@amigang


I have had time to think about, what you have written,

it’s of course a good for a games company to make sure they can deliver their games to as many hardware platforms as possible. Having the same API’s on all platforms, can help that. so continuing to support AmigaOS3.2.x, is a good idea.

But to spend huge amount of resource on less capable platform, semes silly. Most of 68K solutions does not offer much atm. And the platform is not even good for developers anymore, old compilers and bad support.

You talk about “YouTube or Spotify» and talk about how they are better on pi400, well this are not Amiga programs, do you want people to write Linux programs instead? Not sure what value it for Amiga users, if the developers stop writing Amiga Programs.

That fact that AmigaOS3.2 is not overload is good thing, it scales with your hardware as you add more junk on top of it, if you have a fast CPU. If you want higher performance system without lots of eye candy you can do that as well. This its a freedom.

Right now you 4 different 680x0 platforms (Real HW, UAE, PiStorm, Vampire)
that can be are hard to consulate.

Vampire hardware, where promote hardware bagging, means no standard API’s. this manes only custom games, that will crash on everything else.
(UAE can be extended to support some of the new registers. But generally, this bad idea, due to any hardware bagging on computers without this hardware, its forced to catch illegal calls using exception handlers.)

You have the A600GS with is ARM library, I do not know how that works, but its not bad idea, to run more on host CPU native, its good way to proceed on a emulated platform, lots stuff can be accelerated, like sound mixing, alpha blending, font rendering, math libraries, matrix calculations, data types, This can be hacked into EUAE easily. Not sure why it was not done before. The only problem I see with it, is that it won’t be scheduled like it should, it will like running the code in a forbid state. this can create something that looks like micro freezes, if its poorly managed. You also perhaps put bitmaps in host memory, bitmap locks can move into UAE space, stuff like that can be done with memory, like OS4 memory objects, MemCopy, icon rendering. (the MemCopy can be using blitter in back ground, so maybe not much improvement, but it can also be its CPU only driven, if so a patch can speed it up, in any case optimized host native copy routine, is probably going to be faster than blitter emulation)

Then you have PI-Strom, that’s does something similar. With exposed native hardware, and 68k driver on the Amiga side, these drivers won’t be easy for Hyperion to write without being creates of the PI-Storm.

Last edited by NutsAboutAmiga on 12-Aug-2024 at 11:13 PM.
Last edited by NutsAboutAmiga on 12-Aug-2024 at 03:20 PM.
Last edited by NutsAboutAmiga on 12-Aug-2024 at 01:49 PM.
Last edited by NutsAboutAmiga on 12-Aug-2024 at 01:33 PM.
Last edited by NutsAboutAmiga on 12-Aug-2024 at 01:29 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@NutsAboutAmiga

Actually right now i develop a game which will
run on lower platforms (i hope 030 but not sure yet).

I develop on os3.2 though i expect it to run on
other os3.x as well.

On which Plattform it runs i decide per project and
this usually has nothing to do with the os but which
speed requirements of the game.

As to os the primary one on which test is done is
Os3.2 (the one OP left out in his list) though
I usually look that at least one tester for 68k
games also tests on caffeine os and i have a
Uae setup for testing on plain os3.1 myselves.

But os3.2 is the „gold standard“ here so to tell.
Not sure what my vampire testers use, i think
OS3.2 (i still have a Vampire hardware with
ApolloOS here too though but i think my
Usual betatesters use os3.2 and/or caffeine os).

@kolla
Quote:

Amikit XE, Then use it, support it, it’s there already and support is WAY better than anything you can expect from Hyperion.

Yes I do, I kinda admit its become my go to Amiga platform now. I have amikit xe setup for a nice desktop serious stuff I want to do. Then I also have Amiga Game Selector 2.5 on another sd card that pretty much allows me to jump into any Amiga game. With the pi400 being a all in one, it east to take around friends house for some Classic Worms session. (i wish it was a full size HDMI port, I forgot the bloody cable one time and no one has mini hdmi cable)

@NutsAboutAmiga
Quote:

You talk about “YouTube or Spotify» and talk about how they are better on pi400, well this are not Amiga programs, do you want people to write Linux programs instead? Not sure what value it for Amiga users, if the developers stop writing Amiga Programs.

I would love a developer to make a modern web browser or apps that can do all the modern day stuff in native Amiga way, I thank every dev we have left in the Amiga world still making programmings or porting stuff to make the platform better. But I came to the kinda sad but I think very realistic conclusion that the Amiga market is just too small now and there not enough devs left to port all the modern day software back to the Amiga platform, plus the main stream market dont put much effort into supporting non stream OS or even web browsers (more webapps / features are only working on Chromium based browser). Even Linux struggle to get support everything. Now you're right I dont want dev to write Linux app or just do port of apps I would love to see more Amiga apps, like Octmed, PPaint and others getting updated is nice to see. Having a system like rabbit hole where you can still access some modern day feature and feels part of the OS, it just really nice to have and means things like a modern web browser are not as important. I can do Amiga coding or work in Hollywood Designer and have Spotify open and listen to music in the background without the need of getting a second computer.

_________________
AmigaNG, YouTube, LeaveReality Studio

@amigang

not possible. pistorm is too slow.
one third of one rpi core is not enough for modern web browser.

@agami

my sam460 is about two times faster than emu68 on my rpi3.
my sam460 has 3d drivers. on my sam460 amiga os 4 works native not on emulator.
so it was worth spended money.

@ppcamiga1

Quote:

ppcamiga1 wrote: @agami my sam460 is about two times faster than emu68 on my rpi3. my sam460 has 3d drivers. on my sam460 amiga os 4 works native not on emulator. so it was worth spended money.

I think you're onto something, can you please explain your math, like the price of one PPC Amiga vs Pistorm. Perhaps then people would finally understand what's in your mind

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

@ppcamiga1

Quote:

ppcamiga1 wrote: @agami my sam460 is about two times faster than emu68 on my rpi3.

Thank you for chiming in with your anecdotal evidence. Just as I was wondering about where I could get some subjectivity inserted into this objective conversation.

In the spirit of subjective reciprocation, emu68 on my Raspberry Pi 4B is 2x faster that emu68 on your rpi3. So that makes it the same speed as your sam460, and it cost me less than half what you spent on the sam460. So definitely much more worth the money spent.

_________________
All the way, with 68k

@NutsAboutAmiga

Quote:

NutsAboutAmiga wrote: But to spend huge amount of resource on less capable platform, semes silly.

It's not about the capability of the hardware itself, it's about markets.
One can either create a market for a product, or create a product for a market.

The 68k+ market has at least 5x the active users of the AmigaOS 4 PPC. So is it more silly to expend whatever resources might be available on a dead-end market of more capable hardware, or on a reinvigorated market of less capable hardware? Keeping in mind that it is not less capable by a huge margin. Because while the hardware upon which AmigaOS 4 runs is somewhat capable, AmigaOS 4 robs the user of capability by limiting full access to that hardware.

So all things being equally silly, there's more money to be made from a larger addressable market.

_________________
All the way, with 68k

agami Quote:

It wouldn't even have to be that severe. A 68k ASIC which cost $25 in mass production would have better price/performance profile for Amiga 68k software than emulating it on the current crop of ARM SBCs. In the absence of such an ASIC, the emulated path provides the community the fastest 68k Amiga experience at a very reasonable price.

A $25 SoC ASIC is quite an ASIC for a RPi level SBC. Maybe the RPi 5 SoC is in that territory. The original RPi and RPi Zero SoC is a single core ARM1176JZF-S@1GHz which is a superscalar in-order CPU somewhat comparable to the 68060 except it has more caches.

68060 8kiB-I 8kiB-D 1.80DMIPS/MHz
ARM1176JZF-S L1: 16kiB-I 16kiB-D L2: 128kiB 1.25DMIPS/MHz

The SoC also has I/O and a GPU which is more advanced and modern than AGA. The RPi Zero SBC has sold for as little as $5 USD so the SoC is very cheap. The RPi Foundation RP2040 SoC with dual 133 MHz Cortex-M0+ CPU cores, cheaper I/O and no GPU but SRAM for memory can be bought for less than $1 USD in quantity and still uses more transistors than a 68060+AA+. Some people will say the RPi Foundation prices are subsidized but the following post says otherwise.

https://forums.raspberrypi.com/viewtopic.php?f=63&t=9181 Quote:

As has been said on many occasions the cost to build the Raspberry Pi is not subsidised by anyone, the cost to make the board is $25 - X & $35 - X where X is some value that only members of the foundation (& the two manufacturers) know the actual number.

