Amiga RTG support (Update: Released!)

[rhester72]

I'm acutely aware of the marketing push, and how acutely it failed - see the new(ish) owner of ProWrite and ask about how many unsold copies they took possession of as evidence. Good, bad, or otherwise, the Amiga was primarily a games machine in NA as it was anywhere else.

I love RTG. It's in both of my big-box Amigas and I use it semi-actively (more so when it was being developed) in WinUAE - I just can't for the life of me see where it has a place in FPGA, nor have I heard anyone speak to that particular point. People want it, OK - but *why*? More specifically, why FPGA, when what they want has already existed for at least a decade and is readily available with exacting compatibility versus real-world hardware and 100x the performance? What 'the people are clamoring for' is only a download click away.

(What I'm really driving at here: FPGA is fantastic at one thing - hardware synchronicity. An unexpanded integrated-keyboard Amiga can't be done well without it. Once you got up to A2000 accelerators, or the A3000 or A4000, the required tight synchronization between CPU and data busses went away entirely, and thus is perfectly suitable for highly accurate software emulation where FPGA has nothing new to offer and suffers FAR higher implementation complexity.)

I'm a HUGE fan of FPGA. I'm also a fan of using a hammer for nails versus a sledge. The right tool for the job always wins...and when it comes to anything beyond basic AGA, FPGA isn't the tool for the job and really never will be.

[kolla]

I don’t believe anyone today will wish to use Commodore era RTG What people ask for is chunky modes and a P96 driver - “Picasso ‘96” as 1996 two years after CBM was gone.

As for geography - all the three dominating RTG systems for Amiga were developed by competing German companies, and two of them years after CBM was already dead. The only RTG CBM did, was for AMIX, not AmigaOS.

[kolla]

I'm a HUGE fan of FPGA. I'm also a fan of using a hammer for nails versus a sledge. The right tool for the job always wins...and when it comes to anything beyond basic AGA, FPGA isn't the tool for the job and really never will be.

Define “beyond”. AGA is a heck more tricky to implement than a simple high res 32bit framebuffer for RTG, really. The V2 Vampire cards output 1280x720 32bit modes with P96, all done by the Cyclone III FPGA - the MiSTer has a Cyclone V (just like Vampire V4), and can easily do the same. Many other core already go “beyond basic AGA”)

The trickery is in how to implement it and present it to AmigaOS and P96 - as a graphics card (“outside” of the Minimig core, like using MNT2000 or what it was called) or as an extension of the chipset (extending the Minimig core, like in Vampire cards). And on MiSTer there is also a third option, to use “the other side” (the ARM processor) to help with the pixel pushing (like with ZZ9000).

All of these paths can be interesting challenges to developers, but what do they in the end offer that isn’t already available by using software emulation? Earlier, it was discussed quite a bit to do “hybrid emulation”, with the 68k CPU emulated on the ARM while keeping chipset on FPGA, as software emulated CPU could bring higher CPU speed. Same can be said for RTG. So what is left on the FPGA if you want software CPU to get fastest CPU and framebuffer via the ARM to get fastest possible RTG. Audio? But maybe you prefer using RTA and AHI to get high quality sound as well? By then, the FPGA is pretty much redundant and what you want already exists and runs on MiSTer.

[ericgus09]

Yeah, all the things you cannot have with the Apollo core guys, since they share nada. For years Gunnar and his minions (hey there, Eric) have been touting “if you’re so unhappy about the apollo core, why don’t you just write your own” and here we are - someone is both willing and capable, and now it’s suddenly “uhm, no, we cannot allow that”. Well f that, it’s not up to Gunnar anyways, the V2 hardware is NOT his design, nor his product - it was very important for him to point that out earlier.

And where is the open sourcing of SAGA that was announce more than three years ago?

for the record I am an unpaid minion .. I do it for the enjoyment I get out of the V2 and to counter some of the fud and mis information that gets spread around like manure.. most of it is either horribly out of date, patently false or just someone who has a personal axe to grind against the team for whatever reasons (hi kolla)..

Secondly he was a bit hostile with his requests and anyone who knows anything about the apollo knows they wont publish the sort of information he was looking for {seriously think about it and think why they said no -- its logical to most people} (eg the specs so he can write an alternate core) and when he was told .. no sorry not supported, he got angry ... well no surprise there.. I was at least suggesting a viable alternative (mister) but I digress.. I honestly think he was more interested in trying to reverse engineer the vampire than anything else, otherwise he would have taken the suggestion to code on the mister more seriously ..

And where is the open sourcing of SAGA that was announce more than three years ago?

Go ask him .. my guess is as they are still working on it .. it wont be published yet.. but im sure he would be happy to answer coding questions about it, he does that a great deal actually .. (I know I read the forms regularly) .. and actually its coding for Apollo/Saga questions he tends to respond the most on .. Seriously.. go ask him .. see what he says.. but dont be disingenuous about it ..

[ericgus09]

I just can't for the life of me see where it has a place in FPGA, nor have I heard anyone speak to that particular point. People want it, OK - but *why*?

First, personally, I don't want to use WinUAE (and technically as a non-windows user I cant, im stuck with FS-UAE and I dont even want to talk about what sort of disaster that is on the UI/UX front) .. ideally I would really like all my non-classic actual real Amiga needs (eg stuff thats not running in an actual Amiga) to be centralized in one location, right now the Mister does a pretty good job at recreating the subtly of many actual retro machines and its also where nearly all my other retro needs are being serviced from, having limited office space this is great, I dont need 14 different setup with all the associated accessories, monitors, keyboards, mice, etc.. , I want to consolidate what I have in my limited 3mx3m (9ftx9ft space) and more importantly being FPGA so the recreation of said system is very smooth and generally accurate, its not software emulation with its minute but perceptible latency and other issues due to the sequential nature of software emulation that inherently comes along with it .. if I wanted that I would settle for a raspberry Pi and be done with it .. but I have tried that and it just isn't really the same .. it feels "off" .. honestly since getting a Mist (then a Mister after) I haven't really even touched the Pi I had .. its just inadequate now.

Secondly .. there are a number of RTG games, demos and applications I would really like to be able to enjoy (or use in RTG mode -- eg telecom software for internet BBSing that I do) on the Mister in RTG .. currently that's not possible.. fortunately I have other means right now (the vampire) but its not located in an ideal space .. the mister is ..

Again .. because you don't see a value in it .. doesn't mean there isnt some value in it for others who use the machine in a different use case than you do.

[bazza_12]

I find the whole convo really interesting. Was the Amiga a games machine? Was it a productivity platform? Who designed this bit-that bit? Why? etc etc.. Blimey.. If someone wants to develop RTG support for minimig core then great, if they don't then that's great too. Isn't it the same with any core development? No one is being forced to do it.

[mahen]

Hi ! Even if not using RTG for "retro-productivity" purposes, it's EXTREMELY useful to be able to save all your chip ram and browse one's games collection with the Glow icons. Otherwise you are constantly stressed as you don't know if your games are gonna crash or not

As for the "retro-productivity" use cases, I think it's rather for the fun of creating in the same conditions. As I decided not to have any other real computer at home (apart from my wife's), it's still pretty handy for me to benefit from a fast / high res display to chat with friends on IRC, download stuff from IBrowse, read mails, write articles...

Of course, if I had to choose, I'd still prefer to see a 040 implementation but those are not related at all

[akeley]

There truly is no escaping Amiga's Curse. Poor ol' girl had a hex put on her back in the day and it's been like that ever since.

I've run away from eab recently to avoid all these bitter disputes (Vampire, what Amiga is /was, etc) and presto!, here they pop up again

[lomdar67]

Again .. because you don't see a value in it .. doesn't mean there isnt some value in it for others who use the machine in a different use case than you do.

You can bend this type of argument any way you wan‘t.

Because you see a value in it it doesn’t mean there is a value in it for the majority of MiSTer users.

Don‘t get me wrong, if someone want‘s to but his/her time into developing it, fine with me. But being a developer my self I can understand why people won’t do it. And I think discussing this here for days won‘t probably change this.

[robinsonb5]

I've run away from eab recently to avoid all these bitter disputes (Vampire, what Amiga is /was, etc) and presto!, here they pop up again

You need to run further away than that to escape disputes! People failing to disagree gracefully is basically the internet's defining characteristic these days!

As for why people would want this - if someone has an FPGA device for gaming and in the middle of a gaming spell decides to break out some productivity software for a bit of nostalgic fun, it breaks the flow to turn off the FPGA device, go and get a laptop, RPi or whatever, and boot into that. Anything that makes the FPGA device more capable is a good thing, provided it doesn't actively detract from its core functions. If it required an unreasonable amount of time and effort or took up FPGA space that prevented something else from being implemented then maybe it could be argued it's a bad thing - but if core development's being done as a recreational project then it really just comes down to what particular developers feel like working on.

Don‘t get me wrong, if someone want‘s to but his/her time into developing it, fine with me. But being a developer my self I can understand why people won’t do it. And I think discussing this here for days won‘t probably change this.

Discussing it endlessly won't change it - but the thread was originally about perhaps giving financial motivation to a developer - which is something that could potentially change it. I'd step up myself if I actually had a Mister (I don't at the present time, and my hands are currently full elsewhere) but what I'm working on currently should be easy enough to port when the time comes.

[mahen]

Actually, there is someone planning to implement it. But I'll let him announce it.

[kolla]

I guess none of you guys even tried to run UAE under Linux on the MiSTer?

[ericgus09]

Actually, there is someone planning to implement it. But I'll let him announce it.

And (I dont know if its related) but MikeJ of FPGA Arcade fame actually released RTG sources he uses for his implementation of RTG on the arcade .. so a good bulk of it is potentially done.

[ericgus09]

I guess none of you guys even tried to run UAE under Linux on the MiSTer?

Thats cheating .. I do that with Basilisk II for the mac .. but I only do it to mange the boot.vhd file the MacPlus core uses..

[apolkosnik]

So, did anyone try to get the RTG from Mike J going?

[apolkosnik]

[

for the record I am an unpaid minion .. I do it for the enjoyment I get out of the V2 and to counter some of the fud and mis information that gets spread around like manure.. most of it is either horribly out of date, patently false or just someone who has a personal axe to grind against the team for whatever reasons (hi kolla)..
...
Secondly he was a bit hostile with his requests and anyone who knows anything about the apollo knows they wont publish the sort of information he was looking for {seriously think about it and think why they said no -- its logical to most people} (eg the specs so he can write an alternate core) and when he was told .. no sorry not supported, he got angry ... well no surprise there.. I was at least suggesting a viable alternative (mister) but I digress.. I honestly think he was more interested in trying to reverse engineer the vampire than anything else, otherwise he would have taken the suggestion to code on the mister more seriously ..
...

Dude, wtf?! Stop spreading fud and lies.
I wanted to flash something onto my own V2, I asked once. Gunnar told me not supported. I've asked for a hint based on the V1 code base, which is GPL, and Igor told me that they won't open source anything. Fine. They deleted the whole thread, fine with me.
At no point I was angry with them, all I wanted to do was to roll something that could run my code on V2 without investing too much time into guessing pin assignments to ensure I don't turn my V2 into a doorstop. </end drama>

[ericgus09]

Whatever .. i grow tired of discussing that topic with you.. moving on (im sure others are equally disinterested).. I will not comment further on that entire debacle.

[Sorgelig]

MiSTer is fully open but i don't see anyone willing to add RTG.
It's interesting to hack closed platform than write something for open platform..

[chaos]

Hey Sorgelig,

if no one else beats me to it, I was planning to work on RTG. But first I want to implement the 68k CPU emulation on ARM.

[petarku]

hi @chaos , just to ask because the nick is familiar, are you one of the guys that basically created minimig ?

[Mike]

As I mentioned we (fpgaarcade) have decided to support the de10-nano platform.
We will be releasing firmware, binaries and the source as we go along.
The motiviation is to get the Amiga core out, but all our stuff is built with CI so everything should be available.

[apolkosnik]

MiSTer is fully open but i don't see anyone willing to add RTG.
It's interesting to hack closed platform than write something for open platform..

Hey Sorgelig,
if no one else beats me to it, I was planning to work on RTG. But first I want to implement the 68k CPU emulation on ARM.

I'd like to help with any of the menial tasks for any of these things.

Perhaps this could be useful to chaos https://github.com/michalsc/Emu68

[Kitrinx]

As I mentioned we (fpgaarcade) have decided to support the de10-nano platform.
We will be releasing firmware, binaries and the source as we go along.
The motiviation is to get the Amiga core out, but all our stuff is built with CI so everything should be available.

Is fpgaarcade a mister clone? Why not just contribute to the project. I guess everyone wants a piece of the pie.

[lomdar67]

It's the FPGAARCADE replay board it's not a "MiSTer" (or better D10-nano) clone:

https://www.fpgaarcade.com/

You can read more about the D10-nano support here: https://www.fpgaarcade.com/replay2-and- ... o-support/

Maybe Mike can shed some light on it but as far as I understand it they will not support MiSTer but the D10-nano. Their goal ist to bring the Replay ecosystem to the D10-nano. As everything will be available on GitHub this could probably be ported to the MiSTer...

[Mike]

Kitrinx,
We've been developing cores for years - many of them have been ported to mister.
My board has not been available for ages, and while we are working on a new one, we decided to support the de10-nano board that mister uses.
This is simply to let more people use our stuff.

"Why not just contribute to the project?" sometimes choice is good. We do things a bit differently.

Basically it's quicker for me to support the de10 board rather than port my stuff to another framework.
Because we have a common framework and build system, and we've already done the work for the Ardunio Vidor, it's not much additional work.
Of couse, as soon as I release the sources it (like eveything else I do) gets ported, and that's ok, but it results in fragmentation of the code.

Anyhow, off topic. Any questions feel free to ping me or discuss on slack.

/Mike

[mbo77]

Mike, is there any estimation on the timeline?
Only a very rough one, of course.

[Mike]

We are making good progress. I'll post updates on the website.
week or two I hope, but other people are doing uboot/linux and some of the build chain so it's hard to predict exactly.

[mbo77]

@Mike
Thanks. I was merely interested if we're talking several months or some weeks from now on.
So I'm really eager to look into it once something can be delivered.
Thanks!

[mbo77]

About the CPU, what about this project?
http://eab.abime.net/showpost.php?p=141 ... tcount=308

[Mike]

I think Mike is using the TG68K. I get about 80MHz single cycle operation in the Ultrascale part.