@agami

Ever been to RISC-V tracks at open source conferences?

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

Quote:

kolla wrote: Or one can just drop the whole idea of new 68k ASIC, save millions that can be invested elsewhere, and instead use a solution that is provided for free and is open source and runs on the most popular series of ARM boards available.

I'm certainly glad that a substantial amount of people don't think along these lines.
To create a platform that is consequential, and is commercially viable, requires investment in the millions of dollars. Certainly it can be ARM-based, but whether it happens or not, there are some strategic advantages to resurrecting the 68k and modernizing it.


Quote:

kolla then wrote: Only drawback is that it will be wee bit slower than running native - but who cares?!

Maybe people in Norway with their free (or near enough) fjord-powered electricity have the luxury of not caring, but most people should care about computing efficiency.

Last edited by agami on 19-Sep-2022 at 07:54 AM.

_________________
All the way, with 68k

@agami



Quote:

These needn't be part of the first iteration of a new 68k ASIC. The potential project runners would/should design a roadmap of SKUs taking the 68k from its current 32-bit, single-threaded/single-core, CPU only design (greater backward compatibility), through to an integrated SoC and extended to 64-bit, multi-threaded/multi-core design (greater future-proofing).

If you want modern memory access, they most certainly are needed for the first iteration. DDRx controllers don't grow on trees.
Ok, so you want a single core only to begin with? Because if you want multiple designs after the first iteration, that's new ASIC designing, again. You better have very deep pockets.

Quote:

It is certainly the designer's prerogative, and of those footing the bill. While a new and extended 68k CPU architecture should not be designed by a committee, it would be advisable for the potential project leaders to spend some time consulting with certain individuals in the existing 68k market (major stakeholders). Which consulting insights are taken into account and to what degree, would and should be ultimately up to the designer(s) and the investor(s).

Again, who decides what! lol who decides who's right/wrong or what direction. Spend more money finding who you agree with? somone else?
Gunnar already tried in FPGA, lots think it's not the right way to go.

I hate to be the downer here, I love the idea of a new 68K, but (yep, it was coming) it's just not practical, cost wise.
There is too much to be done to get it modern. IF we are talking just keep it the same on a smaller process node, it's not exactly going anywhere new, or improving, is it. It will need logic re-drawn for new socket and IP related hurdles would be a thing too. I almost forgot about the IP. Ergh.

If you want to get moving on a cheap, already proven, modern design, well.....there are SoC already to go, but yes, PORTING to it will be required, just as any new enhanced 68K special-sauce new thing would as well.

Last edited by SHADES on 19-Sep-2022 at 11:55 AM.
Last edited by SHADES on 19-Sep-2022 at 11:54 AM.
Last edited by SHADES on 19-Sep-2022 at 11:36 AM.
Last edited by SHADES on 19-Sep-2022 at 07:56 AM.

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

@kolla

Quote:

kolla wrote: @agami Ever been to RISC-V tracks at open source conferences?

I have not.

I haven't really been anywhere in 2.5 years, and it might be another 12-18 months before I go anywhere.

_________________
All the way, with 68k

@cgutjahr

Quote:

I've never heard of "Wavem" studios. [...] A trailer for their upcoming video "Amiga - Alive and Kicking" doesn't make me go for my wallet immediately - maybe I'll wait for the extended edition. Note how the Youtube video was posted in July and has exactly one comment - from a spammer.

They're legit.

Amiga Alive and Kicking has actually been already delivered to backers, I have no idea about the public version though, which I assume was planned too.

@Turrican3

Amiga Alive and Kicking is available right now to rent and buy on Amazon Prime Video. Great documentary about the Amiga’s next chapter.

Last edited by DiscreetFX on 19-Sep-2022 at 02:40 PM.
Last edited by DiscreetFX on 19-Sep-2022 at 11:32 AM.
Last edited by DiscreetFX on 19-Sep-2022 at 11:31 AM.

_________________
Sent from my Quantum Computer.

@DiscreetFX

Quote:

Amiga Prime Video

That's gotta be an amazing typo.

SHADES Quote:

Again, who decides what! lol Gunnar already tried in FPGA, lots think it's not the right way to go.

Putting Gunnar in charge would not be a good idea. It reminds me of the time on the Natami forum when Gunnar asked about adding more 68k registers and the consensus of developers was that it was not necessary and would take away from the simple design. Gunnar was flustered, continued to argue for it and eventually did it anyway, in an even bigger way with many banks of registers. The additional registers are awkward, difficult for compilers to use, increase decoding overhead, increase power and resource requirements and make code density worse while likely providing minimal performance. The original 68000 ISA was decided by a team of developers at Motorola not by one person even though there were leaders with a vision of what they wanted it to become and this had requirements.

SHADES Quote:

If you want modern memory access, they most certainly are. DDRx controllers don't grow on trees. Ok, so you want a single core only to begin with? Because if you want multiple designs after the first iteration, that's new ASIC designing, again. You better have very deep pockets. [quote] Multiple cores are not much more difficult as most of the logic is cut and paste. Even simple embedded CPUs often use 2 cores. An example is the $1 RP2040 SoC with 2 cores. A first design could be single core for a variety of reasons though. SHADES [quote] It is certainly the designer's prerogative, and of those footing the bill. While a new and extended 68k CPU architecture should not be designed by a committee, it would be advisable for the potential project leaders to spend some time consulting with certain individuals in the existing 68k market (major stakeholders). Which consulting insights are taken into account and to what degree, would and should be ultimately up to the designer(s) and the investor(s).

Feedback from developers and customers about product compatibility and suitability are important for acceptance. This is where the Apollo core went drastically wrong with Gunnar dictating everything. The result is no Apollo core licenses from businesses and no compiler support for the new ISA.

SHADES Quote:

I hate to be the downer here, I love the idea of a new 68K, but (yep, it was coming) it's just not practical, cost wise. There is too much to be done to get it modern. IF we are talking just keep it the same on a smaller process node, it's not exactly going anywhere new, or improving, is it. It will need logic re-drawn for new socket and IP related hurdles would be a thing too. I almost forgot about the IP. Ergh.

There are developers who worked on modernizing the 68k Amiga 2 decades ago (MiniMig ~2005, Jeri Ellsworth Amiga SoC ~2005, Cone-A ~2005, Boxer ~1997). Many have donated their time working for free and some bought expensive materials and equipment out of pocket in the process. FPGA Projects have become more and more advanced, higher tech and more mature. FPGAs are used to design logic for ASICs but there are Amiga users who want to throw it away and use emulation instead of pulling the work and developers together to make it a reality in hardware again. It is way more practical now to make 68k Amiga hardware than it was back in 1985. The whole CPU, chipset, GPU and memory can all easily fit in an inexpensive single chip today which Jay Miner could have only dreamed of.

SHADES Quote:

If you want to get moving on a cheap, already proven, modern design, well.....there are SoC already to go, but yes, PORTING to it will be required, just as any new enhanced 68K special-sauce new thing would as well.

Transitioning to more advanced 68k hardware should be easier than transitioning to little endian ARM which really makes the most sense as it is much better supported. The 68k Amiga hardware could have very good backward compatibility while ARM needs emulation. Early PPC Amiga hardware still had a 68k CPU which retained compatibility and eased porting as parts of the AmigaOS could be run on the 68k. PPC was also big endian.

@matthey

I see you've been advocating for a modern 68K as a reasonable evolutionary path for the Amiga for a long time, but I'm still a bit confused about why you believe it would be a better choice compared to ARM (except perhaps extremely high or even 100% compatibility with native applications - but CPU is only a fraction of the Classic architecture, what should we do with the custom chips? I mean isn't it likely the whole Classic would end up being sandboxed/emulated, and if so what would be the point of keeping a 68K architecture, albeit a modernized one?)

What would the hypotetical company behind it, the developers and (arguably most of all) the end users gain from such a choice?

@AmigaOldskooler:

Quote:

The presentation is available here: https://youtu.be/ZED8_Lfdwuk?t=277

Thanks, but I was referring to the actual PowerPoint file, he announced in the video that he would post that on the net.


@Turrican3:

Quote:

They're legit.

I didn't mean to imply anything else. I was just commenting on the quality of their product.

@Turrican3

Thanx, fixed.

:)

_________________
Sent from my Quantum Computer.

@cgutjahr

Quote:

I didn't mean to imply anything else. I was just commenting on the quality of their product.

You said you had never heard about them so I thought you were somewhat implying the documentary didn't actually exist or something like that (vapourware is unfortunately quite common in the Amiga world), I apologize if I misunderstood.

Having said that, I'd say the documentary is not that bad at all. I think their previous work - The Commodore Story - is more impressive and seems (for a lack of a better wording) done in a slightly more serious/professional way... but I've seen way worse stuff in the past so I can't really complain.

@matthey

Quote:

matthey wrote: FPGAs are used to design logic for ASICs but there are Amiga users who want to throw it away and use emulation instead of pulling the work and developers together to make it a reality in hardware again. It is way more practical now to make 68k Amiga hardware than it was back in 1985. The whole CPU, chipset, GPU and memory can all easily fit in an inexpensive single chip today which Jay Miner could have only dreamed of.

I can emulate an Amiga that actually was on pretty much any HW released in the last 20 years or I can still kinda emulate it on a 20-100€ FPGA system. Results are the same.

I can emulate an Amiga that never was (and never should have been) on sub 100€ HW much better that doing it on >200€ FPGA systems.

As for running an Amiga as it should have been today, well thats not a thing (and never will be) but I can do the next best thing on 17+ year old Macs.