A mass produced $25 ASIC SoC could be wow but lower production drives up the cost as talked about in the interesting thread above. Even if lower production drove up the cost to $25, a 68060@1GHz with somewhat upgraded caches should outperform even the RPi 5 for Amiga purposes and not require a fat heat sink or several times the amount of memory which increase the system cost.

agami Quote:

All platforms have inconsistent performance. Even the leading gaming consoles have adopted a platform approach, where a game can perform differently based on the specific SKU of the user. The latest versions of Windows and macOS run on a variety of configurations: From 4 cores to 32, 5GHz, from 8GB RAM to 1TB, DDR3 to DDR5, spinning rust to PCIe 5 NVMe, and so on. But there is still software on these platforms specifying a minimum spec which excludes many people operating at the mid-to-low end.

Caches, deeper pipelines and MMU usage cause jitter and increased latencies while JIT emulation adds a whole new cumulative layer. Worst case performance can be a problem for embedded use which is why these basic features are often missing in embedded CPUs (MCUs often use SRAM instead of caches, 8-stage in-order CPUs are the standard instead of 10+ stage OoO CPUs, smaller CPUs often have a MPU instead of MMU, etc.). It is possible to create a CPU that is appealing for embedded use. RISC-V SiFive U74 SoCs are a good example.

RISC-V SiFive U74 SoC "embedded" features
o In-order 8-stage CPU cores are practical for embedded use
o branch mis-predict penalty is only 4-6 cycles for an 8-stage pipeline
o load-to-use stalls are mostly eliminated
o some caches including 2MiB L2 can be turned into SRAM memory for use as a MCU
o supports RISC-V compressed encoding for good RISC code density
o embedded oriented S7 core included with 4xU7 cores

Reduced stalls improve performance, jitter and ease of programming, especially instruction scheduling. Deep pipelining isn't necessary for good performance anymore and mostly in-order CPU cores save the transistor, power and financial budget for the GPU. The in-order U74 core is simple and somewhat primitive yet it has near OoO PPC G5 performance and with a few upgrades likely could outperform every PPC core ever made. The pipeline design is a CISC like pipeline design similar to the 68060 but the 68k ISA has significantly more performance potential using it.

agami Quote:

Even without instruction revisions/extensions or vector/tensor units in the emulation soft cores, for which binaries would see optimisation benefits, there is nothing wrong with the Amiga 68k platform having software requiring a minimum spec of 68k+.

Sure. It could simplify the requirement list of software. I'm not against it but it doesn't accomplish much. If divided Amiga camps could be convinced to agree on standards then more could be accomplished but good luck with that. With emulation, any performance gains would be minimal and feature improvements are much different for emulation and FPGA cores. Also, development tools and OSs usually don't support emulation and FPGA features. It's an uphill battle.

agami Quote:

That was @amigang's original point. Amiga OS 3.2.x is not categorised as 68k+ software as it is intended for all Amiga hardware capable of running Amiga OS 3.x. I simply specified a scenario, however unlikely, outlining what Hyperion would need to do to have an OS in the 68k+ rubric. Which would cost a lot of money, which Hyperion do not have, which is why it follows the "if" condition. In the real world, if Hyperion did have the kind of money that would be needed to create a hybrid Amiga OS 3/4 for 68k+, they would sooner spend it on fighting law suits. Such is their nature.

