The above is what I have been into lately and my excuse for light blogging. It is a little microcontroller board based on the NXP LPC1114. Full board details, ordering information, and documentation are available at LPC1114 Devl.
This all came about because there was a project I wanted to do. But I needed some things. So I started designing. The first few designs were failures in various ways. So to keep development expenses down I did the design in sections. That worked. It also meant that a user could add bits (if needed) to get exactly what (s)he wanted.
I suppose I should add a little bit about language support. My friend Clyde (a genius in assembly language programming and his long distance friend Vic) worked many long and hard hours to get the Forth Language implemented on the processor. They did it because I wanted it. And they wanted it. Forth is probably the best language ever invented (so far) for real time programming. It is much easier to use than C and encourages programming in well tested fragments (the way it is supposed to be done). Clyde has been a friend of mine for about 37 or 39 years. Back to the days of the CACHE Club (The Chicago Area Computer Hobbyist Exchange). One of the very earliest computer clubs. We used to get reports on what a couple of hackers in California were up to from other computer clubs. Long before they became household names. Wozniak and Jobs. You might have heard of them by now. We also learned before any one else about another hacker whose name is now also a household word. Bill Gates. Clyde tuned me on to Forth. But I digress.
The Wiki gives the very basics of the Forth Language. It is said of that language that it will make a good programmer ten times better and a bad programmer ten times worse. I don’t know about the worse part. I do know that when I’m competing with C programmers on a project I can run rings around them both in processor economy and speed in getting the job done. Forth encourages good factoring of problems. And good factoring is the key to processor efficiency.
Now about software efficiency. Up ’til now no one cared about software efficiency because a program that barely runs this year will be 4 times faster in two years. But Moore’s Law is running out of steam. As documented here:
As Moore’s Law slows, open hardware rises
28nm node at the final frontier of Moore’s Law
And this short YouTube video that actually asks software developers to improve their product.
I think Forth can do that job. At the very least in the Real Time (hardware control) realm. And possibly others. It is my ambition to Kill C bloatware (nothing ever kills a computer language) for Real Time development. To that end I have a few more projects on the way. I’m working with one of our readers on one. And a noted science fiction author on another. I will announce them here when they get completed.
And if any one is interested in a custom design let me know. I have a crack team that can do amazing things. We would be very happy to port our Forth to your processor. Or design a motor controller for you. Or? And to Eric whose desire for a tool lead to this whole adventure. I have one board to test and another to complete before your design is done. But as you can see I had a few detours along the way. Some things to do. Some tools to make. I probably should also mention that Eric and a reader gave me some seed money to get started. They have been my angels.
And a note: our Forth has been tested on an LPC1114 Xpresso board. If you have one of those you can try it out. It should work on the LPC1115 Xpresso – but we have not tested it.
Comments
15 responses to “Just In Case You Were Wondering”
Hey thanks! Glad to help, even if just a little. (Right now, I am an inch away from sticking my PCB in the oven!)
This brings what to the party? Just a quick search returns Forth compilers for Arduino, PIC and Propeller. Probably lots more. And there are dozens, maybe hundreds of small micro boards out there.
Just to put Moore’s law in perspective, I’ve seen the computer from a Minute Man missile. Two large PCBs, one was an entirely discreet (actual transistors, no ICs) CPU and the other was 500 bytes of core. This could fly to Moscow. These days some lame piece of bloated uselessware takes up gigabytes. I’m all in favor of forcing code monkeys to write efficient code.
MMM,
This brings nothing to the party. Except that it is the start of a system. A system designed for developing products.
Watch for follow on boards.
http://GreenArrayChips.com and
http://greenarrays.blogspot.com
A whole processor, RAM, boot bios and bus; running FORTH; on ONE CHIP 3mm x 3mm!
I’ve been privileged to be on their email list for 2 years now… they’re Fragrant for FORTH
Karridine,
They sent me one of their development boards. Greg Bailey is a friend.
MMM,
The great thing about code monkeys is that they work cheap. If you want non-bloatware, you’ve got to a) find somebody who can write tight code and b) pay him/her to do it. Since everybody else is also trying to find Ms. TightCode and hire her, the paying part can add up.
Some of the projects I’ve done lately have used two processors. One does the critical tasks and the other one is a feature-filled firewalled sandbox for the marketing and software departments to play in, where they can’t hurt anybody.
As for Moore’s Law, the machine that makes it happen, the lithographic stepper was invented across the street from my house and I interviewed at ASML, the successor company to Perkin Elmer last year. More than likely the next decreases in part size are going to cost so much that it’s prohibitive. If you could see what goes into the machines now to get the current densities you wouldn’t believe the workarounds and very exotic technologies. Makes rocket engineering look simple.
-The real limit to Moore’s law is when feature size shrinks to the point that quantum effects like tunneling surpass the desired action. Which will happen if gate oxide layers get much thinner.
-I’ve seen companies try to do engineering with stupid people. Any individual dullard costs less, but 5 of them don’t, and they still do a crappy job.
MMM,
-I’ve seen companies try to do engineering with stupid people. Any individual dullard costs less, but 5 of them don’t, and they still do a crappy job.
That was when I got called in.
Yes, there are a lot of free Forths. However, they do not do the same job as a professional cross compiler and target code. You can now evaluate and use an MPE compiler free for non-commercial use. I haven’t written an ARM/Cortex interrupt routine in assembler for the last ten years using one of these.
Have a look at
http://www.mpeforth.com/xc7.htm
Stephen is quite right.
The free Forth on this tool is not a professional development system. It is a learning tool.
I should note that Stephen was an early consultant on this project. 2012 IIRC.
As a lifelong code monkey, I will tell you why we have bloatware: because users demand cheap, feature-rich applications and massive modularization is the only way it can happen.
Anyways Moore’s law has been about to run out for decades now. I wouldn’t bet against it. That said, performance is far more algorithm-dependent than hardware-dependent.
Funny, the other day we had to disintermediate a gigantic SOAP call from our ERP application to a webservice that we needed to do nothing but return a distance between two zip codes. Just thousands and thousands of method calls, the whole shebang took 250ms to process and that was too long because the process is realtime and we need hundreds of them at once. The Web guys threw up their hands and decided to connect directly. Sometimes performance matters.
More than likely the next decreases in part size are going to cost so much that it’s prohibitive. If you could see what goes into the machines now to get the current densities you wouldn’t believe the workarounds and very exotic technologies.
Fabrication plants are becoming a larger investment, but I doubt they will become prohibitive for a long time yet.
http://en.wikipedia.org/wiki/List_of_semiconductor_fabrication_plants