FPGA Replay Board

[ShapeShifter]

MikeJ will be able to answer in more depth, I'm sure, but my understanding is that the output modes selectable, is something which is entirely determined by the core itself, and then any drivers written in AmigaOS to support what the core supports.

The Replay Board itself is capable of outputting just about anything: it is equipped with a DVI socket on the back capable of outputting both digital and analog signal outputs.  The rest is entirely up to the core.  As for precisely what Minimig AGA supports screenmodes-wise, that's something that only MikeJ or Jakub can answer right now.  As for custom graphics card support, well a lot would depend on what exactly you have in mind.  But my understanding is that if what you want is Picasso-style graphics card support, then I'm pretty sure that support can and will in due time be implemented in FPGA as well.  There's no need for a physical card via the expansion slot, just as there's no need for physical AGA chips to gain AGA compatibility, it can all be done in FPGA.

On the question of Chip RAM, I know it is possible to expand Chip RAM from the standard AGA 2MB up to a massive 50MB(!). There is a Youtube video from last year which which shows this in action; the additional Chip RAM can be activated by executing an 'xchip' command in the AmigaShell.  Chip RAM availability then shoots up from 2MB to 50MB: http://www.youtube.com/watch?v=vn4ZzLH6MpE (it's at around the 1:10 mark.)

[ShapeShifter]

Don't need a video card, Minimig AGA has one built in.

http://www.amiga.org/forums/showthread.php?p=648912

This is what I trying to get across when I said the available output modes on the Replay are controlled in software, by the soft core (the 'hardware') and by OS drivers.  The FPGA can "be" anything we want it to be - it can be a CPU, ECS, AGA - and a graphics card too.  We don't need a physical graphics card to get RTG modes, that's the beauty of this system. The FPGA is like a blank canvas.  It can be anything we'd like to create in silicon!

The only limitation is that someone has to write the core code & drivers to support the particular screen modes we wish to display - and the FPGA needs enough spare capacity to output the mode we're looking to support.  Yacubed estimated the Replay could support modes upto 1,280 x 1,024 - but I am guessing that's in full 24-bit mode, and that higher resolutions would be possible with less colours.

The link above to the Minimig AGA 060 topic explains in more detail, but to sum it up, the only reason that FPGA Replay does not presently support Picasso96 RTG screen modes is because the T68K soft core is not yet 100% compatible in '020 mode, and so the P96 software won't yet run.  But with a 68060 CPU installed via an expansion card, the Picasso 96 *will* run, and thus can support RTG modes people, provided a P96 driver is written for them.

Yacubed is working away at implementing this now, and the aim is to have the P96 software fully running and supporting RTG modes on an unexpanded Replay board in the nearish future.  MikeJ is currently doing some work on different output modes too.

[ShapeShifter]

CPU : 68020 (@ 68030/50 Mhz speed)

At the moment, this is correct, but it is subject to change as they improve the T68K core.  I've been exchanging some interesting e-mails with Yacubed on this subject in order to get a better grip on where things are going.  Here's a little snippet from his email (slightly altered for clarity).  I'm sure he won't mind me sharing with a wider audience

SS: I think for most people, though, what matters is not really what the specific model of CPU you implement -- but getting the maximum possible performance and speed ... Maximum compatibility and maximum speed.

Y: "That's the beauty of FPGA chips - you can create circuits which never existed before. The Tobias Gubener's TG68 can be configured on the fly between 68000/010/020 and EC020 modes. The performance calculated as instructions per clock is already twice as fast as an 030, and we can use hardware multipliers of the FPGA to improve multiplication and shift/rotate instructions. Then it will outperform an 060 clocked at the same rate. The compatibility of [a real] 68060 is limited (when compared with its predecessor) in exchange for it's higher performance. Our softcore CPU doesn't have these drawbacks (although we need some time to fix the last few problems but fortunately there are not many of them).

I guess the specs are more than enough to install Workbench 3.1 on 'HDD' (SDCARD) and then copy a bunch of games and use WHDLOAD to play them

It should just about do, yes...

[ShapeShifter]

Have you decided yet whether you'll be doing another run of boards?

I sure hope so. The Replay is a wonderful little system with so much potential.  Mike has invested a huge amount of time, effort and energy into it's construction, and it would be fantastic to see it truly take off.  

It would be nice if the open source community could embrace this system too, and find new and exciting ways of utilising the FPGA technology.  What about creating a new platform which combines 68K and X86 technology? Could open source OS's like AROS use a hybrid system in some way to develop a multi-architecture OS?

