Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

[PeterS]

I really like my Milan, it's a pity it wasn't more popular.

I have the 68060 upgrade but you can never have too much cpu power. A Coldfire upgrade would have been nice but if an Apollo 68080 plugin cpu was available offering a good speed increase I would be interested.

I remember hearing rumours of a Motorola 68080 years ago but began to think I had imagined it.

[joska]

@joska: Please tell your opinion to the Apollo team. This is the way to get them to the descicion to make the PMMU. They don't follow this discussion here.

I have finally succeeded in logging in to their forum so I will do that. But if they're serious about the Atari version they should get involved in the Atari community too.

[calimero]

As far as I understand, they cluck relation with many Atari developers in past few years...

[Faucon2001]

OL (MyAES), Vincent Rivière and Rodolphe Czuba are in contact with them, at least from what I have seen.

[calimero]

Rodolphe Czuba? ...maybe I miss something...?

[Faucon2001]

Yes, he commented it on Atari French forum Yaronet, seeing it as an opportunity to develop new hardware.
No announcement, but a least a strong interest.

[Faucon2001]

http://www.yaronet.com/topics/186541-am ... re#post-29
Source of what I am saying, but you need to speak French sorry.

[1st1]

It's good to see that you get behind the thing by yourself. See also last entry here: http://atari-forum.com/viewtopic.php?f=97&t=30339

But please let them work on the thing first, they are on a good way, you will love what they do.

It' like I wrote, while you are talking about your worries others are doing it.

[calimero]

@faucon2001 thanks for link! Using google translate (since my french from school is long time forgotten):
"
Rodolphe Posted on 02.09.2016 at 12:12
Well, I am obliged to speak:
I am in contact with several developers with the author of the core Apollo 080 so that atari can use their card!
This is under negotiation and technical update because the author does not know much atari but is willing to become compatible.
We are also seeing to integrate the video STE + Falcon in the chip.
So we can make updates of all the atari of the ST to the falcon!
And also make a new machine one day .... finally ....

I'm taking care of it ! smile
http://www.powerphenix.com"


Czuba did talk some tine ago that he want to make new machine, I think last time it was for SillyVenture promo video...?

edit: https://m.youtube.com/watch?v=Hq-Eesu7YG8

[1st1]

That video is old project idea as it is based on difficult and expensive to get 68060. As noted here, they are working on 68080 now. And now you can add 1+1 if you know which parts of the Atari hardware already have been designed in VHDL. See MiST, see Suska, see Firebee, see SuperVidel. Add all together, rework and optimize and you have something very powerfull. More powerfill as Firebee ever could be(e).

Anyhow, Apollo team still needs to be pushed towards making the PMMU. Of if experienced VHDL and PMMU knowhow exists here, join them to make it.

[oehansen]

A m68k, in silicon, at a speed grade equivalent with arm/intel? ... it would run circles around intel.

[1st1]

I don't think it will reach speeds of current Core-I or even Atom processors. But if you compare the whole environement / ecosystem, having 3 Ghz multiplce core processors and operating systems whicn already reserve a gigabyte or more memory just for themselves, tons of backround processes on the one side, and then having an 68080 with 100, 200 or maybe even 1 GHt and small os (MiNT, Magic, Amiga OS) on the other side, this could be interesting.

[ex68k]

A m68k, in silicon, at a speed grade equivalent with arm/intel? ... it would run circles around intel.

You would need really big & deep pockets, to even dream about competing with the newer designs.
And a lot of time, and even a bigger team.

[ex68k]

A m68k, in silicon, at a speed grade equivalent with arm/intel? ... it would run circles around intel.

You would need really big & deep pockets, to even dream about competing with the newer designs.
And a lot of time, and even a bigger team.

Sorry, hit the "send" button to early

So the idea of using a decent/recent FPGA for the core, updating it with the recent video standards & audio,
is definitely the way to go. And the FPGA can be updated every few years to squeeze more out of it ...

[mfro]

A m68k, in silicon, at a speed grade equivalent with arm/intel? ... it would run circles around intel.

No idea what brought you to this insight, but I seriously doubt it.

MIcroprocessor design made some huge quantum leaps since Motorola left the stage, first steadily following Moore's law, then making the circuits themselves smarter (pipelining, going superscalar).

With sober consideration, Motorola pretty much failed at both with (not only) m68k. Sad, but true.

[1st1]

Like mfro said, orginal Motorola 680x0 CPU is ways behind of modern processor design. It's not only clock speed what makes today's processors fast. Apollo 68080 has now a lot of these modern optimations, like pipelining, branch prediction, out of order execution, maybe also register renaming, parallel execution (68080 can execute some instructions two at one clock) , "multimedia" instructions (SIMD, single instruction, multiple data, see MMX, SSE, 3DNow, ... - and Apollo has AMMX), like Intel and AMD and Arm have them. Apollo at arround 90 MHz can play MPEG files better than 600..650 Mhz PPC (PowerPC processor) made by Motorola (!), Apple and IBM after 680x0 series had been stopped. Thats very promising at about 100 or later 200 Mhz and software which was designed for just 33 Mhz or less.

[alexh]

The problem with Apollo having modern optimisations is that this area is very heavily patented. Broad idea patents (which IMO should never have been granted) which cover practically every conceivable implementation of these ideas.

If Apollo team ever attempted to do anything other than the small retrogaming scene it is my strong belief that if any of their optimisations could be proved to infringe existing patents that the other microprocessor IP developers would slam them hard.

That said there are a lot of microprocessor techniques discovered after the 68060 was designed which were either not patented (there are ideological people in the microprocessor world who publish rather than patent) or the patents have expired (they don't last forever) which Apollo could freely use.

[alexh]

Just had a meeting with the Toshiba lawyers about viral OS (open source). During the Q&A I used the example of a GPL3 licenced HDL design (MiniMig) and if Toshiba could make an an ASIC (or FPGA) with an (unmodified) GPL3 HDL project with our CPU design bolted to it. The lawyers said that would be a "derivative design" and due to the GPL3 licence it would be illegal to sell or distribute without releasing the source for our CPU. I then brought up the concept of providing a precomputed "CPU library" with respect to FPGA development. The lawyers considered it for a bit and we discussed the various technical challenges and decided if we shipped encrypted HDL for our CPU for the end user to build into their own FPGA with the open source GPL3 HDL sources (and never provide a complete FPGA) would avoid the legal minefield.

[calimero]

@alexh - valuable information, first hand!

[kolla]

It's not first hand, Toshiba does not represent GPL, FSF does.

[alexh]

Absolutely but Toshiba's lawyers have much more experience than anyone speculating here. Including myself.

One thing I forgot to mention was they suggested the best solution (for Toshiba) when using open source was to contact the OS HDL author for a an alternative (commercial) licence.

Another interesting point made in the meeting was the lawyers referred to OS licences such as BSD and MIT as non viral, meaning that use in a project (unmodified) combined with our CPU we did not have to disclose our CPU source.

[kolla]

Correct, that is the way to go - that is how we have macOS, for example. Dual-licensing (GPLv2 + BSD) is quite common.

[kolla]

However - Minimig is GPLv3, and hence it is up to FSF to decide what they allow Personally, I don't grasp the point of keeping a 68k soft core aimed at the retro community closed source, really.

[alexh]

Minimig is GPLv3, and hence it is up to FSF to decide what they allow

Surely you mean "it WAS up to FSF to decide what they allow"? No?

FSF decided what was allowed in 2007 when they published GPLv3.

[Sorgelig]

A lot of low-power devices use special macro-commands or HW modules to playback audio and video. Apollo didn't invent anything new here. But still, it's useless. There is no market for their CPU today. It's fully occupied by modern devices.