ATARI ST FPGA announced

[DarkLord]

Lotharek started a new and very interesting thread over at AtariAge:

http://atariage.com/forums/topic/211194-atari-st-fpga/

http://lotharek.pl/product.php?pid=96

I don't recall that announcement here, if it is, apologies.

What's everyone think?

EDIT: Okay, I did a search and finally found some mention of Lotharek being
involved in the MIST message thread.

[alexh]

This is a general FPGA board with an integrated 68000 CPU. It will be able to run other cores such as the MiniMig, Amstrad, C64, Spectrum, Megadrive/Genesis.

[waSTE]

If it does turn out to be cheaper than Minimig as planned I think this would be an absolute bargain, even if it could only run ST / Amiga (though clearly the potential is there for more)

[calimero]

does it have some connection with MiST? http://harbaum.org/till/mist/index.shtml

or Suska http://www.experiment-s.de

...this would be third Atari in FPGA project... ??


EDIT: I just read: it is a MiST

[qq1975b]

It seems it will work as an Atari ST and an Amiga but I have some questions...

It can work as an Atari ST...but as an Atari STe? it seems now is not possible, I hope it will be before selling it.
I suppose it can work as an Amiga 500 only too.

So the question is...In this FPGA systems...are they upgredeable by firmware?

I mean:

Will it be able to work as a TT in the future with no new board design needed? I suppose Falcon is more complicated...
Will it be able to work as an A3000-ECS or A4000-AGA?
Will it be able to improve CPU speed to 68030 or more?

Or all these features need a new hardware design?

[wongck]

It seems it will work as an Atari ST and an Amiga but I have some questions...

Can run as an Amiga, eh?
So here is my chance to poke my nose into the Amiga scene....

[AnakiMana]

It seems it will work as an Atari ST and an Amiga but I have some questions...

Can run as an Amiga, eh?
So here is my chance to poke my nose into the Amiga scene....

I was thinking the same thing. I'm glad my dad brought home an Atari ST when I was a kid. But I later came to understand that the Amiga was worth getting into also. Other than seeing one ONCE at a friend's house as a teenager, I've never experienced the Amiga world. Now that I'm in my mid-30's I want to get a taste of it and satisfy my curiosity. I'll probably get a real one someday just because. But it'll never have that nostalgia effect of being in my heart like the Ataris.

[alexh]

In this FPGA systems...are they upgredeable by firmware?

The FPGA is re-configurable yes.

Will it be able to work as a TT in the future with no new board design needed?

No. It has a 68000 CPU and the TT has a 68030. I don't believe that there is an open-source 68030 HDL project yet. And it is unlikely that the FPGA on the MiST board would have the capacity for one anyhow.

I suppose Falcon is more complicated...

The FPGA won't have the capacity to contain either the 68030 or the DSP56K required for Falcon compatibility.

Will it be able to work as an A3000-ECS or A4000-AGA?

No and no. 68020 & AGA may be possible

Will it be able to improve CPU speed to 68030 or more?

No.

[qq1975b]

All clear now

thanks

[MasterOfGizmo]

Your reply is nearly perfect Thanks!

Will it be able to improve CPU speed to 68030 or more?

No.

He's asking for the speed of an 68030 and that seems to be possible with this kind of setup: http://retroramblings.net/?p=280

[MasterOfGizmo]

Your reply is nearly perfect Thanks!

Will it be able to improve CPU speed to 68030 or more?

No.

He's asking for the speed of an 68030 and that seems to be possible with this kind of setup: http://retroramblings.net/?p=280

The functionality of an 68030 is definitely out of scope, agreed. But most people were primarily interested in the speed boost the 030 gives, anyway. A super-charged 68000 on the other hand is something that probably even more prople are interested in as it's more compatible to a real ST than a 030.

[qq1975b]

Yes. The 68030 functionality would be interesting if the TT or Falcon could be implemented but not for the ST. With more speed only, in my opinion, is OK if we are talking about ST/STe and A500 only.

[MasterOfGizmo]

Yes. The 68030 functionality would be interesting if the TT or Falcon could be implemented but not for the ST. With more speed only, in my opinion, is OK if we are talking about ST/STe and A500 only.

You simply wouldn't want to pay for a FPGA able to hold an TT or Falcon. But FPGAs are getting cheaper and a mature ST/STE HDL implementation sure improves the chances for a TT/Falcon FPGA implementation in a few years when FPGAs have improved.

[qq1975b]

Yes. The 68030 functionality would be interesting if the TT or Falcon could be implemented but not for the ST. With more speed only, in my opinion, is OK if we are talking about ST/STe and A500 only.

You simply wouldn't want to pay for a FPGA able to hold an TT or Falcon. But FPGAs are getting cheaper and a mature ST/STE HDL implementation sure improves the chances for a TT/Falcon FPGA implementation in a few years when FPGAs have improved.

Yes, I know because you explained it me once that Falcon and TT are not possible with this design . I explained myself wrong. I was trying to say that full STe + A500 implementation + "Turbo" mode is absolutly for me because it can be faster than a usual ST and still quite compatible. 68030 is only interesting for a Falcon and TT system but MIST can't handle them (like any other similar FPGA system).

Are ST and A500 capabilities full implemented? "only" STe ones have to be done? Hope it doesn't take too long.

I preordered one MIST weeks ago too

[MasterOfGizmo]

Are ST and A500 capabilities full implemented? "only" STe ones have to be done? Hope it doesn't take too long.

They'll never be fully implemented. But they are both in a state where one can comfortable run most games and e.g. an audio session with cubase. I am pretty pleased with the current state. But there's always room for improvement and new features.

As for the STE: I haven't even started yet. Recently i did the MIDI stuff and some cleanup so the FPGA is now only used to 55% with the bare ST. Then i'll have also to cleanup the DMA a little bit so the blitter can be started. Next then is the blitter itself. Second are extended color modes and then some STE games may already run. But beforehand i'd like to get the turbo mode running to at least the equivalent of a 16Mhz 68000 which in turn is a precondition for MegaSTE.

Don't get me wrong, there are FPGAs capable to hold an entire Falcon ... or two ... or ten:
http://www.digikey.com/product-detail/e ... ND/4158677
That FPGA can hold more than 200 times more logic than the one on the MIST and yes, that's a five digit price tag for a single piece of silicon

[alexh]

He's asking for the speed of an 68030 and that seems to be possible with this kind of setup: http://retroramblings.net/?p=280

Interesting. I'd be curious to know how the improvements stack up on some other benchmarks.

On the Amiga you can make instructions single cycle instead of multicycle and you can adding write caching to get higher performance. But the ST games and demo's rely heavily on accurate instruction and memory timings. While you could activate such a turbo mode. I imagine it introduces terrible incompatibility?

[troed]

But the ST games and demo's rely heavily on accurate instruction and memory timings. While you could activate such a turbo mode. I imagine it introduces terrible incompatibility?

I'm under the impression it won't be demo compatible (and quite a few games then I assume) as it is?

viewtopic.php?f=101&t=24583&start=51

edit: Which is not to fault anything about its development - which is great. I'm well aware we abused the available hardware as much as we could back then and that it's for most practical reasons impossible to simulate correctly. Software emulators are full of specific patches instead of being cycle correct.

[MasterOfGizmo]

I'm under the impression it won't be demo compatible (and quite a few games then I assume) as it is?

viewtopic.php?f=101&t=24583&start=51

edit: Which is not to fault anything about its development - which is great. I'm well aware we abused the available hardware as much as we could back then and that it's for most practical reasons impossible to simulate correctly. Software emulators are full of specific patches instead of being cycle correct.

Correct. E.g. demos using overscan and e.g. opening borders don't work properly. There's no technical reason why it shouldn't be possible to make the shifter and CPU 100% compatible, but currently it simply isn't.