@agami

Quote:

Those that advocate that this or that cheap ARM based solution does the job, the job they're talking about is yet another band-aid on the aging and shrinking Amiga retro HW scene. They're approaching it from a personal aspect. The "Be happy with what you have" crowd.

It is not a band-aid. It is reality. And to understand that this is the reality you can just look at all other retro platforms.
Just like like all other retro fans, amiga retro fans can be split into two deparate markets,

1, one group that collects and uses the original hardware/systems from back in the day. These are the collectors, they will collect original CBM Amiga hardware and they will never consider buying your so called resurrected M68k because it is not genuine CBM hardware and thus they are not interested. Just like a stamp collector is completely 100% not interested in your coin collection.

2, another group, quite a bit larger, that don't care about the hardware and are perfectly happy buying a SuperNintendoClassic. They happy about emulation and just want to relive and replay games from back then. You tell them "but it is emulation running on a cheap ARM, it is not the original Ricoh 5A22!!!" and they will just look confused at you and wonder why they should care? They just want to play the games. It is emulation, so what, it plays the games.

2b, a subgroup of 2 buy more expensive systems, like MISTER and friends so they can get extra features and emulate multiple systems. But they are still part of the "emulation is fine, I just want to play the games".


Why is amiga retro market different than this?
It is not. There are largely two groups. And the same kind of groups. Just like every other retro platform.

Again, where is the market demand for a new resurrected m68k? Why should the industry use it instead of continuing to use the existing ISAs? That is a really important question before you pour money into it. WHY should the industry buy volumes of this hypothetical m68k, what problems does it solve that current ISAs can not solve?


You are basically looking for a market of "collectors" that really want a chip running the m68k ISA that is not emulation and not a FPGA because having M68k ISA in HW is important to them. But at the same time, these collectors will still be ok with running new HW that was not built by CBM. I do not think that market exists.

EDIT: In your next post you mention you want a modern m68k that is fast enough so that you can run modern, demanding applications for the modern age. That sounds like you want an X1000 but it should have a new fast m68k instead of the PPC. Please don't put money into this. There is no market for this product.

Last edited by V8 on 21-Sep-2022 at 05:28 AM.
Last edited by V8 on 21-Sep-2022 at 05:17 AM.
Last edited by V8 on 21-Sep-2022 at 05:16 AM.

@SHADES

Thanks for picking small lines from my comprehensive response and addressing them as if in a vacuum.

There are two major cultural trends in what remains of active and semi-active users of this forum:
- Those clearly interested in continuing to use software (mostly old) which was written for the tragic original Amiga platform. They might do so by maintaining and upgrading original Amiga hardware, or via emulation on contemporary hardware.
- Those who are interested in forging a new path for a more modern incarnation of the Amiga vibe, with different attempts at rejuvenating the Amiga scene e.g. AmigaOS 4 (System54) on expensive Power ISA HW, MorphOS on old PowerPC Mac HW and then where it goes next, AROS on x86 and perhaps then AxRuntime on LInux.

These two cultural trends overlap only where the modern incarnations insist on including some sort of backward compatibility/built-in emulator to run the SW written for the original Amiga architecture. Otherwise:
- The former is happy and content with UAE and promising solutions such Emu68 that allow them to re-experience a long past computing and gaming paradigm, and are generally very price sensitive.
- The later is not content with re-playing some classic game for the nth time, or using productivity software which is generally more painful to use than most modern counterparts on other platforms, be it because their limited feature sets are products of their time, or they lack support for modern standards, and more often than not are prepared to pay extra to get a non-Windows and non-macOS option to reinforce their computing identity.

Original 68k HW is slow when compared to contemporary options, at almost any price. But it's good enough to run all the old SW written for the original Amiga architecture. UAE and FPGA solutions don't need to run at 1GHz+. 060-level performance with a bit extra oomph satisfies 99% of the former group.

New software and computing paradigms require more performance. Even 1GHz is limiting for many of today's standards, programming languages, libraries, frameworks. 2GHz+ with multiple cores/threads is the norm for today's multimedia-rich computing paradigm. Most developers want to work with higher performance platforms.

This hypothetical, and naturally very expensive 68k ASIC project could bridge both cultures. That's how it could possibly help this platform.