For a more modern AmigaOS to be successful, a larger unified Amiga market is needed to encourage developers and spread out development costs. A larger unified Amiga market requires significantly better value hardware from one source. Also, uncertainty created by IP related ownership questions needs to be removed.

agami Quote:

AROS is in no danger of fostering a distro landscape akin to Linux. We might end up with three or four, which is more BSD territory. In summary: - Fast and compatible 68k ASIC = good (but when?) - Fast and compatible 68k soft core = good (because available now) - 68k software (including OS) which takes advantage of above = very much welcomed

No leadership and uncertainty are project killers because they are investment killers. We are mere peasants in Amiga Neverland with 3 kings. One king says let them eat PPC cakes forever, another king lays claim to everything but is an illegitimate and bankrupt fraud owning almost nothing and the last king owns most of the kingdom but spends most of his time fighting off squatters, including the first two kings.

agami Quote:

It's not about the capability of the hardware itself, it's about markets. One can either create a market for a product, or create a product for a market. The 68k+ market has at least 5x the active users of the AmigaOS 4 PPC. So is it more silly to expend whatever resources might be available on a dead-end market of more capable hardware, or on a reinvigorated market of less capable hardware? Keeping in mind that it is not less capable by a huge margin. Because while the hardware upon which AmigaOS 4 runs is somewhat capable, AmigaOS 4 robs the user of capability by limiting full access to that hardware. So all things being equally silly, there's more money to be made from a larger addressable market.

Too much marketing logic when the peasants are supposed to eat expensive PPC cakes.

Last edited by matthey on 13-Aug-2024 at 03:34 AM.

@agami

Some games are just meaningless to port to 68K+ platform, it does not interest me if got it for free or not. I like to see DOOM3 running 68K, LOL…. in your deams perhaps... But please wake up and get back to reality. your hardware can't do it. it will be like trying to pulling timber with a ford escort.

Quote:

least 5x the active users

does not matter, it too slow for DOOM3, will never run, does not matter how many users it has.
what are you going to do, sum up all the fps from all the users?

Can your 68K+ run Blender?
how fast can you encode mp3's ogg files or do any real work.

Last edited by NutsAboutAmiga on 13-Aug-2024 at 11:45 AM.
Last edited by NutsAboutAmiga on 13-Aug-2024 at 11:30 AM.
Last edited by NutsAboutAmiga on 13-Aug-2024 at 11:29 AM.
Last edited by NutsAboutAmiga on 13-Aug-2024 at 11:23 AM.
Last edited by NutsAboutAmiga on 13-Aug-2024 at 11:18 AM.
Last edited by NutsAboutAmiga on 13-Aug-2024 at 11:15 AM.
Last edited by NutsAboutAmiga on 13-Aug-2024 at 11:14 AM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@agami

Quote:

The 68k+ market has at least 5x the active users of the AmigaOS 4 PPC

How do you define market? I would not be surprised if the number of OS4 users willing to pay (and pay more) for OS4 software is higher than number of OS3 users willing to pay for OS3 software.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

I agree that is about the software, less so for actual games..

Given that Hyperion can't even get multiprocessor on the X5000 sorted after 5 years.

Given that to make a productivity appset useful, it will take years and years of coding and OS changes to match, I highly doubt this would be feasible given what we have seen. No sane person would want to start this today for the hopes of a tomorrow.

Final Writer was a very nice surprise on the A600GS, but email, web browser, and other productivity applications needed in 2024 are just non existent. For example, Iris has been being constantly developed for call it 5 years and it has very useful features in the last few releases, don't get me wrong it is very capable email program that I use daily, but things take time and determination to take on projects of that magnitude. I still would like to see features in it, and very big kudos for Jacek for doing so much that I have asked for, but it was an enourmous effort that would be a hard ask. The hit and run we are used to seeing where one or maybe two version we see created to port or create something is more the norm for large projects.

An even if all the above came true, it would have be on a new platform to be truly usable. If all the development happened above, I think it would be likely to get a bounty to move it somewhere.

NutsAboutAmiga Quote:

Some games are just meaningless to port to 68K+ platform, it does not interest me if got it for free or not. I like to see DOOM3 running 68K, LOL…. in your dreams perhaps... But please wake up and get back to reality. your hardware can't do it. it will be like trying to pulling timber with a ford escort.

Doom 3 plays ok on a $55 USD 4GiB RPi 4 (640x480 Ultra quality).

