Quote:

agami wrote: I'm always more interested in forward motion.

QUICK FRIEND QBIT, SEND OVER A FEW VIDEOS ABOUT MAGNETIC MOTORS!!!







.mega

_________________
I HAVE ABS OF STEEL
--
CAN YOU SEE ME? CAN YOU HEAR ME? OK FOR WORK

@MEGA_RJ_MICAL


I send you some Music hahahahahaha

https://www.youtube.com/watch?v=K1T4lXfEfyQ


My suggestion is keeping the classic Amiga Emulation strict for all Classic Amigas in Quickstart with WinUAE.. and adding a Option Custom "Classic experimental UBER Amiga"

Last edited by QBit on 17-Jan-2022 at 01:38 PM.
Last edited by QBit on 17-Jan-2022 at 01:32 PM.

@agami

Quote:

agami wrote: I even discussed with Ben (Hyperion) the possibility of licensing access to the Amiga OS 3.1 source to further research my ideas and concepts. I assured him I was not looking to port it to another ISA, and that I would share any improvements that would be made. Alas, Ben wanted a lot more $ than I had at my disposal.

Oh, you still can use AROS, the 68k version is compatible.

And for Ben, it is interesting to see him juggling with licences, that where granted 'for marketing of AOS4 only' ...

Last edited by OneTimer1 on 17-Jan-2022 at 01:35 PM.

@Hypex

I haven't been using OS 4.x, just 3.1. I haven't even upgraded to 3.2 on WinUAE.

_________________
I sold my SAM460ex lite... waiting for money to buy a Raspberry Pi... or a Classic A1000 with Buffee... or an A1222... and OS4.3 FE update 11

@OneTimer1

Quote:

Oh, you still can use AROS, the 68k version is compatible

Not sure in what state AROS 68k was back when I had this exchange with Ben (4-5 years ago), plus I was really keen to have a look at Exec to see how Carl did it.

Quote:

And for Ben, it is interesting to see him juggling with licences, that where granted 'for marketing of AOS4 only'

The fact is, Amiga OS 3.x source code is not open, and Hyperion are the gatekeepers. It's not so much about the cost of licensing access to the old source code, Ben wanted a lot of $ in escrow in case my team ended up accidentally/deliberately leaking the code on the internet.

Either way, what I told him then is that my project doesn't depend on Amiga OS 3.x. More of a "nice to have", but I could just work with other open source projects such as AROS or even Haiku.

_________________
All the way, with 68k

agami Quote:

It's a nice dream. I have a similar dream, and I even discussed with Ben (Hyperion) the possibility of licensing access to the Amiga OS 3.1 source to further research my ideas and concepts. I assured him I was not looking to port it to another ISA, and that I would share any improvements that would be made. Alas, Ben wanted a lot more $ than I had at my disposal.

Why would you contact a licensee of AmigaOS 3.1, who may soon lose the license, instead of the owner of AmigaOS 3.1? Did you try contacting Michele Battilana?

agami Quote:

@matthey and I have also discussed at length the various options and possibilities for a new 68k ASIC, and there's a lot of promising options there, but of course that too requires quite a bit of investment. Essentially, what you are proposing is, as you say, well within the sphere of "possible", but just because it is possible does not mean it's worth the time (money) and effort. But it would be an awesome passion project nonetheless. By creating a custom 68k core (SMP/AMP/64-bit) in software, you would end up using WinUAE as a hardware DE. I get it.

WinUAE has been helpful in developing and debugging real hardware. The same would likely be true of new hardware although even high end x86-64 hardware may struggle to smoothly emulate future higher end Amiga hardware which is still very affordable. It may be especially challenging to emulate multicore hardware and would require extra programming effort. I framed many of my comments as targeting both 68k emulation and new hardware as they both allow much more flexibility than commodity hardware. The big question is how much of a performance impact would more Amiga compatible custom hardware add and, as I have pointed out, perhaps not as much as even some core architects would have thought. The actual impact depends on the core design which likely would require some changes from traditional designs but they would be good things to explore. As the MIT research indicated, a more sequential memory model for a multicore design had negligible performance differences while being easier to program, less error prone for software, saved chip area and was higher performance for multicore sharing.

agami Quote:

As other's have pointed out, it would also require new software silicon for a new chipset that would work with the new 68k core. Whether or not the new custom chipset design would have backward compatibility, and to what degree, it would still require a large investment.

