The GA144 may actually have a problem with 'grid-lock' as once you get 4 deep into the grid, it takes 3 or more processors to reach an i/o node for output.
The Propeller avoids this issue by having the cog scheme and every processor can reach the outside in the same way.
Communication with hub RAM involves the cog scheme. I/O is direct.
Anyway, on the GA144 neither the SRAM or the i/o from the middle of the matrix can get to the required i/o pins without having several processors involved. The Propeller avoids these barriers.
It seems that with the GA144, one has an added programing task of how to route data through the matrix and that most tasks will require the use of more than one processor. Without a clock, threading multi-tasking seems impossible. One just has to avoid using processors that are assigned another task.
It seems that with the GA144, one has an added programing task of how to route data through the matrix and that most tasks will require the use of more than one processor. Without a clock, threading multi-tasking seems impossible. One just has to avoid using processors that are assigned another task.
The whole purpose of massively parallel constructs is the simultaneous operation of multiple processors each dedicated to their own tasks which can be independant or include multi-tasking.
@Humanoido
That may be true, but without efficent input and output, what is the point of a huge matrix? It seems to actually slow down i/o. And if each process is doing a specific tasks, why should a large portion of the processors also be forced to be a bucket-brigade to pass results to the output? A matrix of 1 x 144 might be more appealing htat a matrix of 8 x 18. I seems to me that things might get stuck in the middle. Even neural nets do not depend entirely on passing to one of four adjacent nodes.
I'd love to see what I am missing here.
The J1 Forth machines is a real head turner as it provides a good interface for LANs. That is a real immediate useful, economic valute. The NVidia graphic cards use massive parallel to handle video graphics, again an immediate useful, economic value ( I guess the i/o goes directly to video ram locations).
I suppose Cray super-computers, Big Blue and whatever have had their glory in serving a useful purpose as well. The true super-computers are doing wonderful things.
But these tiny chips with a lot of processors may be nothing more than a novelty -- especially if my desk top can crunch numbers quicker and handle tasks faster. If the race is won by the swiftest, there are the also rans.
That may be true, but without efficent input and output, what is the point of a huge matrix? It seems to actually slow down i/o. And if each process is doing a specific tasks, why should a large portion of the processors also be forced to be a bucket-brigade to pass results to the output? I'd love to see what I am missing here.
A very fast parallel thinking machine can rely on the sum total of its individual processors without the need to slow down by accessing real world input output. The idea is to do the processing and assemble the result without any i/o. Slower i/o is only used to output the result one time, and is not a factor in an application that suits this technique. This is a method that can be used for some of these parallel Propeller machine projects made from multiple Propeller chips described in these blogs.
@Humanoido
I know you believe that, but does that include reaching RAM indirectly? I suspect much depends on what is inside and what is outside.
It is really difficult to get anywhere talking about processor architecture with just generalizations. At some point, to optimize one has to address the details.
The Propeller has internal RAM reached directly - each Cog has 2K and the HUB is 32K. In the link app, there is no need to wait for external RAM to slow down a parallel process. That's the beauty and the genius of a Parallax Propeller chip array.
I agree, it's rather difficult to generalize about a GA processor that one does not have, has not used, has not programmed, and is not made by or supported by Parallax with their extensive documentation. I would say try the GA Forum but there is none.
You don't get it. If you are building a really advanced artificial inteligence with parallel processing, say an "Aritificial Philsopher" (AP). Then it does not need any input or output. Of course it would not help if it had any output as no one will understand what it has been thinking anyway:)
@Heater
So,... the 144 matrix technology is for virtual perfection, right?
Due to the Chnese New Year holiday, I have yet to have my GA144 arrive and am beginning to feel that User Name was right about the 'bit-banging' style of the Green Arrays.
If you have Random Access Memory attached, it would certainly be more useful if multiple processors could read or write to different portions with at least the same amount of speed, maybe with simultaneous access, and certainly not by passing data through a chain of processors that are excluded from other things.
Since the GA`144 parallel ports are in one processor on the edge, there are actually only 3 other processors that are directly reachable; after that, one has maybe just use another 3-6 processors as data pipes to spread out a bit more and task maybe 9 processors or more (in clusters) with real work.
@Humanoido
There actually is a GA144 forum, but I figure that the comparison is where one really learns something. I have looked a lot at what Forth is doing on microcontroller chips over the past year and so far, the Propeller 1 really offers what seems to be the most viable platform. Other devices have been adapted, but the combination of speed and multiple processors work together extremely well. And it seems it will be even better with the Propeller 2.
Years ago, stuff that was not available was called "Vaporware."
There are many sites dedicated to the philosophy of artificial intelligence. However, artificial intelligence philosophy is a relatively unknown field.
The concept of pure thought within a chip was developed in 2012 with the PE machine that had no parts.
Discussion about the GA144 has led to a point about the restrictive nature that a dedicated multiprocessor chip may have.
I think the great latitude and freedom of taking multiple Propeller chips and wiring for what you want is a godsend. Look, you can take any number of props and do wiring in between the chips, or simply put all the chips in parallel to do things instantly and simultaneous.
If I want some funky high speed processing in my phone or wrist watch or other tiny device. Say speech recognition or whatever. Then I might be able to do it with Green Array chips as they advertise. I might be able to do it with Parallella chips, as they advertise. Perhaps, maybe, such things will come to pass with such devices.
For sure I won't be able to do it with no matter how many Propellers.
iPhone and other phones have already accomplished everything you mentioned. One Propeller chip 1 can already do speech recognition and speech synthesis and you have to admit, even though this prop program has not had millions or billions of money and support from Apple Computer Corp. over several years, it functions well for my projects with speech recognition and speech synthesis requirements.
I have gotten ArrayForth as an emulator on a x86 running. It is also the IDE that the GA144 is dependent upon. So you MUST learn Forth and you MUST learn ColorForth - which is all very exotic in its own right.
After that, you can evoke 'sofesim' to have a simulation of 1/4 of the processors of a GA144. My first impression is that I am watching something that is very similar to a game of Pacman. Why I need to have data snake from one proceesor to the next is beyond me, but it certainly can to that.
There is an alternative Forth, called eForth that might be installed on a GA144 with the right SRAM and ROM, but that is in even a more primitive state than the ArrayForth... and you need the ArrayForth to first load the SRAM and ROM.
About all I can concluded is Charles Moore has been having fun do as he pleases.
Mr. Moore mentions polyForth as being available for the GA144 as well, but that seems to be best explored with the bundle of software provided with the GA144 development board. I only am buying a chip.
~~~~~~~~~~~~~
Architecture is architecture.
The Propeller seems to have a good one, I am not too sure about the GA144 at this point as there are so many milestones in the learning curve, it is difficult to catch on to what is so great about it.
~~~~~~~~~~~~~
Virtual stuff...
Methinks...You are on a slippery slope of falling down a rabbit hole as Alice did. The mind can play tricks on everyone. These days we talk about worm holes that fold space, whereas the inherent meaning of space is a dimension that cannot be folded. And between worm holes and black holes, we have endless dialogue about what might be, but holes are really a 2D concept and space is 3D, and more.
Yes, yes, an ARM on an iPhone or whatever can do many wonderful things. Perhaps my examples are bit too limited. The promise of all these array processors is to be able to do a lot more. And do it in a small space with low power consumption suitable for portable battery powered devices. How that works out we shall see.
Clearly the Propeller is not in that league and will not be for the foreseeable future. Not that it was ever intended to fill that niche anyway.
I suspect that if Chip sets his mind to a massively parallel device it will be something you can actually program:)
He was looking for flat land with a dark sky and wanted to build a domed telescope observatory controlled by a GA chip. However, he settled in the light pollution of Carson City. Let's hope he can now relocate and realize his dream since the law suit is over.
Have patience. You will solve the intricacies of the GA chip after you have one in your hands. It's like you're trying to test the results of various filaments for the light bulb but the filaments have not yet arrived from South Africa. Since you have Forth installed on another platform, why not write general programs that you plan to execute on the chip?
You could also start with Propeller chips connected in parallel to test the concepts of multi-chip programming. I think you could create many algorithms and lay the groundwork for various experiments. You may even find ways to inject deterministic Propeller chips into the GA equation to create more powerful hybrid machines.
Indeed, about virtual, all the above is correct, which is what makes it so interesting. Note, a two dimensional hole printed on a flat paper becomes a three dimensional hole in space which can undergo alteration to become n-dimensions when gravity, space-time, dark matter, or other stuff is intermixed. A black hole or worm hole is not just a hole as we know it, though the simplicity of the word would suggest it.
Just because the human mind cannot comprehend it, doesn't mean it doesn't exist. Take for example, dark matter. For years, cosmologists knew there was something unattributable and incomprehensible until finally it was discovered.
But now, no one knows how it's woven with the quantum reality that surrounds us. Slippery slope indeed! We've opened the door to a Quantum World but we don't know how to step through. This is most likely because the quantum world is not only woven with infinitesimal space with its own physics but also includes the relativity of alternate dimensional protected time with interfacing connections to the EOU boundary. Only now are we beginning to develop the tools necessary to explore and simulate elemental behaviors of quantum reality.
Like the shipments of the filaments, we must have an understanding patience of time to unravel the mysteries of the Universe. We can ask where it will all lead, but we may never know in our lifetime, as yet we may have our hands more full with the interpretation of our own unique meaning of life.
Yes, yes, an ARM on an iPhone or whatever can do many wonderful things. Perhaps my examples are bit too limited. The promise of all these array processors is to be able to do a lot more. And do it in a small space with low power consumption suitable for portable battery powered devices. How that works out we shall see. Clearly the Propeller is not in that league and will not be for the foreseeable future. Not that it was ever intended to fill that niche anyway. I suspect that if Chip sets his mind to a massively parallel device it will be something you can actually program:)
No doubt, with a handful of Propellers and Apple's budget, you could build a great phone. It seems Parallax has already considered the chip size with the P8X32A-M44. (I really don't need a smaller phone)
Chip G. is undoubtedly now hatching plans for a Massively Parallel Propeller chip.
There is no way you could build a great phone, think iPhone or Android, with a handful l of Propellers. At least the ones we have today. The Prop cannot deliver the performance of existing ARMs and GPU's with low power consumption as required by a phone. Let's have a hint of physical reality here.
...and you MUST learn ColorForth - which is all very exotic in its own right. After that, you can evoke 'sofesim' to have a simulation of 1/4 of the processors of a GA144. My first impression is that I am watching something that is very similar to a game of Pacman. Why I need to have data snake from one proceesor to the next is beyond me, but it certainly can to that.
Chuck Moore has provided examples of his learning curve complete with screen shots. The incremental worm from one processor to the next is a validation demo. Perhaps more useful is a parallel hit upon all processors simultaneously.
@Humanoido
I suppose you are right. Ever slowly I a trying to work through all the literature that has been provided in hopes of seeing how to really exploit the might GA144.
ColorForth in and of itself is a bit of a learning curve... even if you know older approaches to Forth. For those familiar with colon definitions, the colon is gone and now a color signifies the naming of a new word. And since only the first 4-5 bytes of a name are actually recognized, these new names are rather terse and not evocative of meaning.
ColorForth requires one to use your keyboard in an entirely different manner, except when in 'text' mode. ColorForth doesn't have files, it has Blocks, just like the book "Starting Forth". Very quaint.
The Core document for getting started is an Introduction to ArrayForth and it clearly states that ArrayForth - a specialized ColorForth - was and is the in house development software for Green Array chips. There is nothing else unless you desire to create your own.
At this point, I am a bit uncertain if one needs to bother with eForth, but I did quite a bit of reading about it because it has a loadable image for the GA144. eForth was supposed to be the ultimate in portable Forth as it has only 32 primitives and the rest is created in Forth Words. I can't quite see the need to port eForth code, but then I never invested in writing code in it. It may be a way to step away from ColorForth and into a more traditional Forth code environment, if you want to extend the 32 primatives into about 200 words... there are materials that might do all this for you.
Also, the bulk of the documentation is the code itself stored in Blocks with Shadow Blocks as explanation and commentary. So you have to really run ColorForth to see what is available. There may be one text available for purchase, but one has to mostly study at a computer an in ArrayForth. One can dump Lists and HTML images of the Blocks, if you want to get away from the computer. A screen dump is available and called PNG.
It seems now that a Block of code can easily be created to load multiple processors for a given task. And it may be the ROM within each processor may be redeployed as one wishes at the initialization. I am now studying the IDE document. If this is the case, the actual loading of the processor nodes is simply done in some miraculous flash and not as big an issue as I feared it would be. Processors are supposedly running at the equivalent of 666Mhz when running run out, so waiting for loads may not be annoying.
I still do NOT have the chip on hand, but at the least there seems to be a lot this chip might do if one gets deep enough into all the items available. It is very, very different and very much Forth in its approach. For instance, rather than have a primitive for the processor provide Zero, one does DUP OR to zero (you just duplicate the existing 18 bit word and OR it).
Either a person will love the novelty and the challenge or hate it. I am just very wary that Forth leads to huge arguments and while that seems to be part of the Forth tradition, I wonder why it must be. Actually having 144 processors available demands a very different point of view, than 8 or so.
Since the GA144 hasn't arrived and it is still Chinese New Years, I've been heavily into reading the tutorials. This stuff is simply so different that one really has to want to work hard to get it.
I now am reading a bit of Chuck Moore's code that attaches a 10Mhz resonator to one i/o pin (maybe any general purpose i/o pin) and allows one to generate VGA images. Very very weird stuff.
The good news is that I checked the USPS tracking today an the GA144 has arrived in Taiwan and cleared customs, so it just depends on when the postman gets back to work. My land line went haywire at the beginning of the new year holiday, so I am also waiting for a telephone repair. And I broke my glasses on the day before Chinese New Years, so I am a bit bleary eyed on a spare pair of oldies.
This will be a holiday to remember... but I feel I am grasping more of what Charles Moore wants Forth to do.
Sorry to hear about the glasses breaking, the phone quirks and the long postal holiday. CNY does have its challenges. Here, the supply of quality food ran out. I can see by the returning pollution in the sky, most businesses are getting back to work and food will be on the table once again.
The GA story reads like an exciting saga and we wonder how it will end up relative to the usage of parallel chips including both the GA144 and the enhanced core multi Propeller, like there could be some joint coop between the two, or giant comparisons of inner workings from one to the other that expand thinking in ways that C. Moore and C. Gracey could appreciate.
I'm looking for less expensive ways to create larger multi processor machines and a cooperation between chips could lead to a working hybrid. If, and this is currently a big if, the GA144 could have the prop algorithms applied to it with a hundred chips, we could extend and expand the Big Brain with over 14,400,000 million additional processors, taking it to the next level. Based on the speed of the chip, the actual performance is more like 1,440,000,000 processors. Exceeding a billion processors between prop and ga could certainly make some exciting experiments for several apps. Ok, just dreaming, but this is how ideas are born and become reality.
Well, the GA144 can easily connect with the Propeller via RS232, SPI, parallel, or GreenArray's SERDES (an extreme high speed synchronous serial). Since the GA144 is a 1.8V device, it is likely to be more friendly with the Propeller 2.
But if you are not interested in learning ArrayForth, there is not much point in doing so. Nobody is providing any alternative IDE for this chip. Dave Hein was wanted C for the chip, but it may never evolve. This is Forth through and through.
Just wondering, where is the link to the Green Array Forum? I found a Yahoo group with only 10 members, and a few postings, but no Forum conducted by GA.
But if you are not interested in learning ArrayForth, there is not much point in doing so. Nobody is providing any alternative IDE for this chip. Dave Hein was wanted C for the chip, but it may never evolve. This is Forth through and through.
Actually, I have little interest in the GA144. The developers might generate more interest in the chip if they provide C tools for it. However, given that Chuck Moore has wore blinders most of his life, and is mired in the Forth world, it's unlikely that that would ever happen. If the developers really want to spark interest in their chip they should come up with something useful to do with it. Instead, they seem to be relying on the Forth community to do it for them.
I may have been mistaken, you may have found the only one or there may be another in Google. GreenArrays has a course registration that I have not yet done, but it is free to all.
What I have done is to get the ArrayForth working in Windows7 and learned from the following:
A. The 'softsim'
B. The ArrrayForth Editor
C. The ArrayForth IDE
D. The introductory ArrayForth htm documents
E. Charles Moore's blog about the GA144
And review material in Starting Forth as ArrayForth uses all the concepts provided therein. (PropForth and TachyonForth do not.)
GreenArrays has a tutorial that was written by a European that is quite helpful. It also has an IDE pdf document that is quite important.
My situation is that I must adapt the ArrayForth to a bare bones device ... no RAM, no ROM, just an RS232 boot port. I don't have their Development Board.
I could have bought the $450 Development Board and there might be more to play with and an easier time with getting up and running, but I am learning more by doing it this way.
My impression is that Charles Moore is having no trouble at all in using a GA144 Development Board at full capacity, but there are very few souls that are in Forums that can keep up with what he is able to do. The GA144 is a challenge.
The Forth Community just doesn't seem to have much 'new blood'. ... participants are about as common as veterans of WWI. I guess that is why I am so curious. Is it worth saving Forth or just letting it disappear is wise? I"d never do this for CPM.
Delivery of my GA144 is still in the making. Then may you have someone outside of GreenArrays doing something. There are quite a few test runs in ArrayForth that I hope to try.
CPM is a lousy file system. MS-DOS was a big improvement. Unix is even better. Some things are just left behind in a natural progression, others die from ignorance of their value.
I suspect Forth is in the latter category. Unix and C evolved for collaboration in large projects; Forth seems to have evolved for independent souls that gaze at the night sky and ponder what a computer might do for them.
What does Forth have that C and Unix does not? Romance.....
Comments
Communication with hub RAM involves the cog scheme. I/O is direct.
I see what you are saying. Good point.
Anyway, on the GA144 neither the SRAM or the i/o from the middle of the matrix can get to the required i/o pins without having several processors involved. The Propeller avoids these barriers.
It seems that with the GA144, one has an added programing task of how to route data through the matrix and that most tasks will require the use of more than one processor. Without a clock, threading multi-tasking seems impossible. One just has to avoid using processors that are assigned another task.
That may be true, but without efficent input and output, what is the point of a huge matrix? It seems to actually slow down i/o. And if each process is doing a specific tasks, why should a large portion of the processors also be forced to be a bucket-brigade to pass results to the output? A matrix of 1 x 144 might be more appealing htat a matrix of 8 x 18. I seems to me that things might get stuck in the middle. Even neural nets do not depend entirely on passing to one of four adjacent nodes.
I'd love to see what I am missing here.
The J1 Forth machines is a real head turner as it provides a good interface for LANs. That is a real immediate useful, economic valute. The NVidia graphic cards use massive parallel to handle video graphics, again an immediate useful, economic value ( I guess the i/o goes directly to video ram locations).
I suppose Cray super-computers, Big Blue and whatever have had their glory in serving a useful purpose as well. The true super-computers are doing wonderful things.
But these tiny chips with a lot of processors may be nothing more than a novelty -- especially if my desk top can crunch numbers quicker and handle tasks faster. If the race is won by the swiftest, there are the also rans.
A very fast parallel thinking machine can rely on the sum total of its individual processors without the need to slow down by accessing real world input output. The idea is to do the processing and assemble the result without any i/o. Slower i/o is only used to output the result one time, and is not a factor in an application that suits this technique. This is a method that can be used for some of these parallel Propeller machine projects made from multiple Propeller chips described in these blogs.
http://humanoidolabs.blogspot.tw/p/about-big-brain.html
http://humanoidolabs.blogspot.tw/p/genealogy.html
I know you believe that, but does that include reaching RAM indirectly? I suspect much depends on what is inside and what is outside.
It is really difficult to get anywhere talking about processor architecture with just generalizations. At some point, to optimize one has to address the details.
I agree, it's rather difficult to generalize about a GA processor that one does not have, has not used, has not programmed, and is not made by or supported by Parallax with their extensive documentation. I would say try the GA Forum but there is none.
You don't get it. If you are building a really advanced artificial inteligence with parallel processing, say an "Aritificial Philsopher" (AP). Then it does not need any input or output. Of course it would not help if it had any output as no one will understand what it has been thinking anyway:)
So,... the 144 matrix technology is for virtual perfection, right?
Due to the Chnese New Year holiday, I have yet to have my GA144 arrive and am beginning to feel that User Name was right about the 'bit-banging' style of the Green Arrays.
If you have Random Access Memory attached, it would certainly be more useful if multiple processors could read or write to different portions with at least the same amount of speed, maybe with simultaneous access, and certainly not by passing data through a chain of processors that are excluded from other things.
Since the GA`144 parallel ports are in one processor on the edge, there are actually only 3 other processors that are directly reachable; after that, one has maybe just use another 3-6 processors as data pipes to spread out a bit more and task maybe 9 processors or more (in clusters) with real work.
@Humanoido
There actually is a GA144 forum, but I figure that the comparison is where one really learns something. I have looked a lot at what Forth is doing on microcontroller chips over the past year and so far, the Propeller 1 really offers what seems to be the most viable platform. Other devices have been adapted, but the combination of speed and multiple processors work together extremely well. And it seems it will be even better with the Propeller 2.
The Propeller 2 looks better and better to me.
There are many sites dedicated to the philosophy of artificial intelligence. However, artificial intelligence philosophy is a relatively unknown field.
The concept of pure thought within a chip was developed in 2012 with the PE machine that had no parts.
http://humanoidolabs.blogspot.tw/2012/02/pure-thought.html
http://humanoidolabs.blogspot.tw/2012/02/propeller-elf.html
Discussion about the GA144 has led to a point about the restrictive nature that a dedicated multiprocessor chip may have.
I think the great latitude and freedom of taking multiple Propeller chips and wiring for what you want is a godsend. Look, you can take any number of props and do wiring in between the chips, or simply put all the chips in parallel to do things instantly and simultaneous.
From 2010 and prior, one can find project development history relative to the Propeller in parallel.
http://humanoidolabs.blogspot.tw/p/history.html
For sure I won't be able to do it with no matter how many Propellers.
Well, untill Chip gets up to the Prop 3 or 4
Is pure thought a virtual process?
iPhone and other phones have already accomplished everything you mentioned. One Propeller chip 1 can already do speech recognition and speech synthesis and you have to admit, even though this prop program has not had millions or billions of money and support from Apple Computer Corp. over several years, it functions well for my projects with speech recognition and speech synthesis requirements.
I have gotten ArrayForth as an emulator on a x86 running. It is also the IDE that the GA144 is dependent upon. So you MUST learn Forth and you MUST learn ColorForth - which is all very exotic in its own right.
After that, you can evoke 'sofesim' to have a simulation of 1/4 of the processors of a GA144. My first impression is that I am watching something that is very similar to a game of Pacman. Why I need to have data snake from one proceesor to the next is beyond me, but it certainly can to that.
There is an alternative Forth, called eForth that might be installed on a GA144 with the right SRAM and ROM, but that is in even a more primitive state than the ArrayForth... and you need the ArrayForth to first load the SRAM and ROM.
About all I can concluded is Charles Moore has been having fun do as he pleases.
http://colorforth.com/blog.htm
Mr. Moore mentions polyForth as being available for the GA144 as well, but that seems to be best explored with the bundle of software provided with the GA144 development board. I only am buying a chip.
~~~~~~~~~~~~~
Architecture is architecture.
The Propeller seems to have a good one, I am not too sure about the GA144 at this point as there are so many milestones in the learning curve, it is difficult to catch on to what is so great about it.
~~~~~~~~~~~~~
Virtual stuff...
Methinks...You are on a slippery slope of falling down a rabbit hole as Alice did. The mind can play tricks on everyone. These days we talk about worm holes that fold space, whereas the inherent meaning of space is a dimension that cannot be folded. And between worm holes and black holes, we have endless dialogue about what might be, but holes are really a 2D concept and space is 3D, and more.
Clever stuff.... but where does it all lead?
Clearly the Propeller is not in that league and will not be for the foreseeable future. Not that it was ever intended to fill that niche anyway.
I suspect that if Chip sets his mind to a massively parallel device it will be something you can actually program:)
Have patience. You will solve the intricacies of the GA chip after you have one in your hands. It's like you're trying to test the results of various filaments for the light bulb but the filaments have not yet arrived from South Africa. Since you have Forth installed on another platform, why not write general programs that you plan to execute on the chip?
You could also start with Propeller chips connected in parallel to test the concepts of multi-chip programming. I think you could create many algorithms and lay the groundwork for various experiments. You may even find ways to inject deterministic Propeller chips into the GA equation to create more powerful hybrid machines.
Indeed, about virtual, all the above is correct, which is what makes it so interesting. Note, a two dimensional hole printed on a flat paper becomes a three dimensional hole in space which can undergo alteration to become n-dimensions when gravity, space-time, dark matter, or other stuff is intermixed. A black hole or worm hole is not just a hole as we know it, though the simplicity of the word would suggest it.
Just because the human mind cannot comprehend it, doesn't mean it doesn't exist. Take for example, dark matter. For years, cosmologists knew there was something unattributable and incomprehensible until finally it was discovered.
But now, no one knows how it's woven with the quantum reality that surrounds us. Slippery slope indeed! We've opened the door to a Quantum World but we don't know how to step through. This is most likely because the quantum world is not only woven with infinitesimal space with its own physics but also includes the relativity of alternate dimensional protected time with interfacing connections to the EOU boundary. Only now are we beginning to develop the tools necessary to explore and simulate elemental behaviors of quantum reality.
http://humanoidolabs.blogspot.tw/search?q=quadlyzer
Like the shipments of the filaments, we must have an understanding patience of time to unravel the mysteries of the Universe. We can ask where it will all lead, but we may never know in our lifetime, as yet we may have our hands more full with the interpretation of our own unique meaning of life.
Chip G. is undoubtedly now hatching plans for a Massively Parallel Propeller chip.
I suppose you are right. Ever slowly I a trying to work through all the literature that has been provided in hopes of seeing how to really exploit the might GA144.
ColorForth in and of itself is a bit of a learning curve... even if you know older approaches to Forth. For those familiar with colon definitions, the colon is gone and now a color signifies the naming of a new word. And since only the first 4-5 bytes of a name are actually recognized, these new names are rather terse and not evocative of meaning.
ColorForth requires one to use your keyboard in an entirely different manner, except when in 'text' mode. ColorForth doesn't have files, it has Blocks, just like the book "Starting Forth". Very quaint.
The Core document for getting started is an Introduction to ArrayForth and it clearly states that ArrayForth - a specialized ColorForth - was and is the in house development software for Green Array chips. There is nothing else unless you desire to create your own.
At this point, I am a bit uncertain if one needs to bother with eForth, but I did quite a bit of reading about it because it has a loadable image for the GA144. eForth was supposed to be the ultimate in portable Forth as it has only 32 primitives and the rest is created in Forth Words. I can't quite see the need to port eForth code, but then I never invested in writing code in it. It may be a way to step away from ColorForth and into a more traditional Forth code environment, if you want to extend the 32 primatives into about 200 words... there are materials that might do all this for you.
Also, the bulk of the documentation is the code itself stored in Blocks with Shadow Blocks as explanation and commentary. So you have to really run ColorForth to see what is available. There may be one text available for purchase, but one has to mostly study at a computer an in ArrayForth. One can dump Lists and HTML images of the Blocks, if you want to get away from the computer. A screen dump is available and called PNG.
It seems now that a Block of code can easily be created to load multiple processors for a given task. And it may be the ROM within each processor may be redeployed as one wishes at the initialization. I am now studying the IDE document. If this is the case, the actual loading of the processor nodes is simply done in some miraculous flash and not as big an issue as I feared it would be. Processors are supposedly running at the equivalent of 666Mhz when running run out, so waiting for loads may not be annoying.
I still do NOT have the chip on hand, but at the least there seems to be a lot this chip might do if one gets deep enough into all the items available. It is very, very different and very much Forth in its approach. For instance, rather than have a primitive for the processor provide Zero, one does DUP OR to zero (you just duplicate the existing 18 bit word and OR it).
Either a person will love the novelty and the challenge or hate it. I am just very wary that Forth leads to huge arguments and while that seems to be part of the Forth tradition, I wonder why it must be. Actually having 144 processors available demands a very different point of view, than 8 or so.
Since the GA144 hasn't arrived and it is still Chinese New Years, I've been heavily into reading the tutorials. This stuff is simply so different that one really has to want to work hard to get it.
I now am reading a bit of Chuck Moore's code that attaches a 10Mhz resonator to one i/o pin (maybe any general purpose i/o pin) and allows one to generate VGA images. Very very weird stuff.
The good news is that I checked the USPS tracking today an the GA144 has arrived in Taiwan and cleared customs, so it just depends on when the postman gets back to work. My land line went haywire at the beginning of the new year holiday, so I am also waiting for a telephone repair. And I broke my glasses on the day before Chinese New Years, so I am a bit bleary eyed on a spare pair of oldies.
This will be a holiday to remember... but I feel I am grasping more of what Charles Moore wants Forth to do.
The GA story reads like an exciting saga and we wonder how it will end up relative to the usage of parallel chips including both the GA144 and the enhanced core multi Propeller, like there could be some joint coop between the two, or giant comparisons of inner workings from one to the other that expand thinking in ways that C. Moore and C. Gracey could appreciate.
I'm looking for less expensive ways to create larger multi processor machines and a cooperation between chips could lead to a working hybrid. If, and this is currently a big if, the GA144 could have the prop algorithms applied to it with a hundred chips, we could extend and expand the Big Brain with over 14,400,000 million additional processors, taking it to the next level. Based on the speed of the chip, the actual performance is more like 1,440,000,000 processors. Exceeding a billion processors between prop and ga could certainly make some exciting experiments for several apps. Ok, just dreaming, but this is how ideas are born and become reality.
But if you are not interested in learning ArrayForth, there is not much point in doing so. Nobody is providing any alternative IDE for this chip. Dave Hein was wanted C for the chip, but it may never evolve. This is Forth through and through.
What I have done is to get the ArrayForth working in Windows7 and learned from the following:
A. The 'softsim'
B. The ArrrayForth Editor
C. The ArrayForth IDE
D. The introductory ArrayForth htm documents
E. Charles Moore's blog about the GA144
And review material in Starting Forth as ArrayForth uses all the concepts provided therein. (PropForth and TachyonForth do not.)
GreenArrays has a tutorial that was written by a European that is quite helpful. It also has an IDE pdf document that is quite important.
My situation is that I must adapt the ArrayForth to a bare bones device ... no RAM, no ROM, just an RS232 boot port. I don't have their Development Board.
I could have bought the $450 Development Board and there might be more to play with and an easier time with getting up and running, but I am learning more by doing it this way.
My impression is that Charles Moore is having no trouble at all in using a GA144 Development Board at full capacity, but there are very few souls that are in Forums that can keep up with what he is able to do. The GA144 is a challenge.
The Forth Community just doesn't seem to have much 'new blood'. ... participants are about as common as veterans of WWI. I guess that is why I am so curious. Is it worth saving Forth or just letting it disappear is wise? I"d never do this for CPM.
Delivery of my GA144 is still in the making. Then may you have someone outside of GreenArrays doing something. There are quite a few test runs in ArrayForth that I hope to try.
Why not? There is a Forth for CP/M I'm sure.
I suspect Forth is in the latter category. Unix and C evolved for collaboration in large projects; Forth seems to have evolved for independent souls that gaze at the night sky and ponder what a computer might do for them.
What does Forth have that C and Unix does not? Romance.....