And while I am middle-aged, I am far away from being senile. In 2010/11, at the height of X1000 pre-launch fever, the enthusiasts of the later culture drowned out the enthusiasts of the former. Some of the former might even have gotten swept up in the gleeful frenzy at the prospect of a "NEW Amiga" and didn't question the "hi-end" specs of the planned X1000, never mind the price tag.

I know that with the decade+ time since then, and more than a handful of disappointments in projects of the later culture, that the voice of the former is much more prominent. Especially since over the past few years there has been new developments and a successful initiative in the production and release of The A500 Mini.

A person can participate in both cultural trends. I, like many others, do have a fondness for the original Amiga architecture and run emulation solutions and have purchased The A500 Mini. But they occupy less than 1% of my annual computing time.
On the other hand I am more interested in what can be done in the future-focused, higher-performance, new takes on the original Amiga concept arena. Mostly because I am unhappy with the greater part of the 99% of my remaining annual computing time on the contemporary platforms of Linux, Windows, macOS, iOS, etc.

_________________
All the way, with 68k

@V8

Quote:

Again, where is the market demand for a new resurrected m68k? Why should the industry use it instead of continuing to use the existing ISAs? And "because this would make a turbo-amiga-m68k asic available for the 25 people that care" is not a valid reason.

Hmm. I largely agree with all of what you said previously but......
I'd love to see like a Pi-xx revision plugged into a cheap expansion backplane/mainboard, running Amiga OSxxx-ARM (pipe-dream?) that I can expand with PCIe and USB etc etc in a nice A500 or A1000 case option. Pi3 was upgradeable in connection to Pi4 compute. Standards make it a lot easier to upgrade. A bit like the AMIGA 4000 CPU slot had in changeability.

Why not offer both? Same footprint board, different case for different oppertunities (add-on cards) such as:
- AGA (that can play nice with the onboard 4k Mali)
- 16 bit multichannel whatever I want USB audio interface (I currently have a Tascam Celsonic 20x20 on USB-3/2)
- Plethora of storage options and WiFi dongles etc (USB)
- Really good graphics resolutions with 4K decode that could be used.
- How about a weird chameleon type plugin board on the PCIe for a FPGA Mister board, that allows you to flash for different systems to entertain all the retro gamer clan?

I mean, there's just so much to create once you have the I/O sorted and the system running with access to it all.
Is there space for a new OS on ARM? Sure there is and by bringing down the cost to get into it all, having a standard that doesn't need to be re-written for every single manufacture's chipset, keeps development code cost/times down and the OS starts to gain features and users because of it all.
Maybe it becomes the Retro emulation OS of choice? There's a niche.

_________________
It's not the question that's the problem, it's the problem that's the question.

@SHADES

Quote:

Hmm. I largely agree with all of what you said previously but...... I'd love to see like a Pi-xx revision plugged into a cheap expansion backplane/mainboard, running Amiga OSxxx-ARM (pipe-dream?) that I can expand with PCIe and USB etc etc in a nice A500 or A1000 case option. Pi3 was upgradeable in connection to Pi4 compute. Standards make it a lot easier to upgrade. A bit like the AMIGA 4000 CPU slot had in changeability.

I understand and can agree with this. But now we are talking about hobby projects becuase it is fun, not major commercial industry-wide re-launch of an old ISA.

I 100% understand that it is fund and cool to make expansions like you describe. 5-6 years ago I personally helped a guy getting network file-system support into the ESP32 he put on the expansion board in his TRS-80, so by running a small program on the TRS-80 it could direct it to get the ESP32 to download and write into memory images of games he had stored on a windows fileserver and on Azure cloud. He basically built an expansion board for his TRS-80 to be able to download and run games from his own cloud-store. An SMB3 offload/accelerator board for TRS-80 if you like.


And that was a really fun and cool project, but no eventhough TRS-80 can now access Azure cloud file-server, this does not mean that TRS-80 will become re-surrected as a platform. sometimes you do things just because they are fun.

Last edited by V8 on 21-Sep-2022 at 05:43 AM.
Last edited by V8 on 21-Sep-2022 at 05:39 AM.
Last edited by V8 on 21-Sep-2022 at 05:37 AM.

@agami

Quote:

Thanks for picking small lines from my comprehensive response and addressing them as if in a vacuum.

I'm not going to bother with that other than to say, none of it matters when you are looking at costs for a new 68K ASIC that outweigh even producing the current A-Eon PPC boards. That's without porting.

