Firebee: the long game?

[BlankVector]

We have it now (2.02 version), I have received it from Didier. This is now in hand of Miro.

Huge step forward. Well done, OL.

[wongck]

We have it now (2.02 version), I have received it from Didier. This is now in hand of Miro.

Olivier

I know this is a FB thread... but is there also the CT60 version or not?
Would be great if there's some updates to the CT60 version as it's my goto Atari machine.

[mikro]

There is a few bits updated for the CT60 but mostly for the CTPCI (drivers). That's a huge step forward nevertheless because the last source code we had was 1.04a from 2009. However the source archive is complete so it is also a snapshot of FireTOS sources from June 2012 (which is surely out of date by now).

[wongck]

Well I look forward to it.
Looks like more just dropped on to your already full plate

[Cyprian]

What a great news.

[Rustynutt]

Sorry for your loss Mathias.
Family matters surely have an effect on community members, yet it isn't something usually discussed.

[Mathias]

Thanks for the kind words! Yes, it is strangely in general not discussed, even though I saw several people quitting any "Atari activity" due to such matters. Perhaps because we are scattered over the entire world, in most cases not knowing each other in person, and thus thinking that we anyways cannot be of great help. In every case I promised myselve to "not give up" and go on with the FireBee – even if it was much less the last time. Anyway it is nothing that needs to be spotted too much, it is only the fact that all our work is based on the fact that we do not have enough manpower for what we would like to achive – in every field of our plattform. So working towards a common goal is permanentely a kind of "administration of shortages", as all of us are working in their spare time and give their best.

[TheNameOfTheGame]

This may sound a little strange and I am not an expert on the matter, but seeing as many Atari projects have a lack of manpower to work on them, would it be possible to set up a Atari AI or use an existing AI and be able to train it on the various Atari platforms? Any AI experts here?

[ThorstenOtto]

Nice idea, but i think the problem would be the "training". If you ask ChatGPT, it will happily answer most of your questions, but most of them are simply wrong.

[TheNameOfTheGame]

Yes, training is the thing. I was thinking OpenAI could work but of course a training paradigm would have to be set and I think it costs some amount to train it to a constrained information set. Not sure if there is a open source AI out there. ChatGPT is very general and not ready for Atari specific problem solving from what I have seen. Maybe in 2-5 years?

Sorry, went off-topic. Just thinking ahead how this emerging technology could help projects like Firebee.

[mfro]

This may sound a little strange...

That's indeed the case. Even if that worked (I'm also pretty sure it doesn't), it most likely would kill most of the fun. At least for me.

What amount of fun could we expect from "AI-generated retro computing"?

[TheNameOfTheGame]

This may sound a little strange...

That's indeed the case. Even if that worked (I'm also pretty sure it doesn't), it most likely would kill most of the fun. At least for me.

What amount of fun could we expect from "AI-generated retro computing"?

Well, sure, I would think of an Atari AI as an "assistant" or such. For instance, AI is going to be very helpful in source code reconstruction in the near future. What if the source to FireTOS was not available and AI could reconstruct it from binary very quickly in C/assembler/etc.? That could be very helpful to Atari devs and take some of the tedious work off the plate and let them do the "fun" dev work.

[mikro]

AI could reconstruct it from binary very quickly in C/assembler/etc.?

Maybe I will live long enough to witness something like this but I assure you, this is decades from now. The best you can do now is to rely on Ghidra, IDA Pro and similar which put years of experience into their products for this one specific task.

But yes, I'm sure if you ask ChatGPT to RE some code, it will answer you with a high level of self-confidence that it is this and this C code. :D

[TheNameOfTheGame]

AI could reconstruct it from binary very quickly in C/assembler/etc.?

Maybe I will live long enough to witness something like this but I assure you, this is decades from now.

Certainly a few years, but decades? Nah. Anyway, I hope we're all around to see it .

[mikro]

People put a lot of unrealistic expectations into the AI models based on visual / textual demonstrations which have nothing to do with the problem at hand. Programming (and the job of software developers) is rare not because it is hard but because it requires following rules which must be fulfilled to the last letter. Even one single byte can have a disastrous consequences. So yes, I don't believe that AI RE will be here sooner than in 2040.

[TheNameOfTheGame]

If you have a binary and the assembler/compiler used to build it, it could be reconstructed through brute force techniques which could take a person a lonnng time but would be very fast for an AI if it learns as it goes. Break apart the routines in the binary, re-construct as C, re-compile and check, repeat if not right. Build up data structures as you go. Yes, that is the simplistic model, but the general idea with the devil always in the details.

It is the training that is the crux, but in the future general purpose AI will be able to communicate and then train itself based on conversation with humans. ChatGPT imo is just the baby steps to get the public used to AI, it is not general purpose AI which is not in the public sphere yet. I do tend to think farther ahead about what is coming which may seem unrealistic, but oh well lol. Personally I think this is very possible for an AI to do very efficiently.

Here we even have a human that does very well with it already - Thorsten .

[mikro]

For anyone interested, this is really a great and relatively easy to understand article how current AI models work: https://writings.stephenwolfram.com/202 ... es-it-work ... a bit of magic, yes but also simpler than you'd have thought. That's why I still think that domain-specific AI tasks of software engineers are still far, far away...

[Luposian]

What retro-teams fail to realize is that Atari, as it once was, is DEAD. No new software development. No new hardware development. All they ever can accomplish is trying to enliven a rotting corpse… and this has been done how many times now?

The ONLY way you will succeed at a “new Atari ST” is to have Atari create new hardware/software. But Atari is only interested in nostalgia… endless revisions of nostalgia. We really need to move on from reviving the past, AS the past, and talk to Atari about creating a NEW Atari computer platform and put some Atari COMPANY behind it.

[Eero Tamminen]

Everybody, give applause for acolyte Luposian, who's cast away his shyness and entered the public arena!

While the current incantation attempt is unworthy of comment, there's a chance he may, eventually, succeed at the necromancy of awaking the (scornful laughter of) long gone platforms, and the companies that once produced them...

----

EDIT: Luposian, I did not quite get your joke. It's like cream cake with a cabbage/onion filling, mismatch of things that do not make sense. Not sure whether my comment makes it any better though...

[Ragstaff]

Edit -nevermind

[Xyla]

The ONLY way you will succeed at a “new Atari ST” is to have Atari create new hardware/software. But Atari is only interested in nostalgia… endless revisions of nostalgia. We really need to move on from reviving the past, AS the past, and talk to Atari about creating a NEW Atari computer platform and put some Atari COMPANY behind it.

We don't need the imposters currently posing as Atari to make anything in order to keep the ST/TOS platform alive (or revive it, depending on how dead you consider it currently). If that were the case, PCs would all still be made by IBM. Besides, if they ever did grow up and get past their 2600 nostalgia obsession, I wouldn't trust them to make anything good anyway.

It seems like we have all the pieces to make a "new Atari ST", but the various projects haven't reached the level of success needed. Firebee is still unfinished, and I have had enough arguments with MiST users that don't have any interest in Atari-compatible cores that go beyond the STE in cabability to know there's not much chance there. I'll keep hoping someone makes the ultimate Atari-compatible computer, but the last people I would expect that from is the people making the exciting new 2600+ console (A console that plays both 2600 and 7800 cartridges??? It's never been done before!!! LOL).

[mikro]

...or we can just have fun. Producing whatever whoever likes and see what happens. It's really getting tiresome to read again and again what has to be done by people who can nothing get done.

[joska]

What mikro said.

[ragnar76]

...or we can just have fun. Producing whatever whoever likes and see what happens. It's really getting tiresome to read again and again what has to be done by people who can nothing get done.

Do what you want to do. As you know, the schematics, the pcb layouts, the BOM and so on are OpenSource and available at the firebee webpage (Section Hardware: http://firebee.org/fb-bin/support?&lng=EN). The KiCAD files are made with KiCAD 5 but they open with KiCAD 7 too without any problems (only some conversions between used V5 and newer V7)

[Gamer]

The Firebee, right now, is the second most powerful Atari computer out there with the Atari VCS being the most powerful. However, as a user, I believe the Firebee needs some things. It needs a more powerful graphics card, updated software to access the Internet and a program that will allow it to run almost all ST software. Of course an updated emulator to run 8-bit software would be great also. I was always hoping that it would be able to run Calamus and that never happened. I love this computer and am glad I own one.