Doom 3 on Raspberry Pi 4 (gameplay)
https://www.youtube.com/watch?v=lJHLFG9sig0

Doom 3 would likely work on a cheaper 2GiB model. The description for the video states, "a small GPU overclock" was used and "the CPU usage is very low" so performance is likely GPU limited. The RPI 4 and 5 have weak low power GPUs compared to the CPUs. This game doesn't need much CPU performance though. WinUAE on high end x86-64 hardware may be able to play it. PiStorm/Emu68 with a RPi 5 and Amiga GPU driver for RPi GPU may come close but Doom 3 uses shaders which the 68k Amiga has no API for. A 68k Warp3D Nova port would give the 68k Amiga the more modern 3D API although the PiStorm/Emu68 setup may be a more difficult target than WinUAE. The easiest target would be real 68k Amiga hardware like the RPi where ~$50 USD hardware can play it.

Modern hardware is cheap when mass produced. The $4 USD RPi Pico using the RPi Foundation's organically produced $1 USD RP2040 ASIC SoC plays the original Doom better than most Amigas.

Multiplayer DOOM on a $4 Raspberry Pi Pico!
https://www.youtube.com/watch?v=vXr7tOR3dis

Doom handheld using RPi Pico for about 10 Euros
https://www.youtube.com/watch?v=V7MJyu1sFCM

Why can't the 68k Amiga have cheap toys that can be used as building blocks like the RPi? A king in Amiga Neverland dictates that we must eat expensive PPC cakes forever!

agami Quote:

The 68k+ market has at least 5x the active users of the AmigaOS 4 PPC.

NutsAboutAmiga Quote:

does not matter, it too slow for DOOM3, will never run, does not matter how many users it has. what are you going to do, sum up all the fps from all the users?

No! It matters a huge amount that there is a large existing 68k Amiga market. It means mass production may be possible allowing prices competitive with RPi hardware. THEA500 Mini success supports this idea. Compared to the tiny market size of PPC Amiga1 hardware after 2 decades of failure, the 68k Amiga market is incomparably better. No sane person would continue the PPC Amiga1 failure. PPC AmigaOS 4 development was stopped for this reason and development pivoted to the 68k AmigaOS market. The illegitimate king is crooked but not insane. Agami knows marketing and has been right on point with his analysis.

kolla Quote:

How do you define market? I would not be surprised if the number of OS4 users willing to pay (and pay more) for OS4 software is higher than number of OS3 users willing to pay for OS3 software.

Software is an investment into purchased hardware in the short term and the platform in the long term. Users stop investing in EOL and dead platforms to save their money for their next purchase of the replacement. The Amiga is in a death spiral and emulation is EOL.

Last edited by matthey on 13-Aug-2024 at 06:53 PM.
Last edited by matthey on 13-Aug-2024 at 06:53 PM.

@matthey

No, you are not showing the 68K version of DOOM3, your showing the Linux version.
if want use Linux, then use Linux, don't pretend your a Amiga user.

Last edited by NutsAboutAmiga on 13-Aug-2024 at 07:50 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

NutsAboutAmiga Quote:

No, you are not showing the 68K version of DOOM3, your showing the Linux version. if want use Linux, then use Linux, don't pretend your a Amiga user.

Only Amiga Raspberry Pi makes it possible. RIP Amiga.

@NutsAboutAmiga

People have use macos on Amiga, if there was a way to run doom 3 seamlessly would it really matter?

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

pixie Quote:

People have use macos on Amiga, if there was a way to run doom 3 seamlessly would it really matter?

There is an important difference between Mac on the Amiga and Amiga on other hardware. The Amiga offered better value than the Mac so it made sense to buy an Amiga instead and emulate the Mac resulting in +1 Amiga sold -1 Mac sold. When an Amiga is emulated it is -1 Amiga +1 other computer. Trevor is responsible for tens of thousands if not hundred of thousands of computer sales with the PPC Amiga1 value. The problem is that they are ARM and x86-64 hardware using Linux and Windows. The developer support goes to the platforms with the largest user bases and Trevor is shrinking the Amiga user base while selling the competitions computers for them.

@kolla

Quote:

kolla wrote: @agami How do you define market? I would not be surprised if the number of OS4 users willing to pay (and pay more) for OS4 software is higher than number of OS3 users willing to pay for OS3 software.