A large investment is relative. It's not that expensive to become a fabless semiconductor business today. It used to be more expensive when CBM was in business but ARM was developed about the same time as the Amiga starting in 1983 with the first CPU ready in 1985 by Acorn Computers. It was already possible by a small businesses with little related experience and turned out to be a tremendous investment. The fabless semiconductor business ended up being worth much more than the hardware manufacturing and software part of the computer business which struggled and disappeared. Fabless semiconductor startups like P.A. Semi and SiFive have also been good investments and there have not been enough of them. They have often been bought out by larger companies before they could shake up the tech market status quo with innovations.

Hardware has been the driver of excitement for the Amiga since its inception. Even Amiga software developers like Carl Sassenrath looked at the hardware diagrams on the drawing board and decided they wanted to be a part of the development. It was primarily the Amiga hardware which sold Amigas for the first few years as the software titles and CBM advertising were lacking. Standard hardware was the reason the Amiga could provide more than the competition for less cost. The Raspberry Pi Foundation has had similar success with standard affordable hardware for the masses. Their investment for the future is in fabless semiconductor production of their own ASIC SoCs.

agami Quote:

I personally do not care about backwards compatibility. I'm always more interested in forward motion. Outside proving some academic point, and as others have mentioned, once backward compatibility is out of scope there are many existing microprocessor architectures that would be more cost effective than back-porting decades of CPU design progress into the 68k architecture. Many millions of dollars would be required to create your software only UBER Amiga, but the returns would be very limited. There was a time, before Commodore went bankrupt, to have a forward looking Amiga that would foster new development while also maintaining backward compatibility with a large % of older software. Back when 5% of global personal computer users were using an Amiga. The way I see it, there are plenty of emulation and physical hardware options to run old Amiga 68k software. What I care about is creating a new pool of developers for a new system which can potentially offer the market an alternative to the Windows/macOS "lesser of two evils" stagnant ecosystem. If a new 68k core can play a role, that would be great, but it's not essential to the plan.

Developing the 68k architecture would have additional up front development costs but then cheaper production costs than licensing available architectures like ARM. Most commodity hardware has another layer of cost in the profit. The actual silicon production costs are very cheap as can be seen by Raspberry Pi products even while they pay licensing costs to ARM and markup costs to Broadcom. The $1 RP2040 SoC ASIC cuts out Broadcom allowing for price reductions. They would have trouble replacing the ARM core to save production costs and ARM really is native as they are somewhat of a RISC OS Acorn hobbyist remake. The Amiga is much more popular and retro, has more and better software especially games, is big endian and no licensing is required for the 68k architecture.

I'm not calling for an UBER software only Amiga and nobody will catch me asking Toni for one. I think it would fail by itself. I don't think there is any software only Amiga product which will succeed on any architecture especially not emulated. People who underestimated the importance of compatibility have usually been wrong. Look no further than the failure of Motorola in switching from the 68k to PPC. Motorola thought it would be easy to convert people to PPC but it was not. Intel stayed with x86 and benefited from the compatibility and loyal customers until they tried to switch architectures with the epic $10 billion Itanic mistake. This allowed AMD to define AMD64 and gain market share with the continuation of the x86 64 bit architecture. I'm not opposed to Amiga on other architectures and for the PC market. I just see it as a dead Amiga with no momentum to penetrate the PC market with a lack of software even if the AmigaOS was modernized to become competitive.

QBit Quote:

My suggestion is keeping the classic Amiga Emulation strict for all Classic Amigas in Quickstart with WinUAE.. and adding a Option Custom "Classic experimental UBER Amiga"

WinUAE usually executes code faster than a real Amiga and many users don't limit the performance or memory except when necessary. There was recently a thread on EAB that talked about Final Fight Enhanced and how it may only be possible with WinUAE removing the '90s Amiga bottlenecks and running faster than selected hardware spec.

http://eab.abime.net/showthread.php?s=d88af227d2ee5ae911fd0db00612d3f0&t=109488

The following is the demonstration video of a nearly arcade accurate Final Fight on the Amiga.

https://www.youtube.com/watch?v=ZfuAvYCAXss

WinUAE can already do this. An AGA Amiga with a 68060 can likely already do this by rendering to a fast memory buffer and then doing a C2P conversion. Even if UBER features were added to WinUAE, it still doesn't revive the Amiga, create a standard enhanced hardware platform and generate excitement like enhanced retro hardware would. It just needs to be affordable enough for Raspberry Pi customers to want to play enhanced arcade quality games like this.

