may approximately increase 312,500 times which is 4.8E11 neurons or around 480 billion. Keeping in mind that three Propeller partitions can handle 150,000 INITNeurons so the addition of the Mac resources increase this by 3,200,000 or 3.2 million times
In order to more readily understand the Big Brain project, a review of its many faces is helpful. The Brain began with a few Propeller Proto Boards and was a portable battery operated Brain.
A Propeller board assembly with a total of 20 boards and 40 props at hand was infused into the project. This phase is the "desktop" Brain.
In the next Brain phase, over a hundred Propellers were connected, making a massive Giant Brain. It was no longer office desktop but rather required a special simple made laboratory bench top. This is the "laboratory" Brain.
After the the giant laboratory Brain and before the desktop Brain, some smaller Brains were made. These are "small" Brains.
The next phase expanded with the connection of Mac computers. This is the "expanded" iBrain.
The next development phase encompasses using wireless, Mac resources, Apple programming, and adding more Apple processors (1,540). This incredibly more powerful version can possibly handle more software neurons into the billions of numbers and can include more functions that are currently unknown (spinoff technology). This is the "ultra" Brain. *the names may slightly change
Phases of the Big Brain
In this phase of the Giant Brain, large desktop macs become part of the Brain, sharing resources, with additional new functions and new programming languages. The concept is, this particular Apple Mac array would set on the desktop or Lab bench and remotely offer a very massive section to the overall Brain. Communications has numerous options, i.e. FireWire 800, ThunderBolt, etc. It is now generally thought that the most powerful version of the Big Brain will draw resources from not only the Propellers and Macs but the internet. This development is still a ways away but there is a plan of direction. In such a phase, Propellers can be detached from Macs and function independently, and visa versa, although the union of the two technologies will make a much more powerful machine.
The Propellers are very good for massively large numbers of I/O. A 3-partition Brain can have almost five thousand I/Os. The Mac doesn't have this I/O hardware built in like the prop.
Mac TFLOPS are remarkable. The number of TFLOPS is actually slightly more because only the GPU card was quoted. One would need to look up the spec sheet for the Quad Core Intel chip and add in that number too. It's probably around 26 GFLOPS in base 3.2GHz configuration.
http://www.intel.com/support/processors/sb/cs-023143.htm
Today the order was placed for more new Terabyte hard drives for Propeller Brain support backup of two Mac "Brain" computers which will contain Brain software, data, apps, programming and various support code. Eventually all USB 2.0 and USB 3.0 will be supplemented with ThunderBolt drives. LaCie Thunderbolt drives made for Apple are scheduled for release this summer indicating how new this high speed technology is on Apple computers.
a plane with quadrillions of Brain controllable states proposed
The interest is in creating a Massive Propeller Brain Trinary State Output Device that can exist in a plane with quadrillions of Brain controllable states. Since expanding the Brain, the number would be far greater and a number so big I don't know how to say it other than it could be represented in some base exponential. As far as exactly how these will be used inside a machine Brain remains to be seen.
http://www.mathcats.com/explore/reallybignumbers.html
http://forums.parallax.com/showthread.php?125757-More-Prop-Pins-amp-States&p=940544&viewfull=1#post940544
Let's examine how the Limbotronic Trinary system can work in the Big Brain. if you put a resistor divider on a pin and set it to input you now have a voltage level of one half vcc. So you can now signal 3 states on that pin to an external device. Use 2 pins and you can signal 9 states, 3 pins 27 states, 4 pins 81 states. So you can get more information out than you can in binary.
In the Limbotronic System, base 3 is very useful for the pin states and can be followed with a program. There is one pin on the Prop chip that is already is a state of Limbo degree, and that would be pin 31 when the chip is connected to USB. So with 31 usable Limbotronic pins in a single chip, the number of possible combinations is over a trillion, 617,673,396,383,947. If we could disconnect USB, then all 32 pins are usable and the number goes up to 1,235,345,972,567,894 which is now over one quadrillion possible combinations using a single prop chip. The Limbotronic Technique works here because it only requires the state wiring of 32 pins and software does the rest. The Brain with only two partitions could represent over one hundred quadrillion states. There's additional methods in representing trinary systems which will be discussed later.
A TRIT in base 3 is similar to a bit in base two. The three state logic system in base 3 is called a Ternary or Trinary base system. It is useful to use Trits, Tribbles, and Trytes. There is a TriINTERCAL programming language that defines unsigned 10-trit (0 to 59048) and 20-trit words. There are numerous Trinary Computers and architecture described on the internet.