@Maijestro

Lol, it is you that is overestimating the situation. Especially when it comes to new users by emulator solutions. And since you will never buy hardware and are just here for a free ride it is of course in your own best interest that this driver gets developed.

Or did I understand you wrong will you now develop software and buy hardware? Afteral you now know what amigaos4.1 tastes like.

If you need advice how to setup your X5k then I'm here to advise you. Can you give me a hint what software you working on?

@trixie

Ah, never heard of pimiga. And isn't that a500mini not just uae with adf loader gui?

Anyways this is comparing apples to oranges. When it comes to classic, you can make uae run on your refrigerator and someone will buy it.

If the direction of os4 is emulation then so be it. I am sure that I can read the retrospective in some future book.
For now three parts From vultures to Vampires are on my menu.

@trixie

Difference here is that the PiStorm still looks like an accelerator and is used in combination with real classic hardware.

Can you say the same for a pc running Linux with qemu?

@Maijestro

That the PPC hardware wasn't done right, considering the size of user base, doesn't mean that it can't be done right. Or do you think that the V4 Apollo hardware was designed and produced by a company like varisys?

Now with €1200 sam460 systems, an entry level could be priced at €500. But with emulator rivaling the top of the line already, you lower that bar even further. To a point that only rpi level prices are enough to make new users buy hardware.

The production batch size of a1222 is about 110 iirc. There are maybe 50 X5040 left. Sam460 is also still for sale despite a small production run. It is safe to say that the hardware base will not grow beyond 1200 in the current situation.

If emulation was the holy grail then there would have been a noticable presence on Amiga38. But I've only counted amikit and cloanto. I couldn't spot any pc running winuae at the stands of those user groups. I bet that 99.99% of its users are just reliving their childhood experience by downloading those adfs. The other 0.01% is using winuae as alternative for cross-compiling.

Amigaos4 wasn't someone's childhood memory. So I honestly don't see any evidence that qemu would do the os4 situation any good. And trying something different from something that hasn't been done right while destroying the last hope for entry level priced hardware doesn't sound appealing to me.

But I'll just lean back and see how it will unfold. I would gladly be proven wrong of course.

Quote:

This depends on how you approach such an endeavor. But that was not the point. The point was if it makes sense at all considering the end result.

@walkero

Quote:

Not sure who the "you" is here. But since I'm reading it, it might as well be me

I'm just being honest here. Like I wrote above. I will probably go for a casual QEMU experience when my X5k breaks down. Maybe even sooner. My personal motivations to develop for OS4 will be fully met with such a solution (access to modern hardware).

In my case, OS4 is not suitable for daily use. I'm also not a software developer. So personally I have no real desire to take OS4 with me on the road at the moment.

Now everybody will have their own personal considerations. And it will surely result in a lot of different opinions and desires.

I'm just curious what the net outcome will be. And how it affects the development of new hardware.

@kas1e

Quote:

Did you check it with recent hardware? Because I can recall that you were using a "5 years old notebook, with 3GHZ one core CPU". And this gives a totally different experience compared to todays 5 GHz+ (single core speed) desktop CPUs. Not to mention the benefit of memory perfomance.

Quote:

True, true. But personally I draw the line when the emulator is faster than the real thing. That's why I left my a1200+bppc behind and went for a sam440 instead.

My questions are not to be confused with criticism on this virtio driver. Emulation is here and cannot be ignored. And if someone sees the opportinuty to make some money then be my guest.

The question that I like to raise is if this will benefit our situation. Or if it will be the final nail to coffin of OS4 hardware.

During Amiga38, there was chatter about the fact that we need lower cost entry level hardware. In the range of $500. The a1222 clearly doesn not fill that gap. Who will give it try now when an emulator rivals the top of the line hardware for virtually free.

At the moment I cannot justify a purchase of the X5040. It will bring me only 10% more raw CPU speed compared to the X5020. (and let's ignore the 30% performance penalty in games for now). This might change if Multi-CPU support is introduced.

If my X5000 breaks down, will I spend $3000+ on an X5040? Or will I spend $xx on a virtio driver? That new PC is needed anyways and I will not settle for less speed like an a1222 or sam460.

@kas1e

Came here to correct my mistake but you were faster

Current QEMU ppc emulation speed is already on par or even exceeding my X5000 on my 3 years old mid level PC.

Make the puzzle complete with virtio and I can put my X5000 in storage next to the sam440, a600 and a1200.
Buy myself the latest and greatest PC. Kids happy because they can play Fortnite during daytime. And I can play around with qemu in the evening.

Even the slow (re)boot is fixed with qemu .

You can wake me up again when os4 is running natively on hardware that exceeds qemu speed.

@joerg

If you want to restrict yourself to windows then that is an excellent idea.

We can justify it to ourselves as something entirely different but both winuae and QEMU are just emulating a machine.

And let's face it. If Toni Wilen (or winuae users) was interested in os4 and put more effort in the emulation of powerup boards, nobody was talking about QEMU and virtio today.


Edit: corrected the winuae maintainers name

@joerg

If you want nonsense then don't get me started about your broken record about pirated stuff and legal action. You know it is just empty threats for years now.

And of course you would need to emulate a complete machine in qemu. Not just the 68k. Just like you need to emulate the pegasos2 marvell Northbridge (and a sm501 currently) as well. Not just the powerpc. But you knew that already.

The point is that if the future is emulation then the powerpc cpu is irrelevant. Better continue with a 10k+ user base then keep on struggling with 500+ users. Don't stop with virtio but port the complete enhancer to 68k. You can count os4 exclusive apps and games with one or two hands. The majority are ports from other platforms anyways.

@flash

And what would this migration look like? A kernel with PPC emulator?

The OS is not owned by Trevor but by Hyperion.

And I have said it many times. Except for Mac Mx hardware, there is no desktop ARM platform. A platform is more then just a SoC. And this means custom hardware.

@Maijestro

Quote:

You mean that there not an Amiga target? Because qemu does support m68k emulation with 68k Mac targets

@joerg

Afaik, Trevor said somewhere that he is only releasing execsg as part of a new os4 release. So if there will be a new os4 release, I do not see why Trevor would not provide a new kernel.

@joerg

Because we get access to existing software. Aros x64 has basically nothing to offer. Not to mention driver status.

And if you need uae anyways to get some applications running why not emulate a 68k directly with qemu. You'll get much better hardware support too.

@joerg

It should be blocked. Fact is that os3.3 was announced on a38. So unlike os4, os3.x is still moving forwards. Legal issues aside. But the point that I am trying to make is that the NG direction becomes pointless when emulation is overtaking NG in terms of performance and access to modern hardware by means of virtual devices. Then why not emulate 68k instead of PPC.

@Maijestro

You are incorrect about the sam460. All PPC hardware currently for sale can use at least HD cards and therefore use warp3dnova. Even an obsolete sam440 can do that with a PCI to PCIe bridge.

The point is that custom amigaos4 hardware cannot keep up with mainstream hardware. So I doubt that custom (ARM?) hardware can leap forward in such a way that makes emulation with virtio drivers struggle and therefore justify a €1000+ price tag. I say custom because there no viable arm SBC available that makes sense at the moment. Not counting those Mx Apples here.
Remember we're not talking about a 10000+ userbase with hardcore fans like classic 68k. We are talking about 500-1000 os4 users here.

So the question remains. What is the point of NG hardware+OS4 when QEMU+virtio can enable the same kind of power to OS3.3 and beyond. It's not like we are swimming in os4 exclusive applications.

@benny

Not with the virtio driver. This is basically a bridge between P96+warp3Dnova to the Host APIs+what er modern Gpu you have in your machine. So while we are limited to RX580 atm. A qemu user gets access to the latest and greatest.

The point of amigaos4 was to break the performance chains of a classic 68k+bus and chipset. Qemu with aos virtio drivers will do a better job in any case.

Hence the question: what is the point of the os4 project anymore when you could do the same for the much larger os3.x user base and have the cpu+gfx power to run both old and new software

Maybe you're confused with vfio. Vfio gives the guest OS access to the host hardware. This means that the guest OS needs to have a hardware driver for each specific vfio hardware. Vfio will normally give the best performance. But it is safe to assume that the Host drivers+apis will not be the bottleneck in this case.

@flash

Missed the whole virtio news on A38. But surprised that it is an official aeon project.

Let's be honest here. A modern PC is superior in processing speed, memory speed, PCIe speed, ssd speed, graphics card support etc. I am convinced that a modern PC with QEMU and virtio drivers will be superior to real hardware. It will run away as well since PCs keep getting faster every year.

I get the point of cross compiling. But what is the point of having such an OS4 environment to develop something for slower real hardware? What is aeon trying to achieve here?
What would be the point of having OS4 at all? Wouldn't it then be better to "reunite" again as qemu os3.2+virtio and move virtually forwards together?

@joerg

Quote:


If Media Toolbox only uses the Scsi mode page commands to request the geometry then we can rule out a driver error.
At least at some point Media Toolbox issues the correct scsi commands:
- read capacity (10) ->opcode 25h
- driver returns logical block address as ffffffffh because the last LBA exceeds 32bit
- read capacity (16) -> opcode 9eh
- driver returns logical block address to the last LBA (64bit value)

The rigid drive geometry parameters page (opcode 04h) is obsolete. So Media Toolbox must calculate its own values for CHS info. And since these add up to 6TB, the correct size has been acquired by Media Toolbox.

Nevertheless, the total number of sectors field is wrong. And this is at least a smoking gun.

@joerg

I now see the red dots where you are referring to. The curious part is that the physical properties are for information only according to Media Toolbox. But they match with a 6TB disk. Could the "for information only" label also mean that Media Toolbox does use total number of sectors instead? This is an unsigned long in struct DriveGeometry. And curiously a large negative value in that screenshot. And the red dots could simply be the result that those positive values are out of range of the negative total number of sectors value?