@matthey

Doesn't the Vampire V4 stand alone and V4 based add-on cards for classic Amigas create a new higher spec Amiga base system that has many new modern features? In my opinion it does. WinUAE, Amiga Forever and FS-UAE do some amazing things too but they are not adding a modern higher spec for CPU/GFX/Audio and so on then what the classic Amiga did without/with add-on cards from that era. WinUAE, Amiga Forever and FS-UAE are amazing Virtual Machines for running Amiga software/hardware and people should certainly buy/support and enjoy them.

Last edited by DiscreetFX on 18-Jan-2022 at 04:46 AM.

_________________
Sent from my Quantum Computer.

DiscreetFX Quote:

Doesn't the Vampire V4 stand alone and V4 based add-on cards for classic Amigas create a new higher spec Amiga base system that has many new modern features? In my opinion it does.

Yes, it does. SAGA has chunky support up to true color, more layers including chunky and PiP with dual playfield enhancements, much higher performance chip memory supporting up to 12MiB currently, 16 sprites 32 bits wide with less restrictions like color palette, an enhanced copper that can use 32 bits of data instead of 16 bits of data at a time, 8 or 16 voice? 16 bit sound with a 24 bit audio dynamic, etc. The Apollo core has high memory bandwidth, is 64 bit, has a fully pipelined but reduced precision and instruction FPU from the 68060, has many newer features the 68060 lacks like larger caches, a hardware return/link stack, more code folding/fusing (the 68060 uses on branches/loops), has a 64 bit wide integer only AMMX SIMD unit, etc. I like the SAGA enhancements which work well in FPGA and are not too limited from what would be used for an ASIC. The Apollo core enhancements are very much limited by FPGA and highly optimized for a FPGA including the ISA. The internal optimization could be changed for an ASIC but the ISA is crippling. For example, a 64 bit wide integer only SIMD unit from the '90s is not competitive for even semi-modern 3D support. No compilers have supported the Apollo core ISA because it is in FPGA only, it is optimized for a FPGA, instruction timings vary by the FPGA used and it is difficult to support (not the sign of a good ISA IMO). It is a good example of one man sabotaging himself with shortsightedness.

DiscreetFX Quote:

WinUAE, Amiga Forever and FS-UAE do some amazing things too but they are not adding a modern higher spec for CPU/GFX/Audio and so on then what the classic Amiga did without/with add-on cards from that era. WinUAE, Amiga Forever and FS-UAE are amazing Virtual Machines for running Amiga software/hardware and people should certainly buy/support and enjoy them.

UAE offers a high performance CPU core on modern x86-64 hardware, modern and high performance memory and chunky RTG enhancements at least. This is significant and gives a more usable computer than many of the original Amigas. The most compatible settings of WinUAE do slow it down considerably though. As good as WinUAE is, it has failed to create a new enhanced or revived Amiga platform. Even if SAGA and Apollo core enhancements were implemented, I believe it would still fail to gain traction. Like FPGA ISA enhancements not being generally supported, the same is true of emulated platforms. Serious developers would at least plan for an ASIC. Planning for and setting standards for a FPGA or emulation is saying we are niche market amateurs.

@matthey

Quote:

The most compatible settings of WinUAE do slow it down considerably though.

Well, d'oh - the most compatible settings is to be _exactly_ like an old 68000 A500, then it will of course also be as slow as an old 68000 A500.

Quote:

As good as WinUAE is, it has failed to create a new enhanced or revived Amiga platform.

You shouldn't really see WinUAE isolated, it is part of a larger eco system that includes not just other UAE incarnations, but also Aranym and other "platform emulators", as well other 68k cpu emulators like musashi and emu68. Because... open source.

Quote:

Even if SAGA and Apollo core enhancements were implemented, I believe it would still fail to gain traction. Like FPGA ISA enhancements not being generally supported, the same is true of emulated platforms. Serious developers would at least plan for an ASIC. Planning for and setting standards for a FPGA or emulation is saying we are niche market amateurs.

In May it will be 5 years since SAGA was announced to become open source - it still isn't, and it is still unclear what SAGA (or SuperAGA, as it says on apollo-computers.com) REALLY is. Is it even finished yet? Is what's currently on V4 the final implementation? Is it really a superset of AGA in all ways, or is it a different chipset that just shares some features with AGA (think venn diagram)? And then AMMX, and what about software? And what's REALLY the point, when all the masses are craving for is launching old games from a high colour workbench and video out on HDMI?

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

SAGA should spark some new 512 colour, 8 channel games as well as be compatible with old ones. Maybe we can get some GBA ports to V4 SAGA?

Last edited by BigD on 18-Jan-2022 at 11:15 AM.

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@OneTimer1

Quote:

Better emulate Amiga on WinUAE in Amiga mode.

That's easier but not the same thing.

@PhantomInterrogative

That would explain why WinUAE feels faster than your Sam. It's not emulating the same system. Unless you mean just running 68K code then not even native big endian will save the Sam there as it's going against hardware that's way faster.

@kolla

who cares?

Most current amiga game devs are not interested in new advanced chipsets but in the challenge and attraction to get most of the old hardware. A new super-classic chipset is not interesting to them. And for commercial development the user base is too small. Additional you can today use RTG + AHI. Lots of game ports do that successfully. There is no shortage of options or a problem because SAGA is not open source but we need more users and developers in general

@OlafS25

As long as SAGA remains “apollo only” any wider adoption and support simply won’t happen. If SAGA really became open source and a commnity wide consensus could be reached regarding spec standardization, so that several parties could implement SAGA - then perhaps.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

OlafS25 Quote:

Most current amiga game devs are not interested in new advanced chipsets but in the challenge and attraction to get most of the old hardware. A new super-classic chipset is not interesting to them. And for commercial development the user base is too small. Additional you can today use RTG + AHI. Lots of game ports do that successfully. There is no shortage of options or a problem because SAGA is not open source but we need more users and developers in general

Even if there were new enhanced Amiga hardware standards, there will always be a few Amiga devs targeting the 68000+OCS/ECS and 68020+AGA because those were Amiga hardware standards. This is not a problem with backward compatibility maintained. Some software enhancements using hardware enhancements are possible and relatively easy to the same software base. This means older games, like arcade and movie games, could become enhanced with minimal rework for example. Yes, the user base needs to grow to create a new enhanced "hardware" standard. Hardware standards need to be "hardware" and competitive. Competitive hardware does not have to be high end. The Amiga is more competitive for low end hardware. The CPU core performance needs to go up and the price needs to come down which points to ASIC SoC. The dual core RP2040 SoC ASIC costs $1 to produce including the ARM license and is likely roughly equivalent to the Apollo core with SAGA in gates. ASIC area is cheap though. A 1GHz dual core 68k Amiga SoC ASIC for $3 production cost may be better suited for the Amiga. Professional design help (which I tried to bring in for the Apollo core project), investment funding and official branding would be valuable for competitive low cost hardware to enlarge the user base.

kolla Quote:

As long as SAGA remains “apollo only” any wider adoption and support simply won’t happen. If SAGA really became open source and a community wide consensus could be reached regarding spec standardization, so that several parties could implement SAGA - then perhaps.

Thomas wrote SAGA and Jens wrote the N68k (N=Natami) which the Apollo core is based on. Thomas and Jens are both professional and reasonable team players. The spec standards are dictated by one person who took control while the Amiga dev community has largely rejected his enhancement standards.

SAGA being open source would not necessarily result in hardware standardization. There are two other FPGA AGA cores which have not resulted in enhanced hardware standardization. Granted, they are very compatibility focused but there have been enhancements to FPGA Amiga cores for chunky RTG, digital video output and audio. The project leaders I have contacted have not been interested in standards while they hypocritically claim they are retro focused. Too much effort was wasted in reinventing the Amiga wheel with three AGA compatible FPGA cores when the extra effort could have been put into improving one. A consumer FPGA based device is better suited for smaller scale simulation where accurate simulation is wanted and enhancements push the scale and cost up anyway. If enough customers want more enhanced and improved hardware, then an ASIC becomes more appealing where many of those FPGA devices would become noncompetitive and disappear including Vampire hardware.

Last edited by matthey on 18-Jan-2022 at 07:53 PM.

@matthey

Quote:

Why would you contact a licensee of AmigaOS 3.1, who may soon lose the license, instead of the owner of AmigaOS 3.1? Did you try contacting Michele Battilana?

As I mentioned in my response to @OneTimer1, my conversation with Ben was 4-5 years ago.
What I knew then is that Hyperion had access to the Amiga OS 3.x source. I was unaware at the time that Cloanto had access to the source.

Neither here or there right now in January 2022. AROS 68k appears to be in very good working condition.

Quote:

no licensing is required for the 68k architecture

Is this a fact?
I remember looking into this (Google searches) a few years back and couldn't find an answer, neither that it didn't require licensing, nor info regarding licensing requirements for 68k.

_________________
All the way, with 68k

agami Quote:

As I mentioned in my response to @OneTimer1, my conversation with Ben was 4-5 years ago. What I knew then is that Hyperion had access to the Amiga OS 3.x source. I was unaware at the time that Cloanto had access to the source.

At that time, Amiga Inc. may have been the owner of the AmigaOS 3.x sources. Cloanto may have been able to grant a license too. Shop around but don't expect Ben to give you the best deal and it's really not a good deal for a sub licensee if he loses his license.

agami Quote:

Neither here or there right now in January 2022. AROS 68k appears to be in very good working condition.

The price is right. It's sad that the AmigaOS wheel has to be reinvented so many times but it's good to put pressure on the lame Amiga hoarders that think it's so valuable when there is a free alternative.

matthey Quote:

no licensing is required for the 68k architecture

agami Quote:

Is this a fact? I remember looking into this (Google searches) a few years back and couldn't find an answer, neither that it didn't require licensing, nor info regarding licensing requirements for 68k.

All the 68k trademarks are expired. There are still valid copyrights but they would apply if copying code or logic from an existing 68k design, copying support code or copying documentation for example. There would certainly be no royalties like when using ARM cores. Gunnar also contacted Freescale/NXP and asked if they had any problem with a new 68k design and they answered no. If they cared which they likely don't, they probably couldn't do anything about a new 68k design like Intel couldn't do anything about AMD x86 designs. The trademarks being expired makes it even safer though. The J-core project has created SuperH cores in a similar manner with the SuperH trademarks expired, most of which were newer than the 68k trademarks. The project provides open hardware natively big endian cores if wanting a core for Sega Saturn and Sega Dreamcast compatibility or just a royalty free big endian core. While simple and compact, it looks like SuperH would have performance issues for a high performance core though. The fixed length 16 bit RISC encoding (instructions) cause an excess of instructions to execute and data cache inefficiency from lack of immediate encoding bits, significantly more than the more common 32 bit fixed length RISC encodings. ARM Thumb 2 fixed this problem with a variable length encoding again (68000 was copied to create SuperH which was copied to create Thumb 2, each in turn taking over the dominate position in the embedded market). The J-core project brings back limited variable length instructions but the damage is done without a re-encoding which would break compatibility and they want to use the existing SuperH developer tools.

@OlafS25

Quote:

Most current amiga game devs are not interested in new advanced chipsets but in the challenge and attraction to get most of the old hardware.

Most don't seem to be interested in AGA either. Even though it's so old now it should be standard. But being the last Amiga chipset also makes it rarer.

Funny, by comparision if the PC had died instead of the Amiga in the 90's; most retro PC fans would be into 286 and EGA, and a few would be into 386 and VGA. 486 and SVGA for the die hard power fans.

It does build on the original chipset so satisfies most gamers. Perhaps better for gamers than game makers. The sound is better but not mixed the Paula way as extra channels and panning with 16-bit is more practical in a 24-bit mixer.

I don't know if the blitter is improved or updated for chunky. As expected chunky is added but is separate. I read about the chunky plane but it doesn't slide into a bitplane slot for transparent upgrade to bitplanes.

Extra sprites. But I'm sure AGA had 64-bit where as I read the SAGA ones are 32-bit. Though if increasing arcade abilites then adding tile planes may be best as most consoles had matrix style graphics layers with tiling and not a simple bitmap with limited parallax abilities.

@Hypex

I am no AGA/SAGA expert but from what was wrote it offers extensions to AGA like full 24bit colour depth and 16bit sound, partly what was planned with AAA but even superior to it. The potential is obvious when you see all the PC ports that at least need 68060 and AHI (and RTG) to run. The same certainly could be done using the chipset, propably more elegant than using CybergraphX (AROS) and AHI. The problem is (from what people told me) that it is somewhere between classic and todays world but not near enough to what is possible on modern hardware.

Hypex Quote:

Most don't seem to be interested in AGA either. Even though it's so old now it should be standard. But being the last Amiga chipset also makes it rarer.

I think the rarity of AGA hardware has a lot to do with it. I believe many Amiga users would like to experience AGA and a faster CPU like the 68060 but emulation isn't the same and some developers are reluctant to make games for a user base which is mostly emulation and not so cheap FPGA hardware. With original AGA hardware, emulation and FPGA hardware, the 68020+AGA base may be larger than what developers realize too.

Hypex Quote:

Funny, by comparision if the PC had died instead of the Amiga in the 90's; most retro PC fans would be into 286 and EGA, and a few would be into 386 and VGA. 486 and SVGA for the die hard power fans.

lol

Hypex Quote:

It does build on the original chipset so satisfies most gamers. Perhaps better for gamers than game makers. The sound is better but not mixed the Paula way as extra channels and panning with 16-bit is more practical in a 24-bit mixer.

Enhancing in a similar way to what CBM was planning should certainly help acceptance as a what if product. AA+ was more practical than AAA as it retained backward compatibility. The following was planned for AAA but I think it is the same specification as AA+ but AA+ would have had a backward compatible register map while AAA breaks compatibility.

Quote:

Updated version of Paula called 'Mary' with 8 voices that can be assigned either to left or right channel; each channel has 16-bit resolution with up to 100 kHz sample rate; additionally it does 8-bit audio sampling input.

https://en.wikipedia.org/wiki/Advanced_Amiga_Architecture_chipset
https://en.wikipedia.org/wiki/Commodore_AA+_Chipset

Even one more generation of advancement with AA+ would have made the Amiga seem much more modern with fewer limitations.

Hypex Quote:

I don't know if the blitter is improved or updated for chunky. As expected chunky is added but is separate. I read about the chunky plane but it doesn't slide into a bitplane slot for transparent upgrade to bitplanes.

My understanding is that the SAGA blitter is AMMX on another Apollo core thread emulating the blitter. AMMX can deal with chunky gfx data although there is no need to start a separate thread to do it. It's a good question what to do with the blitter which may be outdated with the addition of chunky and a SIMD unit. The original blitter has significant startup overhead and increasing the clock speed of the blitter for all programs could cause compatibility problems. I'm not aware of compatibility problems from the multi-threaded blitter emulation. I believe Thomas Hirsch planed to implement a hardware blitter but it was never finished on the working Natami prototypes which used the CPU like other Amigas with a fast CPU.

Hypex Quote:

Extra sprites. But I'm sure AGA had 64-bit where as I read the SAGA ones are 32-bit. Though if increasing arcade abilites then adding tile planes may be best as most consoles had matrix style graphics layers with tiling and not a simple bitmap with limited parallax abilities.

SAGA is compatible with AGA so all sprite capabilities are at least as good. Maybe the 64 bit wide sprites were 2 attached AGA sprites allowing for more colors but dropping the effective number of sprites to 4 while SAGA doubles the 8 to 16 and removes the color palette limitations? The following link has a demo of the SAGA sprite capabilities with some explanation about the capabilities.

SAGA Sprite Repeat X FLAG
https://youtu.be/Is3V4BhwWbU?t=63

OlafS25 Quote:

I am no AGA/SAGA expert but from what was wrote it offers extensions to AGA like full 24bit colour depth and 16bit sound, partly what was planned with AAA but even superior to it. The potential is obvious when you see all the PC ports that at least need 68060 and AHI (and RTG) to run. The same certainly could be done using the chipset, probably more elegant than using CybergraphX (AROS) and AHI. The problem is (from what people told me) that it is somewhere between classic and todays world but not near enough to what is possible on modern hardware.

As I wrote above, SAGA may be more like AA+ which was a more practical and compatible version of AAA. Some enhancements surpass AAA but some ideas are not practical or necessary today. Some limitations can be removed while in theory maintaining better backward compatibility than AAA, AA+ and even AGA.

You are correct that 68020+AGA to 68060+RTG standards can merge on enhanced hardware maximizing the amount of usable Amiga software. There are more than a few compiled ports of PC games coming to the Amiga for 68060+RTG and even a few beyond because of requirements. Cowcat has a 68k+RTG build of Quake 3 working but even a 68060 likely can't give a descent frame rate and it requires 256MiB of memory beyond most 68k hardware.

Quake 3 on Classic
http://eab.abime.net/showthread.php?s=f9af20a75871a9d701aafff3ece75fcd&t=89741

One game may sound insignificant but the same engine is used for Smokin Guns, OpenArena, RTCW, Jedi Outcast, etc. The next bump up in CPU performance and 512MiB of memory may allow for a more modern engine used by several games which can be ported. Around 512MiB or 1GiB of memory is where modern browsers start to become usable too.

Ideally, a revived 68k Amiga with an emphasis on gaming would play most original Amiga games, enhanced arcade quality often bang the medal games, new indy games and compiled RTG ported games. The Raspberry Pi has RTG ported games and some indy games but really doesn't have the retro gaming potential the Amiga has. The Raspberry Pi does have 3D even though it is low performance. There are 3D cores available for licensing which should work but 3D often takes half or more of a SoC chip area increasing cost.