Anyone who thinks that anything Amiga is a big enough market to allow for an ASIC (beyond A500-in-a-joystick, maybe not even that) is delusional.

Anyone who think one could leapfrog the 68k ISA 25 years forward with just a few millions lacks some basic understanding.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@matthey

Quote:

If cheap emulation on RPi is adequate then there is no new Amiga platform, no money to be made or purpose for all these NG Amiga businesses

Find that a odd statement when just this year the A500 mini proves the emulation market can make you money.

Amikit XE according to it dev, the Pi version sold extremely well, Pimiga had 1,000s of downloads.

Then you have Pistorm, different kind of emulation market, but again its a market that done pretty well for the dev, and has massisly brought down the cost to upgrade Classic Amiga

Finally the company that owns Amiga, literally made most of its money selling Amiga Emulation software, Amiga Forever.

This is a business and money to be made in this market.

The missing link to me is combining these markets to one strategy, what if Amiga made its own Arm board like the Pi with that maybe had Pistorm features built in so it be easier to put in real classic Amiga but could also be used on its own, and to get costs down the same board released as a console version sold inside say a A1200 mini but this time featured a much more advance system like Amikit XE out of the box, with a special emulator written that allows it to access more powerful GPU features on the emulated Amiga. Combine that with a online store, that devs could post to and have the same system on Pc, Mac, Linux, Android. With competitions and active engagement with the community to make games and apps for the new system.

_________________
AmigaNG, YouTube, LeaveReality Studio

@amigang

Great ideas!

_________________
Sent from my Quantum Computer.

@agami

Well, RISC-V was topic at conferences long, long becore corona, it didn’t just pop up out of nowhere recently. My point is simply that the architecture wasn’t created in some sort of vacuum, it evolved over the years by playing ball with software developers, in particular compiler developers.

Anyone who wish to revive 68k need to do the same to have any hopes of success.

Last edited by kolla on 19-Sep-2022 at 07:39 PM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@amigang

There already is Emu68 for just stand-alone Raspberry Pi, it just lacks an OS currently. Once Michal has MMU working, you bet all rpi hardware will be made available for linux/68k etc. AROS/68k will eventually support pi too. Nothing will stand in the way for OS3.4 or whatever to do the same, but would be up to the OS owners/developers.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@agami

Quote:

Maybe people in Norway with their free (or near enough) fjord-powered electricity have the luxury of not caring, but most people should care about computing efficiency.

The power does not come from the fjords it comes from mountains, it comes from melting snow and rain, it’s a limited resource.

That was until government agreed to some agreement behind Norwiagines peaples back. Now they sell to the highest bidder, so now all prices is up, and they run out water in the pools, energy companies and government is getting rich and everyone else is getting poor. The industry has no longer any advantage, talk about shooting yourself in the foot.

This winter we will be using wood to keep houses warm, not wherry clamant friendly.
but who cares as long as someone is making money.

Tidal power you’re talking about is not wherry buildout, there is only one powerplant, its in northern part of Norway, that part of PowerGrid is not connected south part of the PowerGrid.

Why Norway was interested in the interchange cables was because import electricity when the water pool is low, that’s has not happened this year. some energy mix in Europa comes from gass powerplants, guess where the gass is coming from? So there is lack of electricity.

Last edited by NutsAboutAmiga on 19-Sep-2022 at 08:13 PM.
Last edited by NutsAboutAmiga on 19-Sep-2022 at 08:02 PM.
Last edited by NutsAboutAmiga on 19-Sep-2022 at 08:02 PM.
Last edited by NutsAboutAmiga on 19-Sep-2022 at 07:57 PM.
Last edited by NutsAboutAmiga on 19-Sep-2022 at 07:56 PM.
Last edited by NutsAboutAmiga on 19-Sep-2022 at 07:56 PM.
Last edited by NutsAboutAmiga on 19-Sep-2022 at 07:50 PM.

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

@matthey

Quote:

Early PPC Amiga hardware still had a 68k CPU which retained compatibility and eased porting as parts of the AmigaOS could be run on the 68k.

Nope, that’s not how it worked/works. Only when running the OS entirely on the 68k and only using powerpc as a co-processor, which was running its own OS kernel (either PowerUP or WarpUP) could 68k and PowerPC coexist. Once MorphOS and OS4 booted on the PowerPC, first thing that happens at boostrapping is that 68k hands over all control to PowerPC and then turns itself off. All 68k code under MorphOS and OS4 when running on CSPPC and BlizzardPPC is done by software emulation, trance and petunia respectively, and NOTHING runs on the 68k, which is very much “offline”.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

And for f*ck sake - being some sort of CPU/hardware Amiga guru and not knowing this shows some amazing level of ignorance, it’s been two decades already, plenty of time to learn the basics - sigh!

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC