Welcome everyone, time for the Gigatron Hack Chat. I'm Dan Maloney, I'll be the moderator today. I want to welcome Walter Belgers to the chat today to talk all about the Gigatron.
Walter, thanks for coming on in what must be a difficult time. We were all so sad to hear about Marcel, and you have our condolences.
Thank you Dan. Yes, his passing is very sad indeed.
How did you two end up working on Gigatron together?
I have known Marcel for over 20 years already
What is on the menu for tonight?
He was working in the semiconductor industry and had lived for some years in Taiwan but returned to the Netherlands again
He was always doing interesting side projects, such as with computer chess, making weather balloons to reach space and more.
After he had bought a bunch of TTL chips and a book about what you could do with them, he thought about starting an interesting electronics project, but he wanted to do it together with somebody.
Actually, another friend that also was involved on some of the other projects was involved in the beginning, but he no longer had the time, and Marcel asked me. I have had an interest in electronics all my life (having soldered since age 8).
@Dan Maloney . Thanks
And to be honest: Marcel was a really knowledgeable and enthousiastic person, and I thought that whatever the outcome would be, I would enjoy "finding out stuff" with Marcel.
When he asked me, he and the other friend had already started, so I was not involved in the earliest prototyping work. My contributions started with the kit version of the Gigatron.
That's the basis for some of the best projects, and partnerships.
Yes I really enjoyed working together, and our knowledge did overlap on some points, but on others we were really able to push each other.
Unfortunately programming is not my biggest hobby, so Marcel did all of that (and I could never have done it as well as he did).
Were you surprised by the community that sprung up around the Gigatron?
Surprised is not the word.. we were thankful and overjoyed :)
How many kits have been shipped?
We had some discussions very early on with other kit builders. Building a kit is a whole different game than building a one-off. And I remember one of the things we were told is that you should do this because you get to interact with all sorts of nice and interesting people.
And when that happened, we were really happy. As I am now seeing many of them in this chat :-)
@Steve Silverman we sold just over 1000 kits
Walter may I say sorry for the great loss - Marcel was always willing to help anyone with their Gigatron ideas or issues
What will happen with some of the work Marcel had started, such as the I/O expansion? Has that been open-sourced too?
@monsonite yes Marcel always wanted to understand every little detail himself, and if others were interested to learn things, he'd be happy to help. As am I. You could say we are old-skool hackers
@pbierhoff the I/O expansion was work in progress. all the software and the schematics can be found on the forum. it's needs a bit of work to be really useful, but it would be great if the Gigatron were to be able to boot games from SDCARD in the future. It's not really in-line with the "retro" design goal, but makes life much easier
Who came up with the name Gigatron and why?
@Paul Hernaus the both of us wrote down possible names, and we gave points to all of the names. The name Gigatron was the one that got most points..
We had 73 candidate names :)
42 would have been suffiicient ;)
Gigatron is a really good name for this device!
What has been your biggest "lesson learned" from working on the Gigatron the last two years? Both positive and negative?
You might not realise, but every tiny part about the Gigatron was thought through very well. The name as well, what does it say, how does is feel, can it be googled, etc
@pbierhoff I can't speak for Marcel, but for me personally, the biggest lesson is that you do not need to know in advance where your project will end up. Sure, there are goals, but instead of trying to reach that goal, it's better to keep taking a step back looking at what you're doing and where you're going and adjusting where needed. That might take you to a better goal in the end
There have been almost no negatives. There were only two people who have been harassing the project. Ah yes and every project has a few set-backs but I would not call that a negative.
I think creating such a design in fewer than 1000 simple TTL gates and showing that it can run 6502 code is pure genius - and illustrates just what computers can do - once you understand them fully
We weren't planning on doing this :) The original goal was to play tic-tac-toe on a 9x9 LED matrix :)
@monsonite that reminds me of the compilers course I took in college - where I learnt that you can write a really simple compiler with a very dirty implementation, and then use that compiler to write a much nicer compiler of itself
I think what really helped the design is that Marcel had not designed a computer before, and because he researched everything to the fullest, he was able to come up with things that others would not think of, because they have ready-made solutions in their heads
I think Marcel took that same lesson and applied it here ... from 8 instructions to a full 6502
that's quite a leap!
@Walter : Were there any community contributions that got both of you thinking "Wow! I did not know this was possible?"
@pbierhoff we thought the GIgatron was not possible :-) But joking aside, there have been quite a few "wow" moments, starting with emulators, then hardware additions, to the recent @at67 basic compiler
I think Marcel proved that you could take the simplest of hardware - and with layers of clever software so much could be achieved. This should be a lesson for every Computer Science student - we don't always need 10 billion transistors and gigabytes of memory to achieve great things
@monsonite True. Some Gigatrons have found their way to high schools and universities, which is nice. I think people can learn a lot. But in "the real world" alsmost everything consists of layer upon layer upon layer where you only see the top layer that you are talking to. Such as the fact that the software on your cellphone still sends hayes AT modem commands to some chip in your phone
(and that layer is then abstracted)
And all the layers that "the cloud" implements are another example
in the end they are just computers sending bits across network pipes
Yes, the world is getting complex. The Gigatron is also complex, especially the intricasies of the software, but you can spend some time and understand it all to the transistor level.
@Walter What do you think will happen now that you "abandoned" the project and open sourced everything. Will Gigatron die down or will it flourish for years to come?
@Paul Hernaus Unfortunately, I cannot see into the future. But I know that in, maybe, three weeks time, new Gigatron kits will be sold by a Dutch company (albeit without the wooden box, manual and controller). @at67 is still putting a lot of time into the BASIC compiler and has shown that writing cool games is now perfectly doable, so I hope that will spin off some cool games
I wonder if a simple tape interface is possible with the existing hardware. In 1984 I got two Sinclair machines to transfer code from one to the other wirelessly across 30m using an FM audio bug
On the forum, people are still discussing possible hardware upgrades, so there's enough going on.
@monsonite the existing I/O capabilities should make that possible I would presume. Requires some software though :)
for transmitting data over FM, you don't really need anything, as the RF radiation from the Gigatron is enormous ;-)
(receiving is a different matter :)
What is the coolest feedback (in your own personal opinion) you received regarding the Gigatron?
So don't use a Gigatron for logging in a ham radio shack!
and from whom?
@pbierhoff there's one very special kit that went to somebody who was a great inspiration to us, and he send some really cool feedback. but he asked not to be named
Marcel wrote me a special version of the ROM - so I could run the cpu at 12.5MHz and still get VGA video output - he hinted that he had pushed it to 15MHz. Almost all of the ICs were upgraded to 74F series, but even with HCT a stock board will run at 10MHz
But we had countless mails from people, and every one of them put a smile on our faces.
Did the feedback alter the final design in a significant way?
We got plenty of pictures as well, of people building the kit, and that is wonderful to see!
Are you still in touch with people like EEVBlog and 8-Bit Guy who provided early visibility on the Gigatron? How hard was it to get them to review your kit?
@steve silverman no the feedback was not on that detail level
@pbierhoff I haven't been in touch much with Dave (EEBlog) but he did some great work. I had to laugh when he opened the box, threw away the manual and just started soldering! But he put everything in the right place and it worked right away.
That was epic indeed - a 4 hour soldering and talking session
I have also been in contact with the 8-bit Guy. He's also busy with designing a system, which is interesting to watch.
Thanks to Dave we had a discussion whether we should switch to 4-layer boards but in the end decided against it. A few were made and a few people had an interest (e.g. when doing overclocking it's a must) but not many.
I liked when Dave (EEBlog) started cutting off all the decoupling capacitors - as an exercise in why they are important
Dave did some really nice tests, the cutting of the capacitors and the move to a 4-layer board, which were both informative sessions.
In many ways the Gigatron is what the Raspberry Pi set out to be in the beginning - a platform for educators to show how computers work
Have you had any interest from Eben Upton and/or others in the Raspberry Pi Foundation?
Maybe a punch-card interface? Not so much because it is retro, but because it makes it obvious that a any permanent memory just pushes bytes in one by one.
Nope. But the Gigatron and RPi are of course completely different.
@Andy Pugh maybe a core memory board that you can put in place of the SRAM :)
Or a nice high speed paper tape reader!
hm still have a punch card reader/writer and paper tape reader/writer here in my home :-)
I wonder if there is a way to use LEDs as the storage elements in core memory?
At least LEDs would non destructive readout
Core memory magnetises metal rings.
I think the best thing was Marcel's tenacity - and his courage in his convictions - regarding the ISA. To go from a set of empty breadboards to a marketable kit with software in 12 months is an amazing feat of achievement.
An LED cannot store information, only display it
I know how core works, but it isn't easy to _watch_ it work.
@Andy Pugh ok so Core with LEDs added to each bit? I see a new hackaday project :-)
And some of the serially-addressed LEDs may have more CPU in them than the Gigatron :-)
A matrix of LEDS with photo-resistors
@monsonite: well we didn't know beforehand it would take so much time. The breadboard design was finished pretty fast (~3 months). But the rest of it took quite some time. The first year we spend about 3 days per week per person on it (next to our daytime jobs).
12 months from conception to shipping kits IS remarkable
I am looking at some new serial non-volatile 1Mbit RAMs from Microchip - I think they might make a useful upgrade to "Pluggy"
The lack of memory in Pluggy was something we noted too late.. it should have had more but it was too much work to change it at that stage.
what's the maximum address space? (16 bit address bus?)
Then again I'm not sure if people will want to write large BASIC programs to store in Pluggy. An arduino hookup would make more sense.
Serial NV RAM is perfect as a backup store or games cartridge. On power-fail it automatically writes the entire RAM array to EEROM
You cat put 64kB in, or you can do bank switching.
Yes - when I say "Pluggy" - I mean any Arduino based support device
@monsonite if we could create something that can fill/save the RAM that can be pre-loaded once, that would be great to store a game library in. that would make the Gigatron appealing to a new group of users I suppose.
As the ultimate tribute to Marcel, someone should send a Gigatron to space in a helium balloon or as the guidance computer on a home-made rocket :)
hear hear :-)
One of our stretch goals is to get a Gigatron into a Quentin Tarantino movie
Well, I have a Gigatron on order, and I've always wanted to try high-altitude ballooning...
(For those of you that don't know, Marcel was part of a team that sent a helium balloon into space about 5-6 years ago. Marcel did the recording part)
A more general question: why do you think such a (wonderful) back-to-basics computer is so appealing to so many people?
@Paul Hernaus Do you think the LED + PR thing could work like CRT memory (Another "watchable" memory tyoe from the distant past.)