|
Modernizing an old computer design is a weird thing. First you strip away all anachronisms and look what's left.
In the case of the original Elf from 1976 that was the processor itself. By pure chance, I got hands on a CDP1802B, which also already existed in 1976, but it was the faster version which used to cost more. It hums along at 6MHz without breaking a sweat, giving me almost four times the performance of the original Elf. I don't have any means on the breadboard to carefully monitor the processor's core voltage and temperature, but overclocking the processor is possible and I may yet squeeze considerably more out of very old silicon.
Toggle switches (or hex keypads) and LED address and data displays are gone, but there is still an interesting aspect about them. The original Elf had no ROM to boot it. Instead, the switches and displays acted as DMA devices and allowed to enter the program manually before letting the processor start executing anything. A modern Elf can have an all RAM configuration, like the original. Why not copy the contents from a ROM to RAM using the same DMA mechanism after a reset? It's still the same old Elf and no typing involved to get that thing running. Plus keeping the advantage of not having a ROM awkwardly floating around in your 64k memory map.
Now, we still need some I/O. For the time being that is a serial port and a VT52 or VT100 terminal (actually an old PC running TeraTerm). That was not part of the old Elf concept, because hooking up a 80$ computer to a terminal that costs at least ten times as much seemed a bit silly. However, that option was added later for those who somehow got their hands on a terminal. The cool thing was, that the Elf never required a UART for serial communication. Serial data was sent and received by the processor itself with carefully timed software routines at 300 baud. Well, Zwölf does it the same way, but I rewrote these routines a little so that they now send and receive at 19200 baud. 38400 baud are possible, but the little processor can't keep up yet when the terminal sends packets of data during a XMODEM upload. The problem is, that there is only very little time to actually do anything with the received data before the UART on the other side already sends the next byte. Still, even at 19200 baud I can load a program faster than with an old C64 floppy drive. Indeed, I'm comfortably in the mid range of the C64's floppy accelerators. Not so bad for a crutch that only has to be sufficient until I have something better.
Up to here we have mostly taken stuff away from the old Elf. Now let's finally add something. RAM. Lot's of it. In 1976 RAM chips were slow, had only a tiny capacity and cost more than their weight in gold. That's why the original Elf only had 256 bytes of RAM. Barely sufficient for the simple trainer it originally was, but it kept the price for your own computer under 100$.
With only a very modest amount of simple logic chips I can access fairly modern SRAMs with a capacity of 512 kb each and up to a total of 16 Mb. Of course the processor can't access so much memory without banking, but a little speciality of the processor (which is often seen as a disadvantage) allows me to do the bank switching without the processor noticing it. I can call code anywhere at any time without having to fear any problems. The code loaded to each memory page now is something similar to a DLL. And this opens up a lot of possibilities for a serious operating system, something also unheard of for the original Elf.
Turns out that modernizing the old ELF is more about taking things away than adding things and about the only thing that was not available in 1976 was fast enough RAM at an affordable price. I was looking for the schematics of the original Elf in archived issues of Popular Electronics magazine. The old magazines were full of ads for computers like the Altair. In one ad for the Imsai 8080, they proudly announced that this computer was expandable up to one megabyte of banked memory. For that, the Zwölf needs only two memory chips for about 11 bucks.
The Imsai needed 16 S-100 cards at only 3900 Dollars each, assembled and tested or a bit cheaper as kits. Plus a truly huge power supply. That's about 63000 Dollars, just for the RAM and the power supply, 306,240.90 Dollars today, adjusted for inflation. I don't know how fast they used to run their 8080, but the Zwölf can outperform the old monster and even run on batteries, despite being mostly made of components from the same time. And it even costs less than the original Elf without adjusting for inflation.
I'm really having fun here. 
I have lived with several Zen masters - all of them were cats.
His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
|
|
|
|
|
Tl;dr 
|
|
|
|
|
Awesome, you're really making progress.
I was looking at some old Popular Electronics last week and astounded at the price of memory, only the rich could afford it. (That definitely wasn't me)
I'm getting on with my Z80 stuff and really enjoying it.
Created a 24 key keypad board using cherry keys.
Cheated though using a 328P to scan and debounce.
Created a 7-segment display board...6 segment.
Also cheated using a MM7219.
Well into writing a monitor program for it.
Waiting on parts to create an EEPROM emulator, got the idea from the guy I bought the CPU board from.
Have a lot of plans.
The less you need, the more you have.
Even a blind squirrel gets a nut...occasionally.
JaxCoder.com
|
|
|
|
|
I remember an ad for a 64K S-100 bus memory board for $1495. I couldn't imagine needing that much RAM.
Crap, I use icons now bigger than that...
Software Zen: delete this;
|
|
|
|
|
I hear ya...and could you image surfing on a 1200 baud modem. Take 24 hours to load a page.
The less you need, the more you have.
Even a blind squirrel gets a nut...occasionally.
JaxCoder.com
|
|
|
|
|
Mike Hankey wrote: a 1200 baud modem
Back in 1991/2 that's what I had to use to connect between my PC in the London office, and the system in Louisville, CO. Made customer support quite a challenge.
|
|
|
|
|
I can imagine that was a challenge.
I tried to do the old BBS's with a 1200 and I remember that being enough of a challenge that I didn't do it long.
The less you need, the more you have.
Even a blind squirrel gets a nut...occasionally.
JaxCoder.com
|
|
|
|
|
I remember paying a fortune just to upgrade my 8k PET to 16k!
- I would love to change the world, but they won’t give me the source code.
|
|
|
|
|
I have a 'fat' TRS-80 Model 100[^], in that I spent $150 to expand it from the original 24K to a full, whopping 32K, which was the most you could put inside the machine. There were folks who made bank-switched external memory modules, but they were very expensive.
Software Zen: delete this;
|
|
|
|
|
I looked for a more specific topic, but didn't find one, so here we are... We went to git. OK, I got past the learning curve. It worked pretty good in VS 2019... until I updated VS 2019 and now there's no git functionality. I can't even find a reference to this problem. Git works. VS 2019 (Community and Pro) work, but VS doesn't talk to git anymore. I can't do compares. VS is not aware of what branch is checked out. I even pulled the project down by URL to VS. No luck. I'd really appreciate any suggestions or links to what might be going on. T'anks, Mike
|
|
|
|
|
Mine works just fine. Check "Options" -> "Source Control" -> Plugin selection - Git
Mircea
|
|
|
|
|
I got that already, but thanks.
|
|
|
|
|
Reading the Chris Maunder comment on the downward spiral of the industry, and along with the latest article by O'Reilly on low-code democratization of programming seems to never end. As with Chris, I have been doing programming since the first Radio Shack computers came out; and before that, using IBM punch cards to create wiring lists, paper tape readers to load programs, as well as cassette tapes.
I guess people are doomed to repeat bad things when they do not have a history of the industry. I can remember low code apps. being sold decades ago. All flops.
For a later lesson, take Microsoft's BizTalk. Preached as enabling power users and analysts to create programs without code. Like a company I worked for that spend 2 million dollars on a BizTalk project that failed because the gazillion objects it created simply choked the databases so bad that the throughput was like snail poop.
After hiring additional consultants and create hundreds of 'functoids', the project was scrapped.
For those not knowing what a functoid is, it's chunks of c# code used to do things that BizTalk could not accomplish in its so-called low-code IDE.
All this relates the infamous 'black-box' approach using some of Billy-Bob's code that is supposed to work. But what happens when you find it not working? Google to see if Handy-Andy's code will work?
Low-code, and especially no-code solutions of any magnitude, by default, are bloatware. As with BizTalk (huge bloatware), it usually chokes through-put.
|
|
|
|
|
But they have their use. Mostly to create consultancy jobs to fix their... products.
Also some of them are actually good, i.e. Simulink, especially in automotive systems paired with AutoSAR. Most of the code is manually written but the entire platform can be designed and configured with a human readable diagram.
GCS d--(d-) s-/++ a C++++ U+++ P- L+@ E-- W++ N+ o+ K- w+++ O? M-- V? PS+ PE- Y+ PGP t+ 5? X R+++ tv-- b+(+++) DI+++ D++ G e++ h--- r+++ y+++* Weapons extension: ma- k++ F+2 X
|
|
|
|
|
Your first sentence is absolutely correct. https://codeproject.global.ssl.fastly.net/script/Forums/Images/smiley_smile.gif
|
|
|
|
|
Member 14840496 wrote: BizTalk
BizTalk, in simple terms, is used for moving data around, transforming it, importing, exporting, etc. It's low-code in that there are lots of built-in objects. But yes, it is highly complex and I have used it successfully so my guess is you may have done something wrong.
Mendix is a no-code platform that I have used and it can do just about anything any other platform can do. Some of it is very fast, like building forms for example. Business Logic can be a little slower to develop compared to .Net. But I find it silly how many developers do not realize how powerful some of the low/no code platforms are. They will change this industry.
You won't build a google.com with one but they can do just about everything else.
|
|
|
|
|
Well I didn't build it. It was first contracted to an outside company, then outside consultants were brought in to try and get it working. They wanted me to join the team, but after looking at the project, I told them that is was doomed to fail.
I did, however, replace it with a C# program.
|
|
|
|
|
Member 14840496 wrote: It was first contracted to an outside company, then outside consultants were brought in to try and get it working.
It is a very complex product. We had to hire a consultant to come in and train us on how to use it and even then it was not easy.
|
|
|
|
|
Well, in the training class I had, the instructor couldn't explain why some of the labs didn't work. That's a bad indicator right there.
Months later I flew to Tampa because the BizTalk project they were building wasn't working. When we met in the room with the numerous programmers, they had literally printed out the workflow of the project and taped it to the walls. Note, I said 'walls', not wall. It went 1/3 of the first wall (about 6 feet), around the corner down the entire next wall (about 15 feet), and around the corner on the next wall for another 3 feet or so.
It was all I could do not to break into hysterics. All I could think was, are you people serious?
|
|
|
|
|
Member 14840496 wrote: in the training class I had, the instructor couldn't explain why some of the labs didn't work. That's a bad indicator right there.
 
|
|
|
|
|
I concur.
I've been using SSIS for the last ten years -- leaving the easy stuff to my colleagues while I implement the difficult stuff in C# "Script Tasks".
Before that, I was on a contract where they used an in-house "rule-based" system -- with a GUI to define which rules to use in what order, branching, etc. What if I need a new rule? Write it in VB.
It's not much different in SQL Server either -- I write a lot of CLR functions in C# to deal with the hard stuff.
Off-the-shelf tools contend only with low-hanging fruit.
If all you have is low-hanging fruit, then an off-the-shelf tool may be sufficient.
But no enterprise of significant complexity has only low-hanging fruit, so highly-skilled developers will still be required. But by all means let the highly-skilled developers concentrate on the difficult tasks while low-skilled developers work on the easy tasks in an off-the-shelf tool.
|
|
|
|
|
Your statement "What if I need a new rule?" is precisely the problem, and it happens all the time.
How do you write a new rule with a black box? Users are very rarely satisfied with the status quo.
I wish I had a dollar for all the times I heard "Can we add this?", "Can we change that?". In the real business world, it never ends.
|
|
|
|
|
PIEBALDconsult wrote: I write a lot of CLR functions in C#
Crikey in all my years of writing TSQL I never had to resort to the CLR, I guess I was only doing the simple stuff.
Never underestimate the power of human stupidity -
RAH
I'm old. I know stuff - JSOP
|
|
|
|
|
|
Mycroft Holmes wrote: PIEBALDconsult wrote: I write a lot of CLR functions in C#
Crikey in all my years of writing TSQL I never had to resort to the CLR, I guess I was only doing the simple stuff.
My thought exactly.
I've been writing SQL for over 30 years and TSQL for nearly 20 and, although I've played with some functions in CLR, I've never found a need for it in production code.
I suppose the over 600,000 lines of SQL under version control in the main project at work is also all simple stuff!
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
