@Karlos

Quote:

It's nice there are some new hardware available, but PowerPC is dead. Just as dead as 68K.

Yes, but it would be kind of cool to resurrect the latter. To give it the chance it never had. To right the wrong.

_________________
All the way, with 68k

@agami

An updated Amithlon, with a new kernel and refreshed ABI for driver development could be perfect OS 3.2 platform.

_________________
Doing stupid things for fun...

Karlos Quote:

An updated Amithlon, with a new kernel and refreshed ABI for driver development could be perfect OS 3.2 platform.

How can emulation ever be "perfect" for the Amiga? What room does that leave for modern 68k hardware?

@Karlos

Amithlon is without chipset. I am not sure if that is really perfect for 3.2. 3.2 is updated 3.1 and that means no RTG.

I think you mix something here... Amithlon was not a real emulation like UAE emulating amiga hardware (as I understand it)

Last edited by OlafS25 on 27-Apr-2022 at 10:57 AM.

@matthey

Are you sure? Amithlon is no emulator in my view. From my limited knowledge, at least not in the sense we understand it. It only translates 68k in X86 like current emu68 project based on ARM (PiStorm)

@ppcamiga1

three times faster than WinUAE with activated JIT on a new i9?

That sounds a little strange, i9 means new hardware and that should be faster everywhere compared to old G4 or G5 Macs. Something sounds wrong there.

@Karlos

relevant is what people can do with a system. PPC is dead end as desktop hardware and it is no problem to emulate 68k on X86 or X64 systems. On my own PC (i5) I can run in emulation everything that is available on 68k.

As I wrote... there are two completely different branches... the traditional 68k branch with V4 and PiStorm as modern extensions and what people call "NG" but in reality is at best "Classic+" because it inherited most of the limitations of the past and is far from what mainstream platforms can offer. On Aros there is at least a experimental version with 64bit and multicore but still missing full memory protection that is today needed. And all niche platforms have common problems... missing developer ressources and that means not enough drivers and modern software. Except you find a billionaire investing lots of money there is no real chance to overcome this. Because of that I always was for concentrating on 68k as a retro platform because I did not see a realistic chance to compete with the big platforms. AxRuntime project from Deadwood is first project that convinces me because it potentially means you can do a linux distribution that includes many amiga technologies and amiga desktop offering look&feel of amiga and modern base, drivers and software from Linux.

Traditional 68k should stay what it is... retro platform offering people the chance to see "what could have been" and dream back or attracting hardware nerds who f.e. like FPGAs. The idea to compete with a super 68k with modern PCs sound unrealistic to me, and be it because of missing modern software and developers.

@OlafS25

I'm completely aware of the lack of legacy Amiga native hardware in Amithlon, it's not as if I've never used it before. My point is that it was, and on compatible hardware still is, the fastest AmigaOS 68K experience. For people that need native chipset support it's not an option. For people that are happy running with RTG and don't mind that it won't run anything depending on the legacy hardware, it's the closest thing you'll ever have to a GHz class 68K machine.

It was cancelled before the full potential could ever be realised. Given that classic Amiga OS had no multicore support, a flight of fancy I had at the time was whether or not a second core
could be dedicated emulating legacy hardware.

Even from a UX perspective, if you've ever used Amithlon, you'll know it's completely unlike using UAE. As much of a pain as it could be to get working, once it is, it isn't an emulation experience in the sense UAE is. And certainly no less "real" than any NG system.

Last edited by Karlos on 27-Apr-2022 at 01:51 PM.

_________________
Doing stupid things for fun...

@Karlos

perhaps we will get something like that somewhen on ARM (RPi) because Michal (emu68) is planning to do that using emu68 as emulation layer and you can boot in Aros 68k or AmigaOS. Amithlon for ARM so to say. Amithlon is no longer supported, you would need a update of the kernel to use it on modern hardware.

@matthey

There's no conflict as I see it. Some people will always want to run 68K on metal and those people will. The only difference with a solution like Amithlon is how close to the metal the 68K code ever gets and how compatible it is or isn't with the legacy custom hardware. Regardless of which, the fact both run 68K code means more interest in 68K, which is no bad thing in my book.

_________________
Doing stupid things for fun...

@OlafS25

Quote:

three times faster than WinUAE with activated JIT on a new i9? That sounds a little strange, i9 means new hardware and that should be faster everywhere compared to old G4 or G5 Macs. Something sounds wrong there.

It's probably because he's either outright making it up to bolster the case for his obsolete hardware preferences, doesn't have the PC hardware he thinks he has, has it misconfigured, or has some particularly pathological UAE configuration.

People like to feel validated in their choices. Which is stupid when it comes to something as subjective as this. If someone prefers a physical 68000 1MB chip RAM OCS machine running 1.3, there's no need to justify that to anyone else, no need to decry alternatives. But you do need to accept that choice has limitations that alternatives may not.

I'm not anti PPC; I have PPC hardware and I've had a lot of fun with it. I'm not anti OS4; I was a contributor and a tester, up to the point where my A1 died. But I recognise the limitations of those choices. Obscure, expensive hardware and performance capped by an architecture that has been shelved years ago. Unlike the 68K, emulation of that hardware is the path less well trodden. In my experience, emulation of PPC is useable but somewhat lacklustre. By comparison, JIT emulation of 68K is like a gift from aliens: breathtaking performance and compatibility.

Last edited by Karlos on 27-Apr-2022 at 03:20 PM.

_________________
Doing stupid things for fun...

@Karlos

Quote:

By comparison, JIT emulation of 68K is like a gift from aliens: breathtaking performance and compatibility.

What do you guys think should a 68k ASIC have to match the performance of WinUAE JIT on a i9-12900KS(Top 1 single-threaded performance on Passmark)? I'm thinking that could be the baseline if ever someone is going to plan to implement a real modern 68k processor because there is no escape of someone comparing it with WinUAE(People's reason of not buying Vampire is WinUAE just beat it in performance).

@AmigaNoob

UAE has never stopped me buying Amiga hardware. What does put me off is the terrifying fragility of the hardware I'd be putting the new equipment in. The fastest ASIC solution you could dream of is no use if the machine it is added to dies a death.

Careful imaging of my original hard disks and some tweaks to config means I can use use UAE to scratch an itch. But, at the end of the day, it's subjectively not the same. The 68K emulation is great but the PPC emulation not so much. Floating point in particular is a lot slower than the 603 240MHz in my A1200 and on top of that, for whatever reason, IO originating from the PPC side is much worse. In the end I had to fool it into thinking my install was from a CSPPC with SCSI attached storage for OS4 to work and it's still significantly slower than the BlizzPPC. No doubt the age of the PC is a factor. I had hoped that I could use the emulation to further work on this https://github.com/0xABADCAFE/proto-bppc-scsi-os4 as my real hardware isn't likely to live long enough.

Last edited by Karlos on 27-Apr-2022 at 06:44 PM.

_________________
Doing stupid things for fun...

@matthey

Quote:

How can emulation ever be "perfect" for the Amiga?

CPU emulation is already pretty darn perfect! What sometimes holds uae variants back is emulation of the chipset when simulating “hardware banging” effects and their almost analogue nature. When left to only do emulation of CPU, they leave “real” 68k and AC68080 in the dust. Bonus feature - with software emulation of CPU, instruction sets and features can be configured for needs of the user and the software on-the-fly.

Quote:

What room does that leave for modern 68k hardware?

Syntax error, oxymoron detected, “modern” and “68k” are mutually exclusive!

The only way you can get what you want is to do just like everyone else, and do it yourself.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

I think we can grant him the various FPGA implementations and any software emulation on a board you can plug in (e.g. ARM) as being "modern 68K" for the purposes of this discussion. I don't think he's implying that 68K itself is modern.

_________________
Doing stupid things for fun...

@agami

Quote:

Yes, but it would be kind of cool to resurrect the latter. To give it the chance it never had. To right the wrong.

Feels strange to apply ethics to computer technology

_________________
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

@Trixie

Quote:

Trixie wrote: Feels strange to apply ethics to computer technology

To me it is quite natural. Computer technology is made by people, for people. The rate at which humans can solve real-world problems, is directly linked to the availability, and affordability of computer technology.

So when a company like intel (prior to AMD's Ryzen come-back) decides not to release all the available improvements to their CPUs in a given year, and decides that it is better for their shareholders if they spread them out over 3 years, they held back humanity's capacity of problem solving and progress, for personal gain.

You see. Computer technology, much like any technology, is very easily linked to ethics.

_________________
All the way, with 68k

@OlafS25

Quote:

And all niche platforms have common problems... missing developer ressources and that means not enough drivers and modern software. Except you find a billionaire investing lots of money there is no real chance to overcome this.

It doesn't need an actual billionaire, though of course more money is always better.

If I had $100 million, I would happily put 10% of that into my plan for bringing to market a viable and sustainable 3rd commercial personal computing platform, for which I would strongly consider, and prefer, newly designed and modernized 68k-based silicon.

Now I'm not saying that all it takes is $10M to make sure all aspects and areas are addressed. That is just the staring capital. It would require 3x to 5x before a 3rd commercial personal computing platform would reach commercial viability and be considered self-sustainable. The extra $20M-$40M would be sourced from other investors.

_________________
All the way, with 68k

ppc is dead - stop this crap
Amiga like solutions on pc compete with windows, mac os x, linux.
It should be as good as windows, mac os x, linux.
As long as there is nothing such on x86, ppc is cool.

Last edited by ppcamiga1 on 28-Apr-2022 at 07:30 AM.

68k.
After almost fifteen years since GvB start natami thing
it is still not fast and comfortable as ppc.
060 75 MHz with not compatible FPU and MMU. Without 3D.
Yes it has 100 MHz RAM and 2D on level of cheap pc from 1995.
but it is still not enough.