Virtio GPU, QEMU, KVM, etc.

Interesting comments and discussion regarding the Virtio GPU driver announcement. So much so that it's taken over the Amiga38 thread. Let's separate it out.

From here:
Quote:
@flash
I'm glad some have realized that Virtio is useful for more than just emulation. Yes, it's possible that we could run AmigaOS 4 natively within a KVM, and take advantage of the host OS' drivers via Virtio drivers. Nobody has tried it yet, but I'm sure someone will in due course.

This might allow us to bypass the "driver problem." A consistent problem for us has been getting working drivers for AmigaOS. It's delayed the release of motherboards by years, and we still end up with unused hardware (e.g., the X1000's onboard ethernet).

I'm envisioning a minimal host OS whose only job is to run the Virtual Machine (VM) containing AmigaOS, and run the actual device drivers. AmigaOS could use any GPU, ethernet, scsi device via their respective drivers. Got an nVidia graphics card? No problem!

Running the OS in a VM isn't ideal, but it's a lot better than new hardware being delayed repeatedly because porting the OS and writing the drivers took forever. In the A1222's case, those delays are so long that they caused a 3-4x price hike. If it had been released when the hardware was ready (pre-supply-chain issues), then it would have been a low-cost entry machine.

@all
This brings us to the "will emulation kill future OS4 hardware" question. They way I see it, we won't get future new hardware without a (large) influx of new users and developers. We need enough users & developers to be able to fund the hardware development (and OS porting) costs. I don't see that happening with existing hardware. The A1222 could have made a difference if it had been released years ago at its original target price.

I much prefer real hardware with the OS running natively, but emulation gives us a much better chance of getting new users and developers. KVM + virtio drivers potentially drops the upfront development costs for porting to newer hardware. Lower upfront costs in turn, could make newer hardware at small volume more feasible. When/if the userbase grows enough,** then we can consider ditching the VM and going "fully native" again (on whatever ISA makes sense).

Hans

** The userbase growing enough would probably have to be more than 20x.

@Hans

Completely agree with you, so go on with Virtio drivers and let AmigaOS be developed in a virtual machine.
A virtual machine does not have hardware failures and year after year gain speed due new new processors released.
It also can use all host peripherals without need a for specific driver.
I suggest you to collaborate as much as possible with Zoltan Balaton.

For native hosts like Power 7/8/9/10 machines, and older Powermacs, could be interesting enable KVM facilities to let code run at full speed without filter of QEMU TCG translator.
Maybe for this task a specific Linux distro with kvm enabled could be a brilliant solution to propose to A-Eon and their Linux specialist guy Christian Zigotzky.

@Hans

Quote:

That is basically what Trevor said in his q&a at Amiwest as well (depends on a1222 sales which we can all guess how that will turn out).

This, coupled with yet another year without progress for Execsg (multicore), OS4 or enhancer. We've hit the end of the road. Mission failed. Emulation can take over the dead platform now.

Let's have a memorial at Amiga 40 and move on on Amiga's with real hardware. Or get ourselves a new hobby

@geennaam

Burial beer, at Amiga40 sounds like good idea.

@All

isn't it just the kernel that is alien (Kernel-based Virtual Machine) that will be emulated so it will bring drivers and api + dual core from linux world, rest will be native os4... the kernel reside in vitual machine on boot. and load os4 from native side like rabbit hole..after kernel is loaded.. or must os4 need to be in the same boat?


or do we we neeed another host computer for it?

In my opinion, AmigaOS 4 is not dead as long as people are using and enjoying it, whether it is on dedicated hardware or through emulation. I'm definitely not going to stop with this hobby just because it is available through QEMU as well.

I'm happy for those who can finally experience it properly without having to pay a huge admission fee. I believe this will bring us new users, who eventually will want to get hold of dedicated hardware. If not, I'm sure some will buy games and software from AmiStore, Alinea, Amedia and so forth.

That AmigaOS 4 is dead has been said for ages now, which is one of the reasons I started writing the AmigaOS 4 Monthly Round-up series. When one looks a bit around, there is plenty of activity, enthusiasm and hopes for what the future can bring. I don't think this will decline with QEMU. I might be wrong of course, this is just my personal view on the situation.

In the end, it is us, the community of AmigaOS 4 enthusiasts, developers and users, that IS the platform. As long as we continue enjoying it, it will continue to live on.

As for the A1222+, yes it is very expensive. A motherboard costs more than the Sam460, but you do get Enhancer, AmigaOS 4 and all the latest graphics drivers included, which costs quite a bit to purchase. If one buys the Sam460, you must pay separately for those (except for AmigaOS 4).

Now, here is an idea (which may be unfeasible, but I'll share it anyhow) I've had pondering this.

One success in the past was the brilliant FlowerPot tool from AmiKit, which made installing and setting up AmigaOS 4 through emulation (WinUAE)on a regular PC very easy.

https://www.youtube.com/watch?v=8cVnw8Txtg4

Here is the link to the site selling it:

https://www.amikit.amiga.sk/store

You will find it a bit down on the page.

I've used it myself several times and it makes it (almost) hassle free to get started with AmigaOS 4.

Maybe such an approach, only with QEMU and the other stuff, could be something A-EON could consider, or possibly get AmiKit on board to make an AmigaOS 4 QEMU-edition of FlowerPot? It would certainly make the threshold for potential new users and developers much smaller. Let us say this FlowerPot version would set everything up for you during the install and you could for example choose which system to emulate and so forth.

That said, I've never used QEMU and I do not know how hard it is to set it up, but shouldn't it be possible through a good script setting all up?

During the install process one could get an option at the end of turning on "Start AmigaOS 4 directly at boot-up every time" or not.

Consider KX Light from Amiga Forever, which starts the environment directly at bootup on the PC, hiding the minimal host OS in the background.

Sorry for long post, but had a lot on my heart concerning this.

The irony of it all is that AmigaOS4 might now """live on""" in a "digital environment" (as it is now endorsed by A-EON). Independant of hardware and host OS. They could call it AmigaDE or amiga Anywhere. (Not sure if these names are already taken since i've only just started reading From Vultures to vampires volume 1.)

Maybe this could ultimately result in a viable business solution. Speaking of which. How is Tao Group doing these days.

@Hans

This is a complex subject.

On one hand it makes sense to follow the path of least resistance, which would be to take advantage of cheap and powerful hardware, and just skip the heavy lifting altogether. That's also fine for newcomers who couldn't care less what the underlying hardware is.

On the other hand, for the existing and mostly original user base, there's a strong preference for native hardware, just because that's how it's always been, change being difficult to embrace for many, which also explains why people still want new Classic HW, 20 years after NG hardware has become available.

When considering cost alone, the first approach is definitely preferable, but there is a middle ground which can benefit everyone including people outside of the Amiga universe, and that would be open source hardware. That approach is working very well in the MCU/IoT world. It allows people with talent to design useful hardware and drivers, while people with enough funds can get the hardware mass-produced, and is typically done with some royalty scheme as it's important to keep the designers/coders on-board to continue improving/expending the product. It works quite well for companies like Sparkfun, Olimex, AdaFruit, and SeeedStudio to name a few.

Maybe that's how Amiga hardware will be produced in the future, I don't know, but I like the idea.

@AlexC

Re open-source hardware: this has already been discussed on the Amigans Discord, and a few people showed interest in helping to set the project in motion. Now, with the latest message "buy the A1222+ or it's the end of OS4 hardware", I'm personally even more inclined to support a community-driven and funded hardware project.

I remember back in the 2000 that people were fighting the emulation by saying that it will kill the Amiga and destroy the community. Only that the opposite happened.
Running Classic Amiga on RaspberryPis and other systems (A500mini, A600GS etc) is based on emulation. The more hardcore amigans would buy original hardware or will make their own.

12-13 years ago, when the X1000 was released, the OS4 ecosystem was quite different. AmigaOS 4 development was pretty active, new hardware was available and companies were working together for a better future. Unfortunately, that is not the case right now.

I am more optimistic about the OS4 future while we embrace the emulation side of things. The work Balaton did with Qemu and the constant support from his side is tremendous. The work Hans is doing with the virtio-gfx drivers is also exceptional.

I do not plan to sell my AmigaOnes hardware for the Qemu, as I like to have the real thing. But I am delighted that I will have it installed on my laptop and use it wherever I want.

I am glad that there is a pretty good option that will help AmigaOS 4 to be tested by, let's say some random numbers, 100 people and 2 of them stick with it and use it as much as possible. These people might come from the Classic camp or the MorphOS camp. Without the Qemu solution, we would never have them. And that's how you increase the user base.

As I said in another thread, we tried the custom hardware for more than a decade and the user base didn't get that big. At least, let's try a different way and see what happens. We have already seen the results, with people joining our community like Maijestro and White, providing useful feedback and experimenting with it. That feels me with a lot of optimism for the future.

@walkero

I still believe that all the talk about new hardware vs. QEMU emulation is moot and pointless unless regular OS4 development gets restarted and users begin to receive updates again. Genuine or emulated, the platform will not keep (let alone attract new) users without signs of life on the OS front.

But I'm driftin off-topic, sorry for that

@trixie
Absolutely. If no more development is done on the OS and on new software, no matter if you have the real or the emulated thing, the result will be to get lost in oblivion

Virtio driver is the only thing they can do "quick". And very important, if you want a portable computer with AOS4.
Especially if you imagine the price of ACube's open PPC laptop. 3000/4000/5000 euros?
Will it hurt real HW sales? IMO, yes. witch will lead to even smaller production runs and further price increases,m
but bring it on. Nothing worst than present situation can happen. Even death is better than coma

@trixie

Fully agreed. I bought 3 classic Amigas lately because its fun and especially because I do not see developement in my preffered platorm, what so ever.
Plus I refuse to spend 900 euros for 10 yeers old mobo and 1200 euros toy
I can play with exact the same AOS4 on my SAM440 (as long as it lives), only slower.

@walkero

Can we get some status update, please. If you know what I mean

@mr2

Quote:

I get what you mean. Dusting off, re-capping my a1200 and getting a pistorm sounds oddly appealing since Amiwest.

@geennaam

Quote:
As far as OS4-related news is concerned, this year's AmiWest was a nightmare come true, yes.

@AlexC

I don't think it's complicated. A lack of OS updates (and massive blunders) have bought us to a place where even Trevor is saying it might not make sense to continue investing in hardware projects. Either we get an influx of new users & developers soon, or it's as dead a platform as geennaam declared.

Also, please don't get too fixated on hardware vs emulation. The first post in this thread being primarily about the possibility of running AmigaOS 4 natively inside a Virtual Machine (VM), using virtio drivers to access hardware devices via the host OS' drivers.


Quote:
Interesting idea. I've bought hardware from all of the companies you listed above. That said, none of that hardware comes close to a new 2+ GHz PowerPC board.

Quote:
That would be nice. Bear in mind that porting AmigaOS 4 to any new hardware requires the cooperation of at least the ExecSG team, if not Hyperion too. Using virtio drivers and a VM would actually make it more feasible to get the OS working on the new hardware. It might even be possible without any changes to ExecSG...

Hans

One comment that I know I'm going to get hate for but I'll make it anyway, is that I don't really get where all these "Expensive hardware" comments are coming from. Yeah, sure, it's not a £500 system which would have been nice, but anyone that has ever built a gaming PC and has been realistic about it will know that PCs never are. For some reason, people seem to expect to put together a "modern" (ok, 10 year old, but still the latest) Amiga system capable of playing all the games and running all the software that OS4 has to offer for the price of some £500 Dell box that wouldn't be capable of running a decent game if it tried.

The only thing I do get is that the pricing of the A1222 has spiralled to the point it doesn't make sense. If you can afford that, then in my opinion, you should stretch just that bit further and get the 5040. But maybe you can't stretch that bit further so ok. And of course there will be people now that just can't stretch to it at all. That is indeed a shame.

Don't get me wrong, you still need to have the money, but I just don't get where all the hate comes from on the cost. I think it's unfounded.

On the topic of emulation (ok, 2 comments then), despite having just bought a 5040 system (which was only about £200 more than my very mid-to-low-end PC) I still welcome emulation and the virtIO drivers if it means I can sit using my MacBook Air in front of the log burner in the winter months working on stuff. The more the merrier!!

@Hans

I see it similarly to some other people here, the availability of new cheap AmigaNG hardware is very poor and of course a bit expensive and will not be able to expand the user base. What remains are the existing users who will buy this hardware again, but we cannot attract new customers for hardware and software.

As already mentioned, a long time ago I owned AmigaNG hardware (AmigaOneXe G4) with the then AmigaOs4.0 as a pre-release version. A lot of things had not yet been ported to PPC and so I sold the machine again after about 2 years. It wasn't until about 18 years later, at the end of 2022, that I started experimenting with WinUae and bought AmigaOs4.1 FE Classic Edition because I was curious about how the system had developed. Unfortunately, I was very disappointed with WinUae because everything was running very slowly on my computer and so I looked for a way to get AmigaOs4.1 running as quickly as possible. When I started, Qemu was mostly unusable until I later contacted Balaton Zoltan through Amiga News and with his help and my testing we were able to improve things significantly in a very short period of time. (6 months).

Qemu was the deciding factor for me to return to the “Amiga world” and I am currently committed to it and of course also prefer real hardware. New users won't invest in an AmigaNG system they don't know just to try out AmigaOs4.1 for a moment. Rather, it will continue to be users who are very familiar with the scene and also know what they are buying.

We have already received feedback from some users that they have even purchased hardware through Qemu, so I don't think this emulation will be the end of AmigaOs4.1 or the end of real hardware. I don't know how long it has been possible to run AmigaOs 68k under WinUae, but here too it has not been possible to "kill" or damage AmigaOs 68k. We know that intensive work is being done on AmigaOs3.x and new ones Projects emerged from this. Why shouldn't something like this also be possible with Qemu?

Qemu with AmigaOs4.1 encouraged me to buy software because it already works well, things like ZitaFTP/Enhancer/ZTools/Emotion/DVPlayer, various AmigaOs4.1 licenses for testing, etc., so I'm not looking for a quick, free solution solution AmigaOs4.1 as I may not have invested in hardware.

I also don't understand the argument that it doesn't make sense to develop software for emulation, but software developed under emulation and running on real hardware makes more sense? It would be much better if all platforms were treated equally and could be operated with software. (software sales)

Finally, I would like to say that I welcome the development of Virtio GPU/KVM and that it could make AmigaOs4.1 even more portable. Using KVM on hardware like (MacPPC) to run AmigaOs4.1 natively would be a good alternative to the current hardware problem. We can't lose anything we haven't tried.

But none of this is of any use to anyone if people aren't shown what AmigaOs4.1 is, how it works and what you can do with it, so I also see that advertising for it will be essential and necessary. And that's why I decided to show things to people on my YouTube channel, they are simple videos and not professionally made, but the interest is still there and some videos have already been viewed over 400 times, videos from October 4th, 2023 (AmigaOne ) XE has already shown an uptrend 117 times.

Basically we all want almost the same thing here, namely to use AmigaOs4.1 because we really like this operating system. So we should do this both on real hardware and with Qemu and help each other realize projects that are meant for us.

What a depressing topic.... i feel sad, and will not accept.

@SkatemanQuote:

+1