Quote:

There are two major cultural trends in what remains of active and semi-active users of this forum: - Those clearly interested in continuing to use software (mostly old) which was written for the tragic original Amiga platform. They might do so by maintaining and upgrading original Amiga hardware, or via emulation on contemporary hardware. - Those who are interested in forging a new path for a more modern incarnation of the Amiga vibe, with different attempts at rejuvenating the Amiga scene e.g. AmigaOS 4 (System54) on expensive Power ISA HW, MorphOS on old PowerPC Mac HW and then where it goes next, AROS on x86 and perhaps then AxRuntime on LInux.

1.- No new users. Slow/non-development of anything other than keeping the status quo. 68K new? not needed.

2.- Possible new users, if it's not at stupid A-Eon prices. Possibility for OS development for newer features etc, again, depending on buy-in

Quote:

New software and computing paradigms require more performance. Even 1GHz is limiting for many of today's standards, programming languages, libraries, frameworks. 2GHz+ with multiple cores/threads is the norm for today's multimedia-rich computing paradigm. Most developers want to work with higher performance platforms.

As already started, this can be had at prices that start from as little as $50

Quote:

This hypothetical, and naturally very expensive 68k ASIC project could bridge both cultures. That's how it could possibly help this platform.

Well, your assumption hasn't got past your first statement of cost yet, and for what? certainly not group one "continuing to use software (mostly old)" There's zero need there.

Quote:

Some of the former might even have gotten swept up in the gleeful frenzy at the prospect of a "NEW Amiga" and didn't question the "hi-end" specs of the planned X1000, never mind the price tag.

Care to wager on how many? or how many NEW users it brought in?
Perhaps on how it sped up OS development for all these "new" users that spare no expense.
It hasn't helped the platform. Fewer people than before use it due to cost let alone, incompatibility of using PPC.

Quote:

I know that with the decade+ time since then, and more than a handful of disappointments in projects of the later culture, that the voice of the former is much more prominent. Especially since over the past few years there has been new developments and a successful initiative in the production and release of The A500 Mini.

Huge difference in cost, doubt it made many more NEW users. Too limited, still expensive for what you get.

Quote:

On the other hand I am more interested in what can be done in the future-focused, higher-performance, new takes on the original Amiga concept arena. Mostly because I am unhappy with the greater part of the 99% of my remaining annual computing time on the contemporary platforms of Linux, Windows, macOS, iOS, etc.

Making more insane costly investment like re-designing a modern 68K at millions of $$ in expense is not going to achieve that.
There is a lot of very good hardware, cheap and fast, multi GHz stuff, right now, available, that would be easy to support with standards that are compatible within generational changes.
Just look at what's being done with Pi-Storm etc. It is clear that with the right approach, AMIGA can be experienced and developed without huge demands on people's earning potential. Even for a modern experience.

_________________
It's not the question that's the problem, it's the problem that's the question.

@V8

Quote:

I understand and can agree with this. But now we are talking about hobby projects becuase it is fun, not major commercial industry-wide re-launch of an old ISA.

The ISA is done. We are using it, it's just not 68K native.
Yep, I don't think the 68K new ISA is going to get off the ground. The cost is just....not happening.

There are better alternatives but, again, the ARM big-endian if you like, does offer a way forward. Modern, I/O, lots of hobby projects of you like, and, a cheap way to get OS interest and new features running.
A-EON could do an AGA PCIe plugin board for retro. Make the back-plane as well? why not? Other Amiga centric whatevers.
Offer different case options for power users (more PCIe lanes for addons or USB options)
It just goes on and on but the platform has to get away from all this expensive obscure stuff.

_________________
It's not the question that's the problem, it's the problem that's the question.

@matthey

Quote:

THEA500 Mini would likely have sold several times what it did had it been half the cost.

Yes i agree and Im pretty sure that its not the hardware that made this a expansive product. Loads of other min computers are a lot cheaper, hell even Retro Games Ltd first product, the C64 mini can be bought for £30 now. That features a Allwinner A20 CPU that could run A500 emulation.

Plus let me just make this point emulation doesn't has to stop at 68000 platform, look at what happened when WinUAE got PPC Emulation added, the Classic OS4 edition sold out. Rabbit Hole show how the AmigaOS can communicate with the host OS to run applications.

Im for a version of AmigaOS that is more aware its on a emulated platform so that things like Ram limits can be gone, maybe new graphics driver could be written to take advantage of the advance features, just like Vampire / Apollo team are doing with AMMX.

_________________
AmigaNG, YouTube, LeaveReality Studio

@Shades and V8
You guys are hung up on the cost of developing the 68k and an ASIC. I found potential embedded partners to help with development of an ASIC which they specialize in and another that has the economies of scale building out IoT infrastructure for the cost of an ASIC to be a cost reduction. One big embedded customer can provide the economies of scales needed. I found businesses that still preferred the 68k to ARM and RISC-V. Yes, the 68k would have catching up to do and yes, those businesses had requirements for what they wanted and needed. Personally, I think you are wrong about demand for affordable high quality retro and hobby hardware but I like to hedge my bets by creating hardware that is appealing to embedded markets. The RPi misses much of the retro market yet it has been a huge success in part because they hit the embedded market. The Raspberry Pi Foundation is nice enough to give us their strategy.

https://www.arm.com/blogs/blueprint/raspberry-pi-rp2040 Quote:

Designing a sub-$5 computer involves several tough decisions. Decisions such as which instruction set architecture (ISA) to use and which class of chip IP to select from within that ecosystem in order to balance price, power and performance. Then there are design and manufacturing questions: who’s going to design the system on chip (SoC) and who’s going to fabricate it? All of these considerations dictate how useful and affordable your final product is. So when it came to selecting a chip to power our first microcontroller-class product, Raspberry Pi Pico, we knew that we needed to push that price-performance ratio harder than ever before.

The choice is for the Amiga to become emulation on RPi hardware supporting and building their platform or to build an Amiga platform using their strategy taking advantage of Amiga strengths and assets they don't have like the tiny footprint of the standard 68k AmigaOS, a minimal logic DMA based standard chipset and a large software base for such a small platform including many retro games. If Amiga assets and resources were combined, we would be starting with more than the Raspberry Pi Foundation had to start with.

amigang Quote:

Yes i agree and Im pretty sure that its not the hardware that made this a expansive product. Loads of other min computers are a lot cheaper, hell even Retro Games Ltd first product, the C64 mini can be bought for £30 now. That features a Allwinner A20 CPU that could run A500 emulation.

I don't think Retro Games Ltd. was able to reduce the production cost of THEA500 Mini as much as they would have liked (they wanted to use a custom Amiga SoC ASIC). I believe they have enough margin to lower the price down to near half of the introductory price and I would expect to see substantial discounts for the holidays when they may sell several times the number of units from Thanksgiving to Christmas than they have since release. They don't need to push down the price as aggressively as a business trying to build a platform. This is a one and done quick to market eye candy product which they will sell as long as there is enough demand at a profitable selling price.

amigang Quote:

Plus let me just make this point emulation doesn't has to stop at 68000 platform, look at what happened when WinUAE got PPC Emulation added, the Classic OS4 edition sold out. Rabbit Hole show how the AmigaOS can communicate with the host OS to run applications.

Did WinUAE PPC emulation support really help the NG AmigaOS 4 market though? Does emulating PPC provide a better NG experience than real NG hardware or is it more of a marketing tool to encourage people to try AmigaOS 4 first while selling a few more copies of the OS in the process?

amigang Quote:

Im for a version of AmigaOS that is more aware its on a emulated platform so that things like Ram limits can be gone, maybe new graphics driver could be written to take advantage of the advance features, just like Vampire / Apollo team are doing with AMMX.

There are no memory emulation limits other than that of an Amiga with 68020+ CPU which is ~2GiB of main memory. Some chipset limits can even be removed like for chip memory. The biggest issue is finding the most compatible place to map the memory. Each emulator may have artificial limits but that is all they are. There are already RTG graphics drivers for emulators which take advantage of RTG features of their host hardware. Optimizing for emulation is mostly pointless as the hardware and timings vary too much.

Last edited by matthey on 21-Sep-2022 at 07:40 PM.
Last edited by matthey on 21-Sep-2022 at 07:38 PM.
Last edited by matthey on 21-Sep-2022 at 07:33 PM.

@amigang

Quote:

https://www.youtube.com/watch?v=ZED8_Lfdwuk So at the recent NWAG meeting David Pleasance (Former Commodore UK Boss) purpose a new project.

At the end of the video David Pleasance asked for the "thoughts" of the attendees. Is there a continuing part of the video?

@matthey

Quote:

Did WinUAE PPC emulation support really help the NG AmigaOS 4 market though? Does emulating PPC provide a better NG experience than real NG hardware or is it more of a marketing tool to encourage people to try AmigaOS 4 first while selling a few more copies of the OS in the process?

AmigaOS 4 on WinUAE is horribly slow and vastly inferior to real hardware. I think it's partially due to the emulated graphics cards being old and not supporting compositing, but the PPC emulation is probably also slower than real hardware. Anything Warp3D Nova related won't work either.

The same goes for QEMU based NG emulation. It can't match real hardware at all. Not even close.

I agree with Shades regarding a 68K ASIC. A "custom" ASIC which is basically putting a set of predesigned ASIC cores & peripherals in one SoC is relatively easy. The more custom you go, the higher the expense. Adapting a core designed for FPGA to ASIC is more expensive. That's ignoring the work needed to interface the core with more modern peripherals or an existing GPU. Designing your own CPU and/or GPU cores and writing GPU drivers would increase the work and cost even more (a lot). The more custom you go, the more customers you'll need to pay for it all.

I'd go the other way, and try to leverage other people's work as much as possible. Switching to ARM or some other CPU architecture allows you to take advantage of the work done by thousands of engineers without having to pay for it all out of our own pockets. Likewise, it's best to switch to little-endian. Fewer and fewer GPUs and peripherals are bi-endian, and less and less work is being done to support big-endian CPUs in drivers. Hence, staying big-endian will mean more work writing our own drivers, and being ham-strung by the endian-conversion overhead (it's hurting our graphics performance).

I'd love to almost put myself out of a job, and be able to recompile libdrm driver code for AmigaOS with little to no changes. That would free my time up for other tasks. We'd finally be able to use drivers effectively written by the manufacturers themselves (or open-source enthusiasts).

Hans

Last edited by Hans on 22-Sep-2022 at 08:36 AM.
Last edited by Hans on 22-Sep-2022 at 08:34 AM.

_________________
http://hdrlab.org.nz/ - Amiga OS 4 projects, programming articles and more. Home of the RadeonHD driver for Amiga OS 4.x project.
https://keasigmadelta.com/ - More of my work.

@Hans

I am no driver developer so I ask

I assume that drivers are software interfacing the OS so what you propose heavily rely on the OS, expecially the graphic part of the OS. To do what you propose would mean to adapt the graphic part of AmigaOS to a system like linux?

@Nonefornow

Quote:

At the end of the video David Pleasance asked for the "thoughts" of the attendees. Is there a continuing part of the video?

Sorry I did'nt record that bit, basically most said it was a good idea, he then played, some vids of the people supporting him, basically the end bit of this vid
https://www.youtube.com/watch?v=cfOrM1JnCs8

_________________
AmigaNG, YouTube, LeaveReality Studio

@amigang

how much money do these "supporters" give to the project? Talk is cheap...

As said common marketing can make commercial sense at some point. But not yet and not in the form of this project

@OlafS25

Quote:

I am no driver developer so I ask I assume that drivers are software interfacing the OS so what you propose heavily rely on the OS, expecially the graphic part of the OS. To do what you propose would mean to adapt the graphic part of AmigaOS to a system like linux?

You'd need some kind of Linux kernel compatibility wrapper. It doesn't need to emulate the Linux kernel, just provide API compatibility for the bits that drivers use. IIRC, that's what deadwood did to get Gallium3D working on AROS.

Hans

_________________
http://hdrlab.org.nz/ - Amiga OS 4 projects, programming articles and more. Home of the RadeonHD driver for Amiga OS 4.x project.
https://keasigmadelta.com/ - More of my work.

@Hans

thanks for explanation

I'm willing to throw in 35 Euro or even 70 Euro for the first year to see where it goes. I think they should take advantage of properties that Trevor already owns like Amiga.org & Amigaworld.net. Maybe that's a better idea then making yet another Amiga website?

_________________
Sent from my Quantum Computer.

@DiscreetFX

if you have money to throw out of the window I can give you my bank data. I always need money

Last edited by OlafS25 on 22-Sep-2022 at 03:37 PM.

@DiscreetFX

Quote:

they should take advantage of properties that Trevor already owns like Amiga.org & Amigaworld.net.

But then they'd lose their only cash-cow, the website.

_________________
The Rear Window blog

AmigaOne X5000/020 @ 2GHz / 4GB RAM / Radeon RX 560 / ESI Juli@ / AmigaOS 4.1 Final Edition
SAM440ep-flex @ 667MHz / 1GB RAM / Radeon 9250 / AmigaOS 4.1 Final Edition

Hans Quote:

AmigaOS 4 on WinUAE is horribly slow and vastly inferior to real hardware. I think it's partially due to the emulated graphics cards being old and not supporting compositing, but the PPC emulation is probably also slower than real hardware. Anything Warp3D Nova related won't work either. The same goes for QEMU based NG emulation. It can't match real hardware at all. Not even close.

Thanks for confirming my expectation that AmigaOS 4 for emulation was advertising. I suppose it wasn't a complete gimmick as it likely was usable for light AmigaOS 4 work and it could be installed on mobile computers. The perspective of emulation by 68k AmigaOS users is high because emulation surpasses the old hardware by ~3 decades while PPC AmigaOS hardware is closer to 1 decade behind modern hardware (GPUs maybe half a decade behind modern thanks to you?). There is likely room for PPC emulation improvements but it is pointless when emulated 68k Amigas already offer better performance, a smaller footprint and better 68k Amiga compatibility with more software. Not only that, but modern features like RTG, 3D, video decoding, etc. can be passed through to the 68k side. Maybe the PPC Amiga emulation experience is poor but the 68k Amiga emulation experience is close enough to compete with PPC Amiga hardware and desktop PC hardware offers more value than PPC Amiga hardware. It would be possible to improve the value of PPC Amiga hardware with mass production but ~2 decades of poor sales is discouraging, AmigaOS 4 lacks competitive desktop features and 68k Amiga compatibility is fair while emulated 68k Amiga compatibility is very good. I can't see any justification for continuing PPC Amiga hardware and it appears the Amiga future is headed back to (or more likely never left) the 68k either with emulation if nothing is done or to 68k hardware with a roadmap to modernization. Emulation will perpetually use original 68k Amiga standards with non-standard modern features and I/O while 68k hardware is fully customizable and an ASIC would set new standards while killing divisions from noncompetitive emulation and niche hardware. That's the reality and those are the choices as I see them. At least we both agree that even the best emulation is "horribly slow and vastly inferior to real hardware" for NG Amigas. We don't seem to be in the majority in this thread though.

Hans Quote:

I agree with Shades regarding a 68K ASIC. A "custom" ASIC which is basically putting a set of predesigned ASIC cores & peripherals in one SoC is relatively easy. The more custom you go, the higher the expense. Adapting a core designed for FPGA to ASIC is more expensive. That's ignoring the work needed to interface the core with more modern peripherals or an existing GPU. Designing your own CPU and/or GPU cores and writing GPU drivers would increase the work and cost even more (a lot). The more custom you go, the more customers you'll need to pay for it all.

Obtaining a more modern and higher performance 68k CPU core suitable for ASIC conversion is the difficult part but there are options. Sadly, I doubt the Apollo core is suitable and significant redesign would be necessary to make it ASIC ready although major portions of logic could likely be used for a professional design under the leadership of a more experienced chief architect. I would approach Jens, and not Gunnar, about the N68050 core which he coded in VHDL and wanted to release as open source while I was part of the Apollo team. I expect Gunnar to be unreasonable but Jens and Thomas are professionals that could be part of a development team. The overshadowed TG68K core could be useful which talented developers like Tobias Gubener, Alastair Robinson and others have worked on. Even under the leadership of an experienced chief architect and using available existing VHDL code, it would likely take considerable time to develop and verify a new in-order superscalar core with modern features. This is where it would be good to look into licensing more mature and ASIC ready 68k cores. The 68060 CPU core is old but a semi-modern design and gained full "MC" certification which requires extensive verification. Licensing or even buying the NXP 68k IP and assets should be looked into. NXP is making minimal if any money off the 68k line and it has minimal potential from their perspective to make money in the future. At the very least, it would be great to buy the 68k (and maybe 88k) IP and assets for a museum and retro logic evaluation and preservation which could gain positive publicity and support from developers. Don't underestimate the retro excitement and market potential. We started to see it on the Natami forum with the Natami "MX Bringup Thread" having 761,487 views. There are at least two 68020 compatible ASIC ready cores that I know of which could be licensed although they are much less advanced than the 68060 or Apollo core.