The Venn diagram intersection of demographic and budget. It's a group of people with discretionary income sharing a common (bell curve distribution) interest in a product/service.

The amount of discretionary income individuals are prepared to spend on a given product/service is based on utility and relative value. I've touched on this before but I'll cover it here again:

If I spend $500+ on a game console, or $1,000+ on a gaming PC, then I will have less issue spending $50-$100 on a modern AA or AAA game.
Whereas for a Playdate at $199 it makes sense to spend $5 on game, but not $50. That's comparative value.

But, if there were magically a game or app that provided unique utility, sometimes known as a "killer app", that turned a Playdate into a very useful tool capable of solving $50 problems, then spending $50 on an app for a $199 console makes sense. Thus high utility influences the comparative value dynamic.

Most Amiga users in the '90s would be familiar with this dynamic because that's how high priced accelerator cards were marketed to us at the time. You could sell or mothball your treasured Amiga, or for several hundred dollars you could expand the utility of your original purchase. That's how I ended up using my original A1200 all the way into 2001.

This is the dynamic which is clear on AmigaOS 4. It is a relatively new platform with a small software library. After spending $1,000+ on an AmigaOS 4 system, it makes sense to spend upwards of $100 for the rare software which gets released, because of both comparative value and utility.

For the past 5 years, as we've seen a renaissance of 68k in the Amiga market, the main focus has initially been on compatibility with existing hardware (Apollo accelerator boards, PiStorm) and compatibility with the vast back-catalogue of Amiga 68k software (AmigaOS 3.2, AROS 68k, A500 mini, A600GS). This focus creates minimal new utility.

Recently however, now that that this market has grown, the focus has shifted to leveraging the performance with new utility which we have not seen on the 68k platform to date. Running Quake at respectable frame rates, new games and media software designed for the 080 + SAGA, game ports from open sourced projects like Diablo and C&C, and ARM libraries for media datatypes in A600GS. If there is software that provides enough novel utility for 68k+ users, we will see them spending just as much money as AmigaOS 4 users do for their software.

Last edited by agami on 14-Aug-2024 at 04:32 AM.

_________________
All the way, with 68k

@agami

Great points apu.

I would say that the "killer app" in 2024 is just being able to use the software for modern uses.

The Amiga market is a tough sell imho, and yes people can make money such as the 600GS, but the big splash days are gone. As we amiga guys and gals get older and start dying off or like myself, I don't have much interest in classic. I went back to 3.9 with what little time I spend, since there is no real value for going to 3.22 it was more of a headache for me. I use MorphOS for work every single day and it works brilliantly.

Based on my definition we have killer apps and I use them everyday... Wayfarer, Iris, PolyOrga to name a few. They are rare in that they do so very much. In Iris I send out invites our daily with the integrated contacts and the .ics functionality built in. You can do stuff with it, which to me works.

The MorphOS model works in that for free to just a few hundred dollars and I get a killer apps absolutely free running at much faster than needed speeds (minus Wayfarer with large sites). The coders have many years of constant development into the apps. Where many app coders add core OS items or MUI items/updates. So you need the OS to be available as well. That is the challenge that all the other Amiga flavors haven't been able to solve, and it is the most vital component. Modern programs need constant improvements to be relevant. Even keeping Iris and Wayfarer on the latest webkit is a huge task in itself, not including all of the improvements added. I don't see anyone that committed to doing it on other flavors for various reasons.

If you watch Chris Edwards latest videos, you see him setting up multi-authenication for google mail using Iris and Wayfarer. He also watches awesome Encore demoes while having wayfarer up on this other screen with his youtube page. All that equals killer app to me at least.

The problem is that most MorphOS users just quietly donate to developers and use it daily without posting a single post about it.

@Acill will be at Amiwest this year with his recently updated PCIe Mac (put a 2.5GHz Processor in his 2.0). He also is raffling off another PowerMac with a MorphOS license...

@Matt3k

Quote:

The problem is that most MorphOS users just quietly donate to developers and use it daily without posting a single post about it.

If only we had our own mosamiga1

Last edited by pixie on 14-Aug-2024 at 01:30 PM.

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga