Better performance and 080 support?

[QuartexNOR]

I noticed that there already was a post regarding a faster Amiga core, but since the dialog there is somewhat angry, I figured I could post my request and reasoning in a new post.

I have been involved with next-generation Amiga developments for the past 9 years. I am the author of Quartex (QTX) media desktop, which is a re-implementation of Amiga OS designed for running on clusters, implemented in C/C++ and Object Pascal, and compiled to WASM and JS.
Since we more or less exhausted the compiler we started with, I had to take 18 months break from the QTX project in order to create a new compiler, RTL and IDE (Quartex Pascal). The desktop system also have a port of UAE so you can run Amiga 68k software, and access it directly in the browser. Or as I do, boot into the cluster with a thin linux bootstrap, and just fire up a Webkit renderer in kiosk mode -- which gives you the exact same experience as using a Linux, Windows or Amiga box.

Building QTX has been, and still is, a wonderful thing, because it allows me to explore possible paths Amiga OS could have taken if Commodore had survived. From pitching 68k as bytecodes (turning UAE into a virtual machine, much like Java or .Net) to abstracting the hardware away completely like I have done with QTX.

There is so much potential in all this, which brings me neatly to the Mister.

Working on real hardware

While QTX is awesome, I also love working on a "real" Amiga (and emulation, im not really a purist). And part of my work involves bringing FreePascal up to speed, and further port over the thousands of Borland Delphi (or Embarcadero Delphi) visual and non-visual components, which is 99% compatible with freepascal.
This means that we can tap into some 25 years of Delphi software development, and get everything from text-editors, word processors, browser engines, networking components, modern graphics packages running on the Amiga platform (including Aros and Morphos).

My challenge is as the title hints to, a lack of power. I am perfectly aware that I could cross-compile using UAE, or install freepascal on the Linux side of things -- but there is something to be said about working on the platform itself. Both for more reliable testing, but also because real hardware is more aggressive (read: "always on") than UAE. Whenever you compile anything substantial with UAE, the CPU hogs all the computing power from the chipset and interrupts.

I fully understand the need for a cycle-exact Amiga, that covers the legacy market and make it possible to enjoy classic titles, exactly like they were back in the day. Here the Mister core developers have done a wonderful job, and I applaud the effort.

Im interested in the future, not the past

There is however another branch of Amiga users. Namely those of us that would like the technology to grow and evolve to modern standards.
The Vampire FPGA is probably the most potent expression of this, where the Apollo team has pushed the Amiga forward towards PS2 standards. This is not much compared to where x86 and ARM is today, but a jump from A1200 to a PlayStation 2 with 512mb ram is a massive evolutionary jump, especially when you consider the time-frame.

With Vampire selling like hotcakes, would it not be a good idea to try and match some of its features?
The 68080 CPU instructions should be doable (without breaking the mold). AMMX operates with separate registers and opcodes, so it should not interfere that much with the traditional 68k mnemonics.

But what I really hope to see, is better speed. Which from what I understand would involve a "cycle agnostic" implementation of each instruction, and I would imagine a bigger instruction cache. I mean, a 040 that chews 68k instructions beyond 68060 (even without 080 support) would be a wonderful thing.

To compare: When doing a clean build of freepascal from scratch, my A1200 spends a week (yes, seven days!) building.
The same build is finished by the Vampire in a few hours. Obviously I don't go around building a compiler every day, but with the DE10-Nano actually having bigger FPGA banks, Linux on board to deal with HID and USB (and more) -- I don't understand why the core is not further evolved? I dont mean that in a negative way, im just curious. With so much potential, why keep the platform in the past?

I just compiled up a ton of networking components for FPC (freepascal) and started my first WebSocket server on a classic Amiga (hurray!).
But porting over a proper webkit renderer, video codecs, office packages ... that sadly requires more power than the Mister currently delivers.

It would be so much easier if we could just agree on a "next generation standard", and then move the platform forward. The 080 instruction set could be reverse-engineered and cycle-agnostic mode implemented (or roll a separate core where cycle accuracy is ignored).

I would be happy to pay a few hundred dollars for such a core. Im *NOT* suggesting closed source, im talking about supporting the FPGA developers during development and show my support. I also run several Amiga groups, and I am positive that quite a few would help finance that.

Beyond that, the Mister is awesome! Thank you so much for the RTG support, that is rock solid and crystal clear. Awesome work!

Thank you

Jon-Lennart Aasenden
A.k.A Cipher-Diaz of Quartex

[chanunnaki]

Interesting read QuartexNOR. I'd love a more "future-oriented/forward-looking" Amiga core.

May I ask, what's stopping you from creating this core yourself or gathering a team for this? You at the very least seem well positioned to at least outline the requirements/vision of such a project and get the ball rolling.

In many ways, Amiga seems like the ideal platform/core to get this kind of treatment as Commodore is still beloved by so many and experienced a untimely death as a company. MiSTer is borne out of the MiST project, and the name of both projects is derived partly from the Amiga name.
The Mega65 and Commander X16 project shows us there is a certain level of interest in modernising older computing platforms, but so far, the MiSTer project has been firmly looking backwards and replicating the past, and not yet looking at what can be done to lead us into the future. There seems to be perhaps some philosophical resistance to the idea of "imaginary" systems on MiSTer, I don't know if that is really the case though. There is the argument that the lack of any software for this proposed "Super Amiga" is the main stumbling block for garnering any interest from FPGA core developers. It would also make sense that replicating an existing system with known, concrete parameters is far easier than conceptualising a brand new system.

The main barrier for something like this to get created on MiSTer isn't perhaps a technical one at all, but a visionary one.

[chanunnaki]

Another thing I would like to add is that I too believe that the "Minimum" core should be left as-is, with future development focused on the refinement of the original Amiga experience.

I am however, all for a brand new separate Amiga core that is more future-oriented and focused on pushing the DE10-Nano as far as it can conceivably go.

[ARCADEAGES]

I don't have too much to add to the technical discussion, but I can point out that the Sinclair Spectrum core on MiSTer has some of the features found in the modernized theoretical-cum-reality "ZX Spectrum Next" dedicated FPGA micro-computer. Eg. fast tape-load, RAM snapshots, and faster clocks going all the way up to 56mhz.

[throAU]

Weren't the Vampire guys working on standalone hardware? Wouldn't that cover this niche?

[caffeinekid]

Another thing I would like to add is that I too believe that the "Minimum" core should be left as-is, with future development focused on the refinement of the original Amiga experience.

I am however, all for a brand new separate Amiga core that is more future-oriented and focused on pushing the DE10-Nano as far as it can conceivably go.

MiniMig, and MaxiMig?

[QuartexNOR]

May I ask, what's stopping you from creating this core yourself or gathering a team for this? You at the very least seem well positioned to at least outline the requirements/vision of such a project and get the ball rolling.

My skills are limited to software development. Verilog is a very different ballgame, and requires electronics know-how. I am not even going to pretend to know that. Sure I could learn, but verilog is not something you pick up over a month. It is a part of cpu and SoC design studies at university.

When it comes to investments, I am in the same boat as everyone else these days. Corona has hit everyone hard and I have just enough to keep afloat.
Which is also why the Mister represents a good value for money. The vampire is awesome, but with shipping im looking at a €700 investment to get one. That is what a new A3000 used to cost back in the 90s.

Having said that, I have no problem donating say $50 a month to the core developer, and im pretty sure other Amiga users would do the same. It all depends on how the R&D is pitched to the community.
In this case the core devs could setup a patreon account, or just use Github's donation system. A blog or website where the author delivers weekly overview, goals etc --- this is how the new economy works.

I just feel the Mister has so much potential going for it, that it would be a shame to keep it prisoner of the past.
The mister developers are now in a very real position, to produce new Amiga machines en-mass without the blessing of greedy patent owners.
If you marry that with the updates to Aros (or Apollo-os as the fork is called), you have a pretty sweet package that bypasses the whole cluster that is the legal situation -- and the community is home free.

You mention some philosophical differences. Yes there seem to be two camps: those that regard the Amiga as pure legacy that will never amount to anything new. And those that want to push it forward. The reason I mention the 68080 cpu/chipset, is because it has already been defined. If Mister aligns with it, we have a standard that is accepted (or available) for thousands of Amiga users.

When it comes to using the new features, I think this is a matter of framework. Old school coding banged the hardware directly, but modern game and demo coding tends to use a framework like SDL to act as a proxy between underlying capability and their code. This way, they can build the same codebase and target morphos, aros and workbench (just swap out the resource-files for older machines).

For example: the width and height of a bitblt() operation is quite restricted on a classic Amiga. So when you call the blit() method in SDL, it figures out how many actual blitter operations is required, and gets the job done. On a Vampire or Mister it would recognize the 080 chipset, adjust the possible width/height -- and as a result the game/demo would take advantage of the better hardware, without the coder having to manually bang the hardware.

There is a ton of projects waiting to be converted over, but people sadly wait around for a standard to emerge.

[QuartexNOR]

Just to follow up with some context on why the Mister core developers should consider this:

Since Commodore went under, different parts of the Amiga patent portfolio is owned by different individuals and companies. The most loud being Cloanto and Hyperion, which has been locked in a perpetual lawsuit for the better part of 2 decades.

The Mister core developers are now in the position where they can end this licensing nightmare. They provide an fpga core that is 100% unaffected by whatever legal mayhem is happening to the original, greedy owners, and evolve it on the users terms rather than some boardroom assholes. And that is a very special place to be, because the Amiga as a platform is extremely versatile (otherwise it would not have survived up to this day).

If we then look at the work the Apollo team is doing with Aros - this can be forked and made compatible with Mister. It is an open-source GPL project after all, so the Aros part can be cloned and adjusted to whatever changes they do for Mister.

Suddenly these FPGA core developers have gone from hobbyists, to actually having a real influence on the evolution of Amiga.
By evolving the platform more in-tune with Vampire, Amiga fans suddenly have options. The Mister also has that sexy linux aspect, where the OS can offload tasks to the ARM side -- which is pretty much the coolest sidekick Amiga can have.

There is a wealth of evolutionary paths that can be explored in this material. And if presented correctly, it can also bring some money to the developers in the form of support and donations.

Note: when I write "more in tune with the vampire", I don't mean licensing fpga code from Apollo. I mean reverse-engineer the 080 instructions so that, eventually, Vampire exclusive software can execute on the mister. So a substantial re-write of the Mister core would have to be refactored and re-implemented.

[Fularu]

Weren't the Vampire guys working on standalone hardware? Wouldn't that cover this niche?

The V4SA is not the answer. It's in far too limited numbers and costs an arm and a leg (about 600eu)

I don't think there's even more than 500-600 units out there (and I'm beeing generous with that number)

[Bas]

Sounds like AO486 but for Amiga. The AO486 is not an absolutely accurate reflection of a 1990's PC but that doesn't stop the DOS crowd from having a wonderful experience on it. Compatibility-wise it has rough edges, as would a big Amiga if it's not an exact replica of a released historical model. This circles right back to wanting to use MiSTer as a platform for conservation of things like AMIX. If a turbocharged "MaxiMig" core can be created I'd be very happy if it can be compatible with actual big-box Amiga software of yore. Neither Vampire or UAE are appropriate platforms for this purpose if you ask me.

[QuartexNOR]

The V4SA is not the answer. It's in far too limited numbers and costs an arm and a leg (about 600eu)
I don't think there's even more than 500-600 units out there (and I'm being generous with that number)

I think there are more Vampire units out there than we suspect. When they visited the Amiga32 event, they had 200 units (stand alone) with them. This was sold in less than 1 hour. I myself have owned 2 vampire units, and pretty much everyone I know from the scene has bought one.

There are more vampire's in accelerator form out there than stand-alone. My guess would be in the ballpark of 2000 stand-alone units. No clue how many accelerator boards have been sold, but it must be in 10.000 range by now.

The Amiga community is much larger than I first imagined, with different communities operating out of different online forums. Facebook has the largest forums (100.000 strong), but you also find large usergroups on reddit, good old newsgroups, irc and a ton more. It is actually quite amazing to think, that a computer 30+ years out of production, is still so wildly popular today. Heck, even my own group "Amiga Disrupt" has 3000 members.

Vampire could be the answer. In fact, right now when I look at the marketplace, only two systems really float to the surface. You have Vampire IV+ Standalone, and then there is Morphos. The latter is very interesting because it exists for both PPC and x86. With some luck, they will also cover ARM or RISC-V some time in the future. Morphos is by far the most maintained and evolved Amiga OS codebase. Aros seem to be all over the place these days.

This is why I really want the Mister to kick it up a notch. Yes, Vampire would suit my needs -- but like you mention, at what cost? I can buy 4 Mister setups for the price of a single Vampire V4SA setup.

I hope and pray that @sorgelig or one of the other core developers picks up the mantle. There is both potential and honest money to be made, as long as the right vision is in place.

[Fularu]

The V4SA is not the answer. It's in far too limited numbers and costs an arm and a leg (about 600eu)
I don't think there's even more than 500-600 units out there (and I'm being generous with that number)

I think there are more Vampire units out there than we suspect. When they visited the Amiga32 event, they had 200 units (stand alone) with them. This was sold in less than 1 hour. I myself have owned 2 vampire units, and pretty much everyone I know from the scene has bought one.

There are more vampire's in accelerator form out there than stand-alone. My guess would be in the ballpark of 2000 stand-alone units. No clue how many accelerator boards have been sold, but it must be in 10.000 range by now.

The Amiga community is much larger than I first imagined, with different communities operating out of different online forums. Facebook has the largest forums (100.000 strong), but you also find large usergroups on reddit, good old newsgroups, irc and a ton more. It is actually quite amazing to think, that a computer 30+ years out of production, is still so wildly popular today. Heck, even my own group "Amiga Disrupt" has 3000 members.

Vampire could be the answer. In fact, right now when I look at the marketplace, only two systems really float to the surface. You have Vampire IV+ Standalone, and then there is Morphos. The latter is very interesting because it exists for both PPC and x86. With some luck, they will also cover ARM or RISC-V some time in the future. Morphos is by far the most maintained and evolved Amiga OS codebase. Aros seem to be all over the place these days.

This is why I really want the Mister to kick it up a notch. Yes, Vampire would suit my needs -- but like you mention, at what cost? I can buy 4 Mister setups for the price of a single Vampire V4SA setup.

I hope and pray that @sorgelig or one of the other core developers picks up the mantle. There is both potential and honest money to be made, as long as the right vision is in place.

All vampire cards combined (V1, V2, A500/600, V1200 and V4SA) it's 5800 units so far since the Vampire V1. At Amiga 32 they had 54 stand alones.

[kathleen]

Vampire could be the answer. In fact, right now when I look at the marketplace, only two systems really float to the surface. You have Vampire IV+ Standalone, and then there is Morphos. The latter is very interesting because it exists for both PPC and x86. With some luck, they will also cover ARM or RISC-V some time in the future. Morphos is by far the most maintained and evolved Amiga OS codebase. Aros seem to be all over the place these days.

Are you sure that MorphOs exists for x86 ??? To my recollection MorphOs was shown running on x86 at a party but so far only the PPC version exists and no news since then of a x86 porting.

[Caldor]

Vampire could be the answer. In fact, right now when I look at the marketplace, only two systems really float to the surface. You have Vampire IV+ Standalone, and then there is Morphos. The latter is very interesting because it exists for both PPC and x86. With some luck, they will also cover ARM or RISC-V some time in the future. Morphos is by far the most maintained and evolved Amiga OS codebase. Aros seem to be all over the place these days.

Are you sure that MorphOs exists for x86 ??? To my recollection MorphOs was shown running on x86 at a party but so far only the PPC version exists and no news since then of a x86 porting.

I am also pretty sure MorphOS still only supports PPC. But I also do think one of the alternative AmigaOS branches got support for some other hardware within the last few years, and I guess it might have been MorphOS. I do not remember what exactly it was, just that I made a note of a rumor that something got ported to another system. Might have been AROS.

[guddler]

I think you've already hit the nail on the head here - and that is availability of core developers. As in, people capable of developing such cores for the DE10-Nano. There are just not that many in the grand scheme of things and to my knowledge none of them are being paid full time to do this. So that leaves the same age old problem of it having to be done in spare time as a hobby. It then follows that the people involved have to really have a passion for the project and want to do it. Then you have to factor into the equation that there are some that are already doing this in / for Apollo and for the Replay board (think I've got that right). By the time you get to this point it wouldn't surprise me if the number of people that are capable are now into the single digits, or low double-digits at the most.

I'm in the same boat. I'm a software guy. I've tried a few times to branch out into FPGA development but it's a different mindset entirely and not that easy. Well, that was my experience anyway. I didn't get far beyond just coding up some test patterns on screen.

I think the idea is great. Sure, why not have two cores? The lower speed (though seriously, not THAT slow) Minimig and some new core. I would question if there is enough power in even the DE10-Nano with it's ARM co-pro on the SOC for what you want though? Doesn't the Vampire have a beefier FPGA, which is part of the reason it costs so much more?

[CeeDeeVee]

I would question if there is enough power in even the DE10-Nano with it's ARM co-pro on the SOC for what you want though? Doesn't the Vampire have a beefier FPGA, which is part of the reason it costs so much more?

No, actually MiSTer has a more powerful SoC, they both use the same cyclone V, however the DE10nano has significantly more (110k vs 77k) logic elements, as well as the ARM co processor. The Vampire has s couple of advantages that I can see, 1) custom board design has more flexibility in terms of clocking without the other peripherals and overhead on of the DE10 (the minor point), commercial development means that people are getting paid to work on the vampire, and the sales are driven by the unique and class leading performance, rather than purely passion project as is the case with most MiSTer development.

I’d also love to see an fast 68080 core, but also very thankful for everything that has been done so far, and it’s totally the right choice to start with accurate reproductions of these classic systems.

[guddler]

Ah, ok - thanks for the correction. I couldn't remember what it had since I sold mine. It was far, far too buggy when I owned it (Dec 2019 to approx May 2020) so I sold it on (at cost, no profit) while there was huge demand and people just couldn't buy them with a view to buying back in should the project mature to the point I wanted. Now, to be honest I've moved on and don't see me buying another.

[ericgus09]

Ah, ok - thanks for the correction. I couldn't remember what it had since I sold mine. It was far, far too buggy when I owned it (Dec 2019 to approx May 2020) so I sold it on (at cost, no profit) while there was huge demand and people just couldn't buy them with a view to buying back in should the project mature to the point I wanted. Now, to be honest I've moved on and don't see me buying another.

The most recent core updates are amazing too bad you sold your unit off .. thats the magic of FPGA any bugs issues are easily fixed with firmware updates.. release 4 and more so release 5 4vsa cores have been pretty major milestones.. I honestly think people forget that the Vampires are not like fixed silicone and can / will be updated fixed .. and new features added over time.. also if things are not working .. post it on discord and get them to look at it.. I have (and others) with some games/demos and all those issues are by in large now fixed.. and those fixes also addressed issues with others .. as I said it just gets better and better. Release 6 core is looking to be amazing too. ApolloOS is coming along nicely and the recent development of the Vampire Bootloader (formerly 4VSA bootloader) allows you to easily and quickly multiboot ALL amiga OSs .. with shared mounted filesystems and such .. its really a great time to own one..

[Michael1260]

Eric, you are everywhere

Totally agree with you !!!

and I can add an excellent support team

[ericgus09]

Eric, you are everywhere

Ive been here for a long time.. and before that the old atariforums mist section,, ;-D but yes..

[ron]

I got the wrong thread when answering, but it's the same. I reply here because basically it's talking about the same thing.

I don't wish what I'm going to say sounds like sarcasm, but ... when there isn't even an open core of 68030, don't you think the expectations are too high?

Forget about the Apollo Team's Vampire, it's a proprietary core and will not be used outside its projects. According to a Finnish colleague, they say the Vampire's core reeks of Coldfire. There is no data to support this claim, but it is clear that MISTer poses a serious threat as far as its intentions are concerned.

Perhaps an alternative is to start with something more affordable, like the M68030, also a PMMU and it would be the gateway to cores that support Atari Falcon 030, Macintosh II, Amiga 3000 and even NeXT.

Dreaming is free, I'm just trying to be realistic with the items currently available. No acrimony, please. The best of the news for those of us who like Motorola processors would be a functional 68030 core. Those are major words.

[Bas]

What Ron said.. exactly that.

[QuartexNOR]

Ron: There is a lot that is unknown about the vampire situation, but the whole coldfire rumor, let me explain:

When the vampire started out there were some very loud open-source advocates in the community that thought the vampire was open-source. So they jumped on it and started to create clones. Obviously this was not popular so a legal threat was issued and they had to stop.

The same people who had started to illegally copying the hardware (with a copy of the then un-protected core) did not respond well to this, even though they are clearly in the wrong here. Instead they turned around and started to blame and thrash-talk the vampire and apollo.
And they have been doing this for years now.

I actually had to kick two of them from my Facebook group, because whenever someone mentions "vampire", they jump in to thrash talk it. I almost did not believe it myself until i saw it, but they do so consistently at every chance they get.

Now fast forward to the Phase-5 ordeal. Some guy bought the rights and started to produce the Blizzard cards.
He made a simple mistake: as you probably know, SBC's are designed in layer, almost like photoshop. And the guy had accidentally disabled the grounding layer before shipping the boards off to print.
The same people I mentioned above went on a full-out assault on the guy. Completely burying him. More and more people got involved, and at one point it was probably 100 guys against 1. It was a disgusting example of online bullying.

But here is the thing, namely the motive: Those people I mention did not do this because of some valid reason. They did this because they make a living creating clones. They WANT everything to be open-source so they can copy and sell them.
It is the most blatant example of greed and pettiness I have ever seen. Seriously, it was like something out of a novel or fiction.

The same people start rumors and go out of their way to make the vampire look bad.
My point being, you should take the rumors with a bag of salt, because in 90% of the cases it turns out to be untrue fluff.

[lordoftime79]

I am a avid Amiga fan and have a few classic machines aswell as the V4 standalone... I think it would good to impliment some of the 080 features, I mean we already have a large pool of fast ram and RTG why not build on that? and yes maybe have it as a seperate core for amiga power users.
I dont know much about FPGA dev but I would help where I could.

[ByteMavericks]

What did happen to Gunnar’s offer of 060 level performance to any project with the option to unlock the 080 performance at a price? I only saw reference to it on a1k.org a while ago

Other than that, there are no better 68k cores than the one we have.

[Fularu]

What did happen to Gunnar’s offer of 060 level performance to any project with the option to unlock the 080 performance at a price? I only saw reference to it on a1k.org a while ago

Other than that, there are no better 68k cores than the one we have.

Mikej from FPGA Arcade Replay is developping a 68040 core and porting it to the DE-10 NAno

https://www.fpgaarcade.com/replay2-and- ... o-support/

[ByteMavericks]

That looks to be a very good way forward: thanks for the info!

[ericgus09]

What did happen to Gunnar’s offer of 060 level performance to any project with the option to unlock the 080 performance at a price? I only saw reference to it on a1k.org a while ago

Other than that, there are no better 68k cores than the one we have.

Mikej from FPGA Arcade Replay is developping a 68040 core and porting it to the DE-10 NAno

https://www.fpgaarcade.com/replay2-and- ... o-support/

He's been at that for some time, I know he took a detour to create that 060 daughter board for his upcoming FPGA Arcade 2 .. not sure where he is on the 040 core..

[kolla]

Oh lordy, I just realised who QuartexNOR is and that exlains all the revisionism regarding "the vampire situation" and all the rest...

Anyways - all this writing and wishful dreaming of a faster 68k core for MiSTer is pointless, for many reasons...
* the development of the 68k cores used are mostly done by atari people who don't frequent here so much
* they mostly hang around atari-forum.com, not here
* if you want fast 68k CPU, try software emulator the ARM first, I dare you, I double dare you
* most people "demanding" faster CPU claim they cannot do it themselves because they are "software people", so again I challenge you to try software 68k emulation on the ARM!
* Sorgelig doesn’t care that much about Amiga aside from playing games, so don’t expect much
* if you want faster CPU core, you most likely must do it yourselves.

[Fularu]

What did happen to Gunnar’s offer of 060 level performance to any project with the option to unlock the 080 performance at a price? I only saw reference to it on a1k.org a while ago

Other than that, there are no better 68k cores than the one we have.

Mikej from FPGA Arcade Replay is developping a 68040 core and porting it to the DE-10 NAno

https://www.fpgaarcade.com/replay2-and- ... o-support/

He's been at that for some time, I know he took a detour to create that 060 daughter board for his upcoming FPGA Arcade 2 .. not sure where he is on the 040 core..

The 060 DB was for the first replay.

The replay 2 has a full FPGA implementation of the 68040/60 (the same processor really) per Mikej. This has been discussed here and you can find his posts on the subject.

I see kolla's here to mention how pointless such an endeavor is as usual