The Amiga chipset is easier as there are already at least three AGA compatible cores. Most, if not all of them, support chunky RTG and digital video output (upgraded Amiga 16 bit 8+ voice audio is also available). The chipset would need extensive verification and testing before being used in an ASIC and more modern IO added but a FPGA could be used for the chipset if not ready in time (FPGA capabilities for other chipsets would be a major asset for retro and embedded markets anyway). Could 3D hardware be added on top of the Amiga chipset chunky modes? I think so. Emulators are already doing it and a FPGA or ASIC can do all the same logic. Emulators have also shown that chip memory can be expanded while retaining excellent compatibility and increased performance. Emulation has few enough limitations that retro 68k Amigas are competing with PPC Amiga hardware and many 68k Amiga users think emulation is good enough for the Amiga platform. Emulation is good enough for a retro Amiga platform but not for a NG Amiga platform even though the price drop of hardware from an ASIC is good for retro markets as Retro Games Ltd. planned but couldn't pull off.

IP would need to be licensed and is available. Depending on licensing costs and requirements, this is often worthwhile. HDMI licensing is pretty much a given and reasonably priced for mass production. Standard ASIC blocks are commonly licensed and practically required for an advanced ASIC. SerDes IP needs licensing for high speed digital I/O and isn't cheap but necessary for high performance and efficient I/O. Licensing a 3D GPU core would save a huge amount of time. The Imagination Technologies hybrid ray tracing GPU looks almost too good to be true considering their low power claims which may allow fanless ray tracing for the masses. "Ray tracing" is a great marketing feature for the Amiga where PC ray tracing started even though modern hardware ray tracing uses different algorithms.

Hans Quote:

I'd go the other way, and try to leverage other people's work as much as possible. Switching to ARM or some other CPU architecture allows you to take advantage of the work done by thousands of engineers without having to pay for it all out of our own pockets. Likewise, it's best to switch to little-endian. Fewer and fewer GPUs and peripherals are bi-endian, and less and less work is being done to support big-endian CPUs in drivers. Hence, staying big-endian will mean more work writing our own drivers, and being ham-strung by the endian-conversion overhead (it's hurting our graphics performance).

If porting the AmigaOS to ARM for NG hardware, I agree that would be better to go full little endian as it is native (code is little endian), better supported on that architecture and big endian support is practically deprecated with AArch64. The Amiga platform will remain more divided as you leave behind much of the retro market just like before with PPC hardware. The hardware should be cheaper using the RPi but Amiga businesses would become software only businesses trying to sell AmigaOS in a market where most of the OSs are free. At least the trade off of desktop features for better performance and a small footprint is not as bad for low end RPi hardware but desktop Amiga emulation will have better performance, better pass through features and better retro compatibility (MorphOS x86-64 will also have a performance, feature and maturity advantage). There is no problem for little endian GPUs in an ASIC as the data lines could be swapped. There is only one GPU to support and one driver to write. The Imagination Technologies GPU is scalable so an upgrade would just have more parallel GPU hardware.

Hans Quote:

I'd love to almost put myself out of a job, and be able to recompile libdrm driver code for AmigaOS with little to no changes. That would free my time up for other tasks. We'd finally be able to use drivers effectively written by the manufacturers themselves (or open-source enthusiasts).

A standard GPU would also free up your time as only one driver is needed. In the case of an ASIC, you should get loads of documentation and example code for once instead of wasting time with reverse engineering.

Last edited by matthey on 22-Sep-2022 at 09:46 PM.
Last edited by matthey on 22-Sep-2022 at 08:32 PM.

@Hans

Quote:

and being ham-strung by the endian-conversion overhead (it's hurting our graphics performance).

I see no harm in sending LE textures / bitmaps to the GFX card, why convert in the driver? I can just as easily work with GBRA as can with ARGB format, older driver used offer PC/LE modes. And if you’re converting 8bit to 16bit LE or BE makes no difference, its only palette lookup table that’s needs to be different (a table of 256 is quick to convert.)

Last edited by NutsAboutAmiga on 22-Sep-2022 at 08:49 PM.
Last edited by NutsAboutAmiga on 22-Sep-2022 at 08:47 PM.
Last edited by NutsAboutAmiga on 22-Sep-2022 at 08:45 PM.

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