minator Quote:

Other nice to have, I don't see the logic of doing an ASIC either.

Get a microscope.

minator Quote:

But without the equivalent 30 years of architectural improvement. So, it'll run slow and hot.

No. The 68060 is designed for lower power using many of the same low power features used today.

M68060 User’s Manual
https://www.nxp.com/docs/en/data-sheet/MC68060UM.pdf Quote:

In addition to substantial cost and performance benefits, the MC68060 also offers advantages in power consumption and power management. The MC68060 automatically minimizes power dissipation by using a fully-static design, dynamic power management, and low-voltage operation. It automatically powers-down internal functional blocks that are not needed on a clock-by-clock basis. Explicitly the MC68060 power consumption can be controlled from the operating system. Although the MC68060 operates at a lower operating voltage, it directly interfaces to both 3-V and 5-V peripherals and logic.

The Superscalar Architecture of the MC68060 Quote:

Power management The processor has many design features that reduce power during normal operation. Computing engines on the device share common input flip-flops that supply operands to the ALU structures. Since the design is static, each unit dissipates power any time the input flip-flops are updated. The chip includes special circuitry between the common operand registers and the computing engines to minimize unnecessary signal switching if the engine is not in use on a given cycle. The device carefully controls cache power dissipation, disabling each cache when there is no access to it. Also, it implements logic in both the instruction and data caches to determine in advance of an access which portion of the cache will be used and enabling only the necessary portion. All caches use an electrically stable, six-transistor static RAM cell. The caches use self-timing circuit techniques to achieve high performance with no external clock requirements. These circuit techniques eliminate the need for complex clocking to control cache operations and allow the caches to operate at any frequency. Ours is a fully static design. Its clock system has no internal phase-locked loop. As a result, the device can operate from its maximum frequency to 0 Hz, and switch operational frequency rapidly. This allows low-speed operation to conserve power. The processor implements an ultra-low-power sleep mode. A new instruction, the LPSTOP (low-power stop) instruction, puts the part in a low-power standby mode. In this mode, we can isolate the part from the external bus by tristating the outputs and isolating the inputs from internal circuitry. The external clock can also be stopped for maximum power saving. Since the processor is a fully static design with a clock system that does not use an internal phase-locked loop, it is possible to reapply the clock and resume full-speed operation from LPSTOP mode, triggered by either an interrupt or reset.

To summarize and add some features not mentioned, I created the following list.

1. low power at the time 3.3V operation
2. full static CMOS design allows clock speed to be reduced from max all the way to zero.
3. LPSTOP low power stop instruction
4. extensive power gating of internal units/blocks (turns off unused units/blocks)
5. reduced instruction fetch power with a 4B instruction fetch, buffer and decoupled IFP and OEPs
6. very good code density reduces instruction cache misses and expensive DRAM accesses
7. branch cache and prediction (~90% correct) reduce mispredicted instruction fetch and execution

Modern cores have additional low power features but most of them come from chip fab process improvements.

1. lower voltage (mostly made possible by process improvements)
2. more low power states (savings when transitioning to sleep state but minor)
3. better branch prediction (process improvements allow larger branch caches/histories)
4. 8T SRAM can be used instead of 6T SRAM (process improvements allow more transistors/area)

There is not much the 68060 is missing for low power and modernization could easily add these features. The 68060 has an 8-stage in-order core design which would be lower power than most deeper pipelined OoO RISC cores.

minator Quote:

Without out of order execution, modern prefetch or branch prediction it'll be constantly stalling. Without the power saving tech, you wont be able to save power while the processor is stalled, and it'll stay hot. This on top of CISC processors being inherently inefficient because of the big decoders.

The stalls of in-order cores are mostly from instructions and data not being in caches but most 68k programs are small. A modernized 68060 would likely have 4 times larger L1 caches (32kiB I+D) and at least a L2 cache because the 68060 8kiB I+D caches were such a bottleneck for the Amiga. The 2002 ColdFire V5 has 32kiB I+D caches, 8B/cycle instruction fetch, improved branch prediction and and a hardware return stack for a very similar design to the 68060 but more modern. OoO cores have more active transistors all the time to avoid stalling compared to in-order cores which is why they use more power and dissipate more heat. RISC designs often gain more from OoO execution because they have more stalls.

1. data cache miss and stall (even for CISC and RISC)
2. instruction cache miss and stall (good code density reduces cache misses, advantage CISC)
3. load-to-use stall (most CISC no and most RISC yes, advantage CISC)
4. branch misprediction stall (even for CISC and RISC)
5. store-to-load forwarding stall (even for CISC and RISC)

The above stalls are the often multi-cycle stalls that waste the most stall cycles. It looks to me like CISC has the advantage for stalls. The classic fat RISC architectures like Alpha, PA-RISC, MIPS, SPARC, PPC and original ARM being dead while AARch64 and RISC-V replacements have improved code density suggests that code density is important, which the 68060 has. SiFive figured out that load-to-use stalls are detrimental to performance and their in-order series 7 RISC CPU cores use a 68060 like design to eliminate load-to-use stalls. The 68060 design and features look like they were created with a crystal ball yet the design is left to bit rot unused because ARM is easy

minator Quote:

I think there'd be some market, but it'll be pretty small. It might be more interesting with a more modern design, but that'd be so expensive there really would be no market for it.

Do you believe THEA500 Mini sold over 100,000 units with a nonfunctional toy case, no Amiga branding, no AmigaOS and too high of price? Do you believe the Sega Genesis Mini has sold over 1.5 million units not counting Sega Genesis Mini 2 sales? Do you believe the equivalent of over $24 million was raised for the X68000 and how big of a crowd do you think it would take if true? How small do you think the 68k market is?

https://gamerant.com/sega-genesis-mini-2-low-limited-supply/ Quote:

According to sales data, the 2019 original Sega Genesis Mini (or Sega Mega Drive Mini depending on the region) sold over 1.5 million units. Some consoles even had region exclusive games depending on where they were sold. Much like the original, the Sega Genesis Mini 2 has a number of retro titles, although this time will feature some classics that never made it onto the system as well as some Sega CD games. Despite this being a product that would be greatly anticipated by older Sega fans and prospective buyers, the mini console will be in less supply compared to its 2019 release.

It would be nice to see updated sales numbers for these devices and the Commodore 64x for comparison.

@thread

Quote:

I asked Retro Recipes if they bought Commodore, could they fix this and they said yes they think they can.

Source:
Legal dispute: Hyperion can be held liable for copyright infringements

#6

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@matthey

Quote:

matthey wrote: No. The 68060 is designed for lower power ....

The "Let's Buy Commodore" project is not about reviving the 68060, it is not even about building a new 6510 or a new C64, it is about someone interested in the brand name, so it might be usable for Commodore related products.

#6 Quote:

FAQ added

There are a couple of interesting answers in the FAQ.

Quote:

Q: If Commodore International is gone, how can new products be considered truly “Commodore�? A: What if we owned the 47 trademarks going back to 1983, when Jack Tramiel was CEO? That’s the plan. But beyond that, what if we worked with original Commodore management, engineers, reps, tech support, and product devs? