@NutsAboutAmiga

Quote:

NutsAboutAmiga wrote: @cdimauro Quote: So, you're missing one of the key elements which makes/defines an Amiga. And you dismiss part of Amiga’s upgrade history.

I don't. I've made PRECISE claims that match with the history.
Quote:

Beside already disproven by PiStorm, and hardware upgrades, you don’t need a 680x0 CPU, and get by with incompatible BIOS. Just look at EmuTOS.

This is a completely different thing and I've nothing to say about that. In fact, I'm supporting emulation since decades...
Quote:

Quote: They aren't ported likely because they hit the hardware and they are written in pure assembly. So, not worth the effort. Nope, people got old, most people I talk to at work, or other places do not use an Amiga anymore, now are you shocked? People had time in there 20’s maybe lacked cash back them. so they started getting kids when where 25 or older, now they don’t have time, that’s pretty much the story. Some people quitted 1990’s, and some quitted in 2000’s, the Amiga thing is declining hobby. With fewer and fewer people using and developing for etch year. Just a fact of life. And I expect is not going to get better. And the demo codes I know that use AmigaOS4.x, mostly do not make demos anymore, they write libraries, or mess around with SDL and games, programs, maybe a age thing or maybe it’s not as big priority as other things.

OK, this explains it as well. However don't underestimate the efforts required for porting an Amiga demo.
Quote:

Quote: Why do you need a FPGA? Easier adaption, and acceptance, with better legacy support, PowerPC path might feel like more natural / smoother path.

FPGAs are slow for implementing processors, if this was your idea. You need ASICs for that.

Whereas they are good enough for implementing old chipsets. However that's good if you implement (or use) also the processors that those systems used. I mean: I don't see any sense on "hybrid" implementations (e.g.: a PowerPC CPU + the Amiga chipset).
Quote:

Quote: Aren't PowerPCs powerful enough to emulate the Amiga's chipset? No issues, just we running on one CPU core, the emulation takes up everything. the GFX convention is pretty complex thing on its own, P2C, 8bit + Color palette to 16bit or 32bit. The copper thing is its own secondary interpreter. UAE in general has updated much for AmigsOS4.x, most developers writing software for AmigaOS4.x/PowerPC instead.

That's normal: an emulator usually uses just one core for emulating the entire machine. That's if perfect compatibility is a goal.

If full compatibility isn't a concern, then some additional cores might be used. However and likely this will never be the case for a tightly integrated system like the Amiga.
Quote:

PiStorm or Putinia / Trace don’t need to deal with that, and run faster then,

Exactly: because you don't have the burden of implementing the chipset AND doing it in the most compatible way.
Quote:

any UAE installed on same system, not sure why any wonts install UAE on PiStrom but for fun you can install the 68k version of UAE and see how well that one runs, LOL..

UAE used on an Amiga... isn't something new.

cdimauro Quote:

FPGAs are slow for implementing processors, if this was your idea. You need ASICs for that. Whereas they are good enough for implementing old chipsets. However that's good if you implement (or use) also the processors that those systems used. I mean: I don't see any sense on "hybrid" implementations (e.g.: a PowerPC CPU + the Amiga chipset).

FPGA: Apollo core CPU ~100MHz in FPGA
Emulation: G3@300MHz from a 3GHz modern CPU (less than 10% of performance of host CPU)
ASIC: more than 10 times the performance of emulation

The CPU magic bullet is not a FPGA bullet but a silver ASIC bullet to kill the Vampire. CPUs in FPGA primarily have two purposes.

1) develop an ASIC
2) create a low production SoC with relatively low CPU performance requirements (fewer chips & smaller boards reduces cost)

If a high performance CPU is desired then there is only one choice which is an ASIC. No modern high performance CPU means no modern Amiga. That is the reality. ASICs are cheap and likely to become cheaper with all the new fabs planned. The question for Amiga NG hardware is whether to develop native hardware in a custom ASIC with superior compatibility, use commodity foreign ASICs with compromises or throw in the towel to stop embarrassing the Amiga legacy. So far, the commodity ASIC route has been embarrassing likely missing over 90% of the Amiga market judging by evidence of 68k demand. Hyperion knows as they have numbers on 68k AmigaOS sales vs PPC AmigaOS sales. We should know when Hyperion has to pay hundreds of thousands of dollars to Amiga Corporation for the money they made off the 68k AmigaOS. Ben Hermans is a gambler that likes to bluff when everyone knows who has the better hand so he is likely to lose everything.

Last edited by matthey on 23-Aug-2022 at 08:41 PM.

@matthey

Quote:

matthey wrote: cdimauro Quote: FPGAs are slow for implementing processors, if this was your idea. You need ASICs for that. Whereas they are good enough for implementing old chipsets. However that's good if you implement (or use) also the processors that those systems used. I mean: I don't see any sense on "hybrid" implementations (e.g.: a PowerPC CPU + the Amiga chipset). FPGA: Apollo core CPU ~100MHz in FPGA Emulation: G3@300MHz from a 3GHz modern CPU (less than 10% of performance of host CPU) ASIC: more than 10 times the performance of emulation

It depends on what you need to achieve.

As I've said before, IF you need to reproduce an entire system, especially having full compatibility as the key goal, then emulation is slow because the CPU has to do too many things. And usually using only a single core.

But if you just need to reproduce only the CPU, then the emulation path gives very good performances, not so distant from the original system.
Examples: Apple's Rosetta (PowerPC emulated on Intel) and Rosetta2 (Intel emulated on ARM).
On the Amiga side we have Michal's Emu68 which clearly proves it.
Quote:

The CPU magic bullet is not a FPGA bullet but a silver ASIC bullet to kill the Vampire.

That belongs to the first case: reproducing an entire system.
Quote:

CPUs in FPGA primarily have two purposes. 1) develop an ASIC 2) create a low production SoC with relatively low CPU performance requirements (fewer chips & smaller boards reduces cost) If a high performance CPU is desired then there is only one choice which is an ASIC. No modern high performance CPU means no modern Amiga. That is the reality.

Sorry, but there cannot be a modern Amiga by definition: Amiga is very old technology which has nothing modern and cannot be modernized without throwing away everything.

But if you're talking from a purily performance perspective (so keep all defects of the Amiga platform: hardware & software), then yes: I agree with you. An ASIC is need to have an Amiga which is comparable to modern systems.

So, it's "just" a matter to decide what's the goal for the post-Amiga market.
Quote:

ASICs are cheap and likely to become cheaper with all the new fabs planned. The question for Amiga NG hardware is whether to develop native hardware in a custom ASIC with superior compatibility, use commodity foreign ASICs with compromises or throw in the towel to stop embarrassing the Amiga legacy. So far, the commodity ASIC route has been embarrassing likely missing over 90% of the Amiga market judging by evidence of 68k demand. Hyperion knows as they have numbers on 68k AmigaOS sales vs PPC AmigaOS sales. We should know when Hyperion has to pay hundreds of thousands of dollars to Amiga Corporation for the money they made off the 68k AmigaOS. Ben Hermans is a gambler that likes to bluff when everyone knows who has the better hand so he is likely to lose everything.

I fully agree.

Quote:

ppcamiga1 wrote: Real "classic" amiga have to ...

The term 'Classic Amiga' was introduced by AOS4 evangelists to alienate the Amiga and made the AmigaOne look like a successor, when the original brand name was still held hostage by Amiga Inc.

So as long as it has not the name Amiga on it, it's is not an Amiga.

If you want to speak about 'Amiga Clones' like Minimig or Vampire you can't use the word classic.
If you want to speak about 'Amiga Software Compatibles' you can mention the Dragon because it was not a clone.
You can mention computers using AmigaOS4 with AmigaOS3 compatibility (AmigaOS4) or Amiga like OS (MorphOS) with AmigaOS3 compatibility.

You can even mention Amiga compatible OSes like 68k AROS or API compatible OSes like AROS x386 or UAE systems running AOS3/AOS4 but it is not 32 bit that makes AROS a 'classic Amiga', really that's stupid.

cdimauro Quote:

It depends on what you need to achieve. As I've said before, IF you need to reproduce an entire system, especially having full compatibility as the key goal, then emulation is slow because the CPU has to do too many things. And usually using only a single core. But if you just need to reproduce only the CPU, then the emulation path gives very good performances, not so distant from the original system. Examples: Apple's Rosetta (PowerPC emulated on Intel) and Rosetta2 (Intel emulated on ARM). On the Amiga side we have Michal's Emu68 which clearly proves it.

Hmm. CPU emulation gives "very good performance" at less than 10% of the host CPU performance...

cdimauro Quote:

Sorry, but there cannot be a modern Amiga by definition: Amiga is very old technology which has nothing modern and cannot be modernized without throwing away everything.

Yet FPGA devices have compatible modernized chipsets that do not seem to be a limitation at all. The biggest limitation to running modern software continues to be lack of CPU performance (and too high of cost reducing the user base). WinUAE has better CPU performance with emulation and worse emulation of the chipset but still poor performance compared to the monster host CPUs and this is not a compiler target for development. Just like with CBM not upgrading the performance of CPUs leaving the Amiga as an under powered CPU platform, the same is true for the 68k Amiga today except it is now more over priced and less competitive in performance/price. The majority of performance comes from the CPU and the quickest way to modernize a 68k Amiga continues to be with better CPU performance. There is no other change that could be made which would do more to allow modern software than increasing CPU performance. Limiting CPU performance to less than 10% of a modern high performance CPU using emulation is more of a handicap to running modern software than any Amiga chipset limitations. Even planar graphics is not as much of a handicap for modern software as a CPU at less than 10% of modern high performance CPU performance. Doom and Quake play with better fps on a 68060 CPU with AGA than a 68030 or even 68040 with RTG chunky and VRAM. Modern software is even more CPU intensive. CPU performance was the biggest advantage of NG Amiga like systems over the 68k Amiga to run modern software but now that better PPC performance can be found in an emulator, it is obvious that PPC performance is closer to older 68k CPU performance than modern CPU performance but without as much 68k Amiga compatibility.

Last edited by matthey on 24-Aug-2022 at 12:24 AM.

@matthey

Quote:

matthey wrote: cdimauro Quote: It depends on what you need to achieve. As I've said before, IF you need to reproduce an entire system, especially having full compatibility as the key goal, then emulation is slow because the CPU has to do too many things. And usually using only a single core. But if you just need to reproduce only the CPU, then the emulation path gives very good performances, not so distant from the original system. Examples: Apple's Rosetta (PowerPC emulated on Intel) and Rosetta2 (Intel emulated on ARM). On the Amiga side we have Michal's Emu68 which clearly proves it. Hmm. CPU emulation gives "very good performance" at less than 10% of the host CPU performance...

You can take a look at Michal's Emu68 running on PiStorm with a lower-end RPi3B+ to see how good an emulator can perform when it has to focus only on pure CPU emulation and not the chipset.
Quote:

cdimauro Quote: Sorry, but there cannot be a modern Amiga by definition: Amiga is very old technology which has nothing modern and cannot be modernized without throwing away everything. Yet FPGA devices have compatible modernized chipsets that do not seem to be a limitation at all. The biggest limitation to running modern software continues to be lack of CPU performance (and too high of cost reducing the user base).

The CPU is only one side of coin / problems.

The Vampire, which has the most powerful Amiga chipset + RTG implementation, struggles with high-resolutions and color depths. Not even 720p is an option; in fact, Gunnar recommends using it at 1/4 FullHD (960x480).

Whereas I use WinUAE at FullHD since years and only because my monitor doesn't support 1440p o UltraHD.
Quote:

WinUAE has better CPU performance with emulation

It's the exact opposite: CPU performance on WinUAE is limited because it has to emulate both CPU and chipset. And the more accurate is the chipset emulation, the less performance will have the CPU emulation (obviously).

This is the biggest problem when you have to emulate an entire system, and not just the CPU.
Quote:

and worse emulation of the chipset

Sorry, I don't get it: WinUAE has the best chipset emulation. What's worse there?
Quote:

but still poor performance compared to the monster host CPUs and this is not a compiler target for development.

See above: that's exactly my point!

If you have to emulate an entire system, then the CPU performance is poor. That's the WinUAE case.

Whereas if you have to only emulate the CPU, then you get very good performances. That's the Emu68 case. But you can also take a look at OS4's Petunia or MorphOS's Trance JIT.
And it can be done even better, because more optimizations are possible.
Quote:

The majority of performance comes from the CPU and the quickest way to modernize a 68k Amiga continues to be with better CPU performance. There is no other change that could be made which would do more to allow modern software than increasing CPU performance. Limiting CPU performance to less than 10% of a modern high performance CPU using emulation is more of a handicap to running modern software than any Amiga chipset limitations.

CPU performance can be easily increased if the emulator should not fully emulate the system.

Which is also possible for the Amiga, IF the focus is on purely o.s.-friendly applications, which I assume are the ones that you were talking about (the original Amiga games don't need big resources to run).
Then a super light-weight Amiga emulator is also possible (but someone has to write it ) which exploits the host CPU power (using also 2 or 3 cores to offload some tasks) and getting very good performances (running 68K code).
Quote:

Even planar graphics is not as much of a handicap for modern software as a CPU at less than 10% of modern high performance CPU performance.

Well, modern CPUs like my Alder Lake have specific acceleration for planar graphics. With AVX-512 there are some instructions which implement bitblit operations, just to give the more notable example in the emulation side (WinUAE could get a nice boost for the Blitter implementation for big blits using the Blitter immediate mode).
Quote:

Doom and Quake play with better fps on a 68060 CPU with AGA than a 68030 or even 68040 with RTG chunky and VRAM.

That's because the latter had to deal with the CPU vs VRAM connection. Packed/chunky isn't relevant here.
Quote:

Modern software is even more CPU intensive. CPU performance was the biggest advantage of NG Amiga like systems over the 68k Amiga to run modern software but now that better PPC performance can be found in an emulator, it is obvious that PPC performance is closer to older 68k CPU performance than modern CPU performance but without as much 68k Amiga compatibility.

Indeed, but do you want to keep full compatibility with the Amiga? Or limited compatibility (see above: only o.s.-friendly applications) is your target?

For full compatibility you've a big price to pay.

@OneTimer1

What makes classic Amiga classic Amiga is 32 bit big endian cpu.
It is last and only thing from original Amiga hardware that is still important
and should be preserved.
Classic Amiga have to have 32 bit big endian cpu.

aros x86 is worth nothing crap that is not binary and not source compatible
and is not modern.

@ppcamiga1

The custom chips made the Amiga special. RTG and 060 was nice but lost some of the magic to get acceptable performance for productivity programs.

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

@ppcamiga1

PPC can and will be replaced by Arm or x86-64.

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

@ppcamiga1

Quote:

ppcamiga1 wrote: @OneTimer1 What makes classic Amiga

Again? Another PARROT here!

There's NO "classic" Amiga. There were and there's only AMIGAs.
Quote:

classic Amiga is 32 bit big endian cpu. It is last and only thing from original Amiga hardware that is still important and should be preserved. Classic Amiga have to have 32 bit big endian cpu.

Chapter 1 pag.1 (and 2) of Commodore's Amiga Hardware Reference Manual gives another definition: an Amiga is made of a 68K processor and the Amiga chipset.

Are you better than Commodore on defining how its machines are made?
Quote:

aros x86 is worth nothing crap

Again, you're an AROS hater. So much envy, I've to say, or... just fear for this competitor.
Quote:

that is not binary

Guess what: it was already expected! It's for... ehm... x86: which is A LITTLE bit different from a 68K. Like ANY other processor (PowerPC included).
Quote:

and not source compatible

That's plainly FALSE. As usual, when you talk of AROS.
Quote:

and is not modern.

Is there ANY post-Amiga/like o.s. which is modern?

@matthey

Quote:

FPGA: Apollo core CPU ~100MHz in FPGA Emulation: G3@300MHz from a 3GHz modern CPU (less than 10% of performance of host CPU)

With all respect, you have compared apples to oranges here. Apollo core is m68k architecture, g3 is ppc. The comparison you have proves only that ppc emulation on x86_64 architecture is inefficient.

On the contrary, m68k emulation on aarch64 can be much more effective. From Emu68 - up to 1 MIPS per 1 MHz, depending on the code.

@BigD

Quote:

PPC can and will be replaced by Arm or x86-64.

Yes, but you are not talking about anything that will be happen soon.
while you sit around waiting, you can be enjoying AmigaOS4 running on a real PowerPC.

Or enjoy AROS with its limitations, regarding backward combability on x86-64.

So many things will never be recompiled, source code is lost, there is slight hope some decompile some of the old programs and write patch, or recreate some of it. but doing so requires huge dedication, for most things that’s not really an option.

It seems to me almost every path has been taken, there little has not been tried.

We turned every stone, image format, ABI’s and API’s and emulation approaches, full and practical. There is not a lot to be debated, except going in circles.

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

Can we rename amigaworld.net to cdimauroworld.net please?

_________________
Sent from my Quantum Computer.

@cdimauro

Quote:

The Vampire, which has the most powerful Amiga chipset + RTG implementation

No, the V4 maybe, but no Vampire.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@NutsAboutAmiga

I'm going to talk about numbers in this post I've done a little research over the past month or so and have come up with some conclusions:

1. The Apollo Core team has sold at least 10,000 units. This number may be a little bit understated as they are constantly selling units

2. The Amiga 500 Mini has has probably sold less than100,000 units. They have not sold as many as they anticipated. It seems to be selling very well in the UK, but not so much elsewhere. The spoiler seems to be theC64Mini/Max combo. Somebody developed a PCUAE setup for TheC64 system which among other things allows people to emulate the Amiga on the C64. So many people have not bothered to buy a 500 Mini.

3. As mentioned above many people are using the C64 Mini for getting their Amiga fix. As the C64 Mini/Max combo has sold around a half million units, there are many users on that platform. I estimate about a third to half use PCUAE.not only to emulate the Amiga but for some other stuff swell.

4.There are probably about 4000 users of NG Amigas. So the market share of NG Amigas is less than two per cent. I have deliberately not included general emulation users because there is no way to make a rational estimate of their numbers.


One thing I should note is that most of the PCUAE users as well as many of the emulation users use OS1.3 rather than 3.1. The reason for this is simple. Most Amiga games were designed to run on1.3. They simply run better in that environment.

@Neuf

I don’t know the market share, and don’t care. Frankly I’m enjoying my system, have no time argue about numbers.

But what I know there is lot people whit RTG Amiga Systems, people have upgraded with Amiga Systems, that have sound cards, lots of systems FPGA systems, many of system are fully 16bit enabled using AHI. If people won’t listen to 16bit CD quality they have hardware but they can’t. because of some gatekeepers.

Gatekeepers are people who fight upgrades, open API’s and portability or do everything to hold Amiga back, that makes me sad.

I written many times, I think it’s good every time someone buys Warp060 or Applo or Vampire, the more people who have upgraded there Amiga System the more people will demand better software. I believe people who invest are not the typically gamers.

Gatekeeps who use none standard oppcodes, hardcode devices, general prevent people from using hi resolution and good audio options, will get lots of nagging and complains. When choice not support open standards, they will have to suffer the criticism.

I know AHI too slow for Amiga500, but plugins enable you not worry about that, you have PAULA player, or AHI player, or MIDI player, working on the same game, or demo, if people used plugins, that be no issue at all even on slow 7Mhz Amiga500.

Now I do applaud people who do upgrade their software, and do listen to complaints, there a few people care about there software, and what people thing about what they produce. Sadly now everyone does.

Last edited by NutsAboutAmiga on 27-Aug-2022 at 08:12 PM.
Last edited by NutsAboutAmiga on 27-Aug-2022 at 08:02 PM.
Last edited by NutsAboutAmiga on 27-Aug-2022 at 07:58 PM.
Last edited by NutsAboutAmiga on 27-Aug-2022 at 07:57 PM.
Last edited by NutsAboutAmiga on 27-Aug-2022 at 07:47 PM.
Last edited by NutsAboutAmiga on 27-Aug-2022 at 07:45 PM.

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

cdimauro Quote:

The CPU is only one side of coin / problems. The Vampire, which has the most powerful Amiga chipset + RTG implementation, struggles with high-resolutions and color depths. Not even 720p is an option; in fact, Gunnar recommends using it at 1/4 FullHD (960x480).

Modern graphics resolutions and color depths use a lot of memory bandwidth and a SoC shares memory bandwidth between the CPU and GPU. Vampire/Apollo-core hardware has more CPU memory bandwidth than a X1000.



A1-x1000 CPU PA Semi PA6T-1682M 1800Mhz, 2MB L2, 64K L1 CACHE
Total RAM: 2 GiB External Bus (FSB) Speed: 900 MHz
Radeon HD 7700 2GB DDR5

.RAM Disk:> bustest size=8m fast
BusSpeedTest 0.19 (mlelstv) Buffer: 8388608 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
fast $5E298000 readw 13.1 ns normal 152.4 * 10^6 byte/s
fast $5E298000 readl 14.0 ns normal 285.6 * 10^6 byte/s
fast $5E298000 readm 17.9 ns normal 223.4 * 10^6 byte/s
fast $5E298000 writew 12.7 ns normal 157.0 * 10^6 byte/s
fast $5E298000 writel 14.1 ns normal 283.7 * 10^6 byte/s
fast $5E298000 writem 17.2 ns normal 232.2 * 10^6 byte/s

I wouldn't be surprised if this beats an X5000 too. The saving grace for NG hardware is that there is another memory controller, memory and separate memory bandwidth for the GPU. More SoC bandwidth is possible and likely easy for the Vampire but a low end FPGA is used to reduce costs. High resolution are possible with the current bandwidth but either colors or refresh rate needs to be reduced just like many of the early RTG graphics cards for the Amiga.

cdimauro Quote:

Indeed, but do you want to keep full compatibility with the Amiga? Or limited compatibility (see above: only o.s.-friendly applications) is your target? For full compatibility you've a big price to pay.

Chunky graphics modes can be chosen which bypass any planar deficiency even though planar modes are available for compatibility. The Amiga chipset with enough memory bandwidth is not a bottleneck to the CPU either. CPU performance is a bottleneck in FPGA and chipset performance is a bottleneck with emulation but those are due to deficiencies of low cost FPGAs and emulation respectively. Where is full compatibility a problem for an ASIC?

michalsc Quote:

With all respect, you have compared apples to oranges here. Apollo core is m68k architecture, g3 is ppc. The comparison you have proves only that ppc emulation on x86_64 architecture is inefficient.

I'm afraid that my comparison doesn't even prove that PPC emulation on x86-64 is inefficient as someone could come along and write a more efficient PPC emulator for x86-64.

michalsc Quote:

On the contrary, m68k emulation on aarch64 can be much more effective. From Emu68 - up to 1 MIPS per 1 MHz, depending on the code.

With that kind of emulation performance there really isn't a need for an intermediate language. The executable code is super small due to the 68k code density too. Maybe we could use 68k Amiga binaries as portable modules to save networking and disk bandwidth. We could call it AmigaEverywhere and everyone will buy it and switch to the Amiga because it uses the "Amiga" name. The binary copy bandwidth savings should offset the emulation overhead, right?

I appreciate the effort but I don't see where an emulated Amiga does much more for the platform that what we have now. Yea, a few more ex-Amiga users that can't afford used $50 x86-64 hardware may buy a $15 RPi Zero 2 W but I don't see a big draw. I hope I'm wrong.

Real affordable hardware that is the ultimate 68k retro gaming, hobbyist, hackable hardware that replaces the Raspberry Pi is what is needed. The Amiga 1200 has never looked better as the following video shows.

Solas RGB LED Controller for the Amiga (1200)
https://youtu.be/bykoQt1D_KY?t=117

Now add 68k CPU performance that is better than emulation and Amiga chipset performance that is better than FPGA and we can forget the Amiga gimmicks, unless it is an LED add on board for Amiga hardware that is competitively priced with RPi hardware.

Last edited by matthey on 28-Aug-2022 at 01:27 AM.
Last edited by matthey on 28-Aug-2022 at 01:24 AM.
Last edited by matthey on 28-Aug-2022 at 01:24 AM.

@DiscreetFX

Quote:

DiscreetFX wrote: Can we rename amigaworld.net to cdimauroworld.net please?

Care to explain why, dear EAB "moderator" (sich!) Pyromania?

The Censorship runs strong in you and in your "family" (cit.)


@kolla

Quote:

kolla wrote: @cdimauro Quote: The Vampire, which has the most powerful Amiga chipset + RTG implementation No, the V4 maybe, but no Vampire.

Vampire Accelerator cards


@matthey

Quote:

matthey wrote: cdimauro Quote: Indeed, but do you want to keep full compatibility with the Amiga? Or limited compatibility (see above: only o.s.-friendly applications) is your target? For full compatibility you've a big price to pay. Chunky graphics modes can be chosen which bypass any planar deficiency even though planar modes are available for compatibility. The Amiga chipset with enough memory bandwidth is not a bottleneck to the CPU either.

Indeed. Since long time graphics moved to packed (not even CLUT anymore) because there was/is plenty of bandwidth, so it doesn't matter anymore thinking about saving some of it using planar formats.
Quote:

CPU performance is a bottleneck in FPGA and chipset performance is a bottleneck with emulation but those are due to deficiencies of low cost FPGAs and emulation respectively. Where is full compatibility a problem for an ASIC?

The problem stays on realizing the required mask(s) which need a lot of money, and if you find issues / bugs or missing something for the current ASIC, then you need again other masks.

ASIC is very very good for "definitive" chips, but aren't recommended for retroplatforms like the Amiga, where compatibility enhanced and still enhances over the time.

Maybe it could be used for a processor (where issues could be addressed in software, via SetPatch / new or fixed libraries, or microcode), but not for the chipset.
Quote:

Quote: michalsc [quote] With all respect, you have compared apples to oranges here. Apollo core is m68k architecture, g3 is ppc. The comparison you have proves only that ppc emulation on x86_64 architecture is inefficient. I'm afraid that my comparison doesn't even prove that PPC emulation on x86-64 is inefficient as someone could come along and write a more efficient PPC emulator for x86-64.

As I've already said, Apple did it with its first Rosetta, which run PowerPC code (included the Altivec one) on new Intel-based MacBooks at around half the speed of the equivalent PowerPC Macs.

So, it's definitely possible doing the same for WarpOS, MorphOS, and OS4. But someone has to do it, of course.
Quote:

michalsc Quote: On the contrary, m68k emulation on aarch64 can be much more effective. From Emu68 - up to 1 MIPS per 1 MHz, depending on the code. With that kind of emulation performance there really isn't a need for an intermediate language. The executable code is super small due to the 68k code density too.

No, it's not code: the 68k code is just like ordinary data for the ARM processor which has to interpret it.

When the real, native, code is running, its AArch64 instructions which are running and consuming instruction cache size & bandwidth.
Quote:

Maybe we could use 68k Amiga binaries as portable modules to save networking and disk bandwidth. e could call it AmigaEverywhere and everyone will buy it and switch to the Amiga because it uses the "Amiga" name. The binary copy bandwidth savings should offset the emulation overhead, right?

Is it really a bottleneck on the SSD-NVMe PCI-Express 4/5 & 1-10Gb/4G/5G network age?
Quote:

Real affordable hardware that is the ultimate 68k retro gaming, hobbyist, hackable hardware that replaces the Raspberry Pi is what is needed. The Amiga 1200 has never looked better as the following video shows. Solas RGB LED Controller for the Amiga (1200) https://youtu.be/bykoQt1D_KY?t=117 Now add 68k CPU performance that is better than emulation and Amiga chipset performance that is better than FPGA and we can forget the Amiga gimmicks, unless it is an LED add on board for Amiga hardware that is competitively priced with RPi hardware.

How could you add a performant 68k? ASICs. Which is super-expensive for a niche market like this...

aros clowns as usually do not accept reality.
"classic" Amiga works on 32 bit big endian cpu.
It is one and only thing from original hardware that is still important.
Only on 32 bit big endian cpu os may be source and binary compatible
with good old Amiga OS from Commodore and 68k Amiga apps.
I include Amiga NG in "classic" because it is something like classic on steroids.
Amiga like from Commodore only better because many times faster.

aros x86 is worth nothing crap that is not binary and source compatible.
and at the same time more than 25 years behind oses from pc.
aros x86 is not binary and source compatible and has not memory protection.
It is incredible stupid. Nobody need this shit.

szulc di mauro and all rest start working on something usefull.
made good amiga gui open source replacement then port it to unix.

@matthey

You are making some extremely wild clams, while people are comparing about not being able to use 1080p modes on workbench on the Vampire.

I can tell you one thing, Youtube video, where your selecting different topic, looks slow, you practically seeing gadgets being redrawn, top / down. Is this the ultra-fast Amiga your talking about?????

Last edited by NutsAboutAmiga on 28-Aug-2022 at 07:18 PM.
Last edited by NutsAboutAmiga on 28-Aug-2022 at 07:18 PM.

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