Okay, I'm just imagining possibilities and engaging in wild speculation here, but it's because what I love about FPGA technology is that it's theoretical potential is as vast as the imagination of the people using it.

Personally, I'm very much looking forward to using AmigaOS on this thing, but will have fun playing around with other systems/cores which are released for it over time.  But as well as bringing older technology snack to life, a part of me is very much hoping that boards like this will lead to entirely new creative projects being undertaken

[ShapeShifter]

Can't wait for one of these to be nestling comfortably at home.

Likewise!  Theres a lonely, empty space underneath my living room TV,  just itching to be filled by an object approximately 17cm by 10cm in size - give or take a few CM's.

This empty space just doesn't look right.  I keep looking at it and thinking 'this must be filled somehow!'

I do hope I can have the situation rectified by 25 December 2011 :lol:

[ShapeShifter]

I would like to echo AmigaDave's comments and say that I think, actually, it is quite incredible what MikeJ has been able to achieve so far.  For one person to personally design, and then build, refine, source a supplier for components,  arrange a business deal to get it produced in quantity, arrange funding to pay for it all, and then get a production run of 50 units produced, and then  personally thoroughly test each one, and then ship them all out himself without any support from any other person...is quite incredible!! Especially if, as I believe, MikeJ is holding down a full time job at the same time.

I realise it is very frustrating that we can't all have our new wonder-Amigas right now, and the frustration you're expressing it probably made worse by all the false promises of the past by other vendors - promised machines which never materialised - but this isn't MikeJ's fault.  He's in many ways done a lot more than any single person alone would ordinarily be able to achieve.  I for one am very impressed by his work so far!

I am sure he would like to order hundreds of these boards and ship them all out, but I suspect he probably wants to see how gets on with 50 for now.  There's only so much a single person can handle at a time.  It's a lot of work, and I'm sure what he doesn't want to do is to end up ordering 1,000 of these boards, and then find he can't test so many units all by himself, and has tp ship them out as-is, and find that many have faults which cost a lot to fix (for example.) Or to find he can't sell all of them and would make a massive loss on this project.

The responsible thing to do, in my opinion, is exactly what he has done.  That is: to produce a limited run of boards to begin with, so that he can then personally test them all, and start sending them out to customers, and then see how the units perform 'in the wild', and see what the reaction is like from his customers.  IF it all goes well, and there are no problems, and all the units perform as expected.. and there's still a lot of demand for the Replay.. and he's confident enough to place a much larger order, and be able to trust and rely upon the workmanship of the factory producing them without personally testing hundreds of units, then he'll be able to ship out more... THEN he can place that much larger order, and everyone can get the Replay they desire.

I think we need to ask ourselves, could we do what MikeJ has done? Do we have the skills to develop a board like this? Would we have the patience to work at it for so long? Would we be prepared to travel abroad and sign business deals, personally test 50 units, etc? We should't expect from others what we couldn't imagine doing ourselves.  MikeJ's done a lot more than most of us could ever do, and he's created an amazing little board with so much potential.  Let's cut him a little slack, give him a chance.  Let's not expect perfection or demand the impossible.  He's doing the best he can, and he's done a remarkable job so far at that IMO

[ShapeShifter]

The FPGA Replay will no more 'kill' the original MiniMig board than the release of the Amiga 4000 'killed' the Amiga 1200.  It's simply a new option for us to choose from. The Replay complements the MiniMig, it doesn't make it redundant.  The Replay will appeal to those of us for whom the original Minimig board was 'not enough', in the same way the Minimig will appeal to those who don't need the extra functionality/abilities of the Replay and are more concerned about getting their 'retro Amiga kick' on a budget than they are about expandability or high-end features.

It market terms, the Minimig is like what the A500+ used to be: a low-cost budget machine above the original Amiga 500, whereas the Replay is the Amiga 1200: the powerful and expandable mid-ranger, whilst the NatAmi is like the Amiga 4000: it's high-end in features but it has a price tag to match.   NatAmi will be the most full-featured option and will offer the users most out-of-the-box, but it will also be the most expensive option of the three, because it will offer many features which most people may not even want or have a need for (do we really *need* a full 512MB of RAM, plus IDE support and a floppy disk drive, when we've large SD card capacities now, and can simply ADF up any disks we wish to transfer over and use?)

The Replay is ideally placed in between these two.  It leapfrogs the original MiniMig in specifications, offers AGA and 030/040+ speeds, and is much closer to what the NatAmi will offer, whilst keeping costs down by removing unnecessary features which aren't required by the average user.   Thanks to it's daughterboard expansion slot and FPGA design, it can also be expanded in both software and hardware capabilities in much the same way an A1200 was expandable into something which could not only match, but even exceed, the out-of-the-box capabilities of the top-of-the-line A4000.

From my point of view, the Replay offers the perfect balance of power, performance, features and expandability and most of all, it does all of this at a very reasonable price.  For those who only want A500 capabilities to replay old Amiga games at the cheapest possible price, the original MiniMig design still remains the best option.

[ShapeShifter]

The average Amiga enthusiast is always looking to add extra RAM, a faster CPU, a bigger hard drive, etc regardless of what he buys.  
[..]
I'm always happy to support a great project and the FPGA Arcade had me drooling as soon as I heard about it.  As a long time Minimig user, I love the FPGA based machines and they get more use than my real Amigas... however I do treasure my A2000 and A4000.  

We're totally on the same page here Darrin It's strange, but the "Next Gen" PowerPC Amigas and OS4.x for some reason just never excited me in the same way the FPGA Replay has.  I think it's because I have always been a "Classic Amiga" fan, and I have never really had much interest or saw much benefit in buying a big, bulky, expensive AmigaNG system just to be able to run a small number of extra apps.  I'd rather see evolution and improvement of the classic, and at a low price.  Replay is much more in the spirit of what I've been looked and waiting for.

It will replace my expired A4000 very nicely, it's modern, it's compact, it's powerful, it's expandable, and it's NOT emulation but "the real thing."

haywirepc, if expandability is important to you, there's no contest between the two; you have to get a Replay.  The MiniMig just isn't designed for expansion. But if the ability to expand isn't important to you, and A500/ECS is enough, then Minimig is perfect, as Darrin said

Nope, I just turned mine on today after installing it in my new fancy minimig case.

It fired right up...

:roflmao:

[ShapeShifter]

Can I make an order or pre-order right now? How?

If you're interested in getting ahold of a Replay, the best bet is to send a PM or e-mail to MikeJ.  He's in the process of getting the latest batch of 50 out to people on his pre-order list.  Once that's finished, he's planning to produce a second, much larger batch of Replays.  There should be enough for everyone, but it wouldn't hurt to let MikeJ know you're interested

[ShapeShifter]

@ Mikej and freqmax:

Thanks, I'm sorry if I've caused trouble ! I was and I am  an almost complete noobie in Amiga World (Im basically a Sinclairist); anyway I think that  the implementation at least of original pluggable keyboards  of  various Amiga models (4000, CDTV, etc) could be quite....interesting.

I asked Yacubed about this, as I too would love to use a real, genuine Amiga keyboard with the Replay.  I was told that, whilst it would be possible to support it in the Minimig core, the onscreen-menu of the Replay itself also requires a keyboard and not just the Amiga core itself.  

So then the question becomes: Which keyboard protocol should be the default at power-on of the Replay board itself? If it's Amiga, then PS/2 keyboards used by most people won't work.  If it's PS/2, Amiga keyboards won't work.  Since the Replay is meant to be used to power many different cores (not just Amiga), it's easier just to stick to PS/2 keyboard support for the time being.

Oh, and welcome to the world of Amiga, Cignale! It's always nice to welcome someone new to our humble little community

[ShapeShifter]

It's also worth adding to Hattig's elegant description, that one of the unique advantages of reproducing the Amiga technology in this particular way is that the hardware designs aren't set in stone (or rather silicon, in this case!).  This means the design can tweaked, refined, upgraded and improved as developers create ever-better ways of doing things.  This can include finding ways of sidestepping old limitations, or even removing them altogether.

For example: The Amiga AGA chip set was coded to support a fixed 2MB of CHIP RAM.  This limited the resolutions etc. which could be supported.  That limitation has been removed completely in Minimig AGA / FPGAArcade Replay, which can support unto 50MB CHIP RAM.

Also, one question which comes up a lot is what kind of CPU the Replay etc. reproduces.  This question can actually result in confusing or imprecise answers, because it's possible to do things in FPGA which were never done in silicon.  For example, what would you rather have: A 68060 running at 100MHz but which is less compatible with software than a 68020 -- or an 68020 which runs at speeds which outperform an 68060, yet has the advantage of offering better compatibility?

I think we'd all prefer the 68020 - but if it makes us feel better, we can call it an 68050 or an 68070, or if NatAmi objects we can call it something else, it really doesn't matter.  What matters is what kind of performance we can achieve whilst preserving maximum compatibility too.

In the same way that CPUs can be tweaked and improved over the original designs, the Amiga's graphical and sound capabilities can be enhanced, as well.   We can incorporate chunky modes into our FPGA implementation alongside enhanced AGA modes.  It can be like having the AAA chipset and a Picasso graphics card as well, all coded into FPGA!

So, not only is the FPGAArcade Replay not emulation, but it also goes beyond simple cloning too.  This allows for the Amiga's design, chip set, to be further improved beyond the capabilities of the original.

[ShapeShifter]

Any A.T.A for the new core mike ?

Now, now - don't go bugging MikeJ about the new core. You're not supposed to do that.  That's my job :lol:

Seriously, I'm kind of laid back about waiting for it, because I know MikeJ will deliver on his promises, and he won't keep us waiting longer than necessary.  There's something very refreshing about that: someone meaning what they say, and delivering on it, in this community

When you think about it, Mike has probably written more code - and designed more hardware - than even Amiga, Inc. have done over the past 12 years*!

(* not counting hardware / software produced by third parties)

[ShapeShifter]

This is never going to be a mass market item.  I don't see that as a negative, though.

It doesn't have to be mass market to be very, very successful, though.  I see the Replay potentially succeeding in the same way the Amiga succeeded: it may not saturate the entire market, but it will find itself achieving some real success and popularity within certain niches.  It's already exciting the Amiga world; there's also the Atari world, the C= 64 fans, arcade and console fans, etc.

The nature of the Replay itself also means it has more potential than simply being whatever retro system you want it to be; it can become any system you wish it to be; retro with modern elements, a convenient development platform bridging two different technologies, etc.

We won't know exactly where and how the Replay will distinguish itself until it has started to ship in real numbers, however, and people get their hands on it and start to mess around with it.  That's when we'll see the real excitement and ipotential take off, IMO.

[ShapeShifter]

What we need now is decent cores which is usable for every retro-fan out there, so that the question of what board to use is not important, but what fpga board has the MOST cores out there which are maintained and fully working.

We need to get an Atari Core at least to get the Atari guys to jump on this board. Hoperfully, the fpga-replay board won't be promoted as a "Amiga Board". We need ANY user out there, no matter what core they like to run. And we need people to make and maintain good cores.

I couldn't agree more! For a very long time now, going back to the late 90's, I have hoped that it would be possible for Acorn, Atari, Amiga, and Apple (hey, all A's!) to come together in some fashion and work together for our common good.  

In the FPGA Replay, we've actually got a device where the success of one particular core is good for everyone else who's using the system, whichever core(s) they happen to prefer.    Random example: the likelihood of 68060/Coldfire/PPC accelerators and other expansions which are useable by Amiga fans is increased if there should be a large number of Atari users which these expansions would sell to.  And not just hardware expansions, but software/core projects useable by one community could be of benefit to another.  

The larger the market for the Replay, the better for everyone using it, no matter our core.

I truly hope the Replay achieves some big success, and my intuition is telling me that it should infact do very well

[ShapeShifter]

PAL LCD TVs handle both 50 and 60hz signals.

NTSC LCD TVs probably only do 60hz.

Unfortunately, it's not as straightforward as knowing which country you're based in and what the predominant TV standard is in that country.  I know it should be, but sadly it's not.  You have to check the TV's supported modes to see what will work.

With the present core and Replay hardware, the horizontal refresh rate is being scan-doubled from 15khz to 31khz to enable PAL/NTSC modes via a VGA-compatible monitor or flat screen TV.  But not all TV's will accept  PAL 50Hz input modes, even if you DO live in a country with a PAL 50Hz standard.  They often require 60Hz or more when being fed a signal via a VGA socket. This is particularly crazy when the TVs in question WILL quite happily display 50Hz refresh rates when the input in question is coming in via composite or component input sockets.

Of course, we have the option to display video via component or composite or other analogue output modes, but here we may have a new problem.  TVs might well be happy with 50Hz vertical rates from those particular input sources, but they will choke when fed the 31KHz scan-doubled horizontal rate - even though the TV supports that rate via the VGA socket!

In short, know what modes your TV is capable of supporting, and you will then have a better idea of how you should hook up your Replay unit to your TV.

Of course, I should add that none of this will be an issue in future! MikeJ is working on a new core which I'm advised will allow us to configure output modes by means of altering a config file stored on the SD Card.  This will allow us to select whether the Replay enables or disables scan-doubling, and presumably whether it is outputting a PAL or NTSC signal.  This should make future setup and configuration of the Replay much easier.