Any forth users interested....
prof_braino
Posts: 4,313
...in going in on an order for the greenarrays forth chip?
Recent posting on hackaday has put a bug in my bonnet. Not like I need another major project, but this has been on the list since Chuck Moore started talking about a successor to the Sh-boom
Minimum order is 10 chips at $20 each.
http://www.designspark.com/content/hands-144-core-processor
We still have to deal with solder the chip onto a board etc.
Sal and I would claim two of the ten, are there 8 more folks interested?
These would be cheaper than using 18 props to get 144 cores. Using propforth, we would actually need 24-28 props to get 144 usable cores.
The greenarrays architecture is significantly different from the prop, so propforth is very different from arrayForth. But at least they at both forth, so your brain only has to bend in one direction.
Reply or PM me.
Recent posting on hackaday has put a bug in my bonnet. Not like I need another major project, but this has been on the list since Chuck Moore started talking about a successor to the Sh-boom
Minimum order is 10 chips at $20 each.
http://www.designspark.com/content/hands-144-core-processor
We still have to deal with solder the chip onto a board etc.
Sal and I would claim two of the ten, are there 8 more folks interested?
These would be cheaper than using 18 props to get 144 cores. Using propforth, we would actually need 24-28 props to get 144 usable cores.
The greenarrays architecture is significantly different from the prop, so propforth is very different from arrayForth. But at least they at both forth, so your brain only has to bend in one direction.
Reply or PM me.
Comments
As such I belive there is no future for this green array effort.
( fignition.com That one comes with a complete IDE onboard, mostly written in Forth. )
*Sigh*
Really need to find the time to play with it soon...
Perhaps Forth is a fundamental property of the computing ethers that can be summoned by the faithful in any form so required??
LOL,
You know, I used to think that every programmer should learn assembly language. To get down to the basics, the fundamentals, to understand what a computer really is. Then they could continue with their C/C++, Java, BASIC, Python, whatever with a good understanding of what they are doing.
Recently I have realized that this is all wrong. An assembly language is basically adhearing to an architecture, an instruction set that itself is a language just like C or C++ or Javascript. Yes it is simpler and more fundamental and yes it is built into the hardware but still it is just a language that is whatever it is according to the whims of the CPU designer.
So even down there in assembler we are far removed from the "computing ethers":)
Sorry, I'm tired and getting a bit philosophical.
Some sort of BOB would be a necessity.
Besides, if I'm honest with myself, I've already got far too many pending unfinished projects going. ;-)
The good news is that eForth and Polyforth are running on it now. Both AFAIK need external SRAM.
I really wish Moore would have brought the F21 to market or at least kept the Mup21 going. Both were very hacker friendly and could be grasped in afternoon. Moore forgot about us mere mortals and instead concocted a chip that makes sense to him and handful of others but that's it.
@prof_braino: I would have jumped at this opportunity a month ago. Now I'm up to my neck in other projects, including a Forth engine on an FPGA.
Well, In order to get programs to run on computers one has to be pedantic or it won't work. If one is not pedantic the machine will do it for you:)
Am I right in assuming you are confirming the truth of my statement?
I don't mind Forth being more a philosohy than a language. I have respect for anyone who can put together such a develoment environment in Forth or otherwise and would not want to deteract from that.
It's was a misunderstanding on my part and now that I know better I will not be expecting Forth to provide cross-platform, portable, solutions. Which probably means I don't have much use for it.
Clint Eastwood was in it. Can you guess where he went?
There's a difference you know or are you just slagging Forth because it doesn't meet your particular ideology?
What is an embedded system? And/or what is the difference between one of those and a desktop system?
With the Propeller we are dealing with a system more powerfull than the desktop systems I was using in 1982 http://en.wikipedia.org/wiki/Intertec_Superbrain Which happens to be the year I first picked up Leo Brodie's book "Starting Forth" and did my first experiments in Forth on said desktop machine. http://home.iae.nl/users/mhx/sf.html
So perhaps I was a bit surprised to find that after 30 years of Forth we have three or more implementations on the Propeller that are incompatible.
Sorry if my "slagging" was a bit over the top. Can we call it "playing devils advocate" as I really don' have an ideology about programming languages or processor architectures or such like?
There are a whole bunch of languages that call themselves Basic which are mutually incompatible. The same is true for Scheme, Lisp, XLisp, and Clojure. So it seems like Forth is in good company.
Actually, the standardization of C and JavaScript strikes me as the exception, not the rule. Both of these propagated via the network effects of Unix and the Web respectively. Java didn't fragment because Sun aggressively sued nonconformist implementations as a violation of their license agreement. Now that Google won the lawsuit over their clean room version, that's over with, so programs compiled for Android won't run on other JVM's.
Let's not confuse Java and JavaScript though they are two totally different things.
I would not worry so much that Java compiled for one VM won't run on another as long as the source code is portable between the two. After all we can compile Java to native code as well. Of cource that does not apply to Android apps as the entire class library system they live in is different from the standard Java classes. "Write once, run anywhere" How I laughed when I first read that discription of Java years ago.
Don't foregt FORTRAN, COBOL, Ada, and so on are pretty well standardized as well.
Seems that anyone proficient in programming eventually realizes how bad the language they are using is, no matter what it is, and a has a desperate urge to create a new better language. As a result we now have thousands of languages and dialects.
Still this is all way off topic here I guess.
So, I guess Heater. doesn't want one. Only 4 or 5 left as of post #3.
FigForth is kind of a "kitchen sink" implementation closer to what Heater is looking for, its not something you master so much as start to dig through. It was made by committee, some of the desicions are misleading as they point in different directions.
Propforth and colorForth (and so I think arrayForth as well) were made by one guy more or less, and follow a more coherent design. And colorforth was made by Moore, so its exactly the minimum necessary and suitable for forth. Propforth is as close to as we can get to Moore's forth without getting Moore to do it for us.
True. It should be "every MICRO controller program should learn assembly", workstation programmers simply need to learn to design properly.
Looks like that covers the on-topic part of the list so far, and I am getting drawn off-topic as well.
So far 5 of 10 are claimed. Remember, we still have to buy a schmart board, and do some surface mount soldering. One guy wants to pay me to solder his as well, so that's also an option. I think I'll get better each unit, but its still a risk.
No confusion on my part. Right now the future looks brighter for JavaScript than it does for Java. Java is mostly confined to Android and Servlet programming, In the Servlet space it will likely be replaced by Groovy while no successor to JavaScript is on the horizon.
Here's something to consider:
Q: What's the definition of a natural language?
A: A dialect with an army.
It's probably true for programming languages as well. Except the armies are called user's groups.
I think you have to consider the above chip as a study in parallel processing tasks, with Forth as the means to explore creatively. Forth really requires you have a vision for what you want to control, or output, or whatever. Just writing big dictionaries that are supposed to be self-documenting gets you into perpetual read/write/read some more.
Mastering FigFORTH is a rather insane undertaking - tries to be all and everything to everyone (something the OOP have gotten in trouble with also doing).
But if you consider the control of an un-manned ROV in the deep sea, that can be done quite nicely.
If and when a crisis occurs, you have options to creatively modify your systems, to investigate damage, to control damage, and ultimately get the darned thing back without having to entirely redevelop your firmware.
One could argue that C++, C##, and so forth are getting into the weeds. Somebody is always tweeking software that is where the 'soft' creeps in.
(I guess I am an advocate for the other devil. Sorry Heater.)
This all started with The Tower of Babel in the Old Testament. Languages actually cause preferences of communication channels, and quite a bit of confusion, and require endless amounts of learning. Just try learning Mandarin Chinese as the modern world changes it.
Armies? Well, being widely accepted by a large user base always seems to be a winner, unless you have something you don't want everyone to understand. Then we get slang and jargon and its back to the Tower of Babel.
Could I get 3 of those $20 chips sent to Taiwan? I'll send you money for cost and shipping in advance and we can see if Customs refused to let me have them.
It does seem that each will require a $34.95 Schmart board to do anything intelligent. Please advise if we are buying this in quantify as well. After all, if the board is not available, what are we to do?
Oh snap!
Thanks,
David
I wonder the same thing. There are two approaches - solder paste and a toaster oven. Or buying some liquid flux, tinning the board, and then working the areas quadrant by quadrant (meaning the four side of the board). I've had some luck in the past with tiny SMDs of about 20 leads. But this would be a new adventure.
From past experience, if you load the board with the right amount of solder by tinning the fine pitch leads, the chip can have opposite corners tacked into place and then everything else is fairly easy. No more solder is required, just a very gentle touch with a nice clean low watt soldering iron. A flat tip might make the work go faster.
Lots of liquid flux that can be washed off with water. I'd paint the board and paint the chip leads with it with a small sable brush.
I already committed to buy 3 chips, but after thinking about this - I will buy 3 if that is what is needed to get to 10. But if others want to buy in, I might be wiser to buy 2 or 1. It isn't really the cost of the chip that the problem. It is that $34.95 for the Schmart Board. We can reach 10 and order, then late comers can buy chips away from me. It looks as if ordering Schmart Boards independently might be the reality.
I really can not imaging using 144 cores for something besides video or encryption. But both are interesting topics. 1.8 volts is adequate output to directly drive video, isn't it?
A clockless processor? Well, I am just going to have to hold one in my hands and see if I can absorb some knowledge that way.
Really, I'd like to see someone offer to solder the chip onto the Schmart Board for a fee so I don't have to do it myself. I'm sure I'll fry both the chip and the board if I try it.
BTW Flux is very cheap.
If one needs to build up some courage, some other lesser ambitious Schmart Boards might be purchased. I have some SMD SX28 chips that could use being mounted. Mistakes would be less costly, more easily recovered from. Only after practise,and a good nights sleep to sort out what I really learned would I do the 88 pin device. It is all about baby steps first, then walk.
ECN article about the chip and the board being paired up for hobbyists.
In other words, you can order here at $20 for the chip and $10 for the board OR you get both at Schmart Board for $34.95.
I am not sure how this all sorts out for shipping. But it might change everyone's plans. And read the brief. And there is a Schmart Board soldering video for this chip.
http://www.greenarraychips.com/home/documents/greg/GA144.htm
http://www.youtube.com/watch?v=lKahRk9-uB4&feature=youtu.be&noredirect=1
"The GA144 chip is designed to give you options that have never before existed and to place them under your control by writing software"
I believe Video requires about 1 volt or less to a VGA or NTSC. I would need to review some of the Propeller video documentation to be sure about exact levels. This chip does ADC and DAC, so video should not be a hardware issue.
The soldering is just a psychological barrier. A little practise and the fear goes away.
I haven't done SMT work for ages and then I was using professional tools and getting paid for it. I need to try it again with my new tools
I'm curious, If this beast is running without a clock how on earth are you going to get the accurate timing required for video?
Further, can these 144 cores actually pump data out of one "hole", as it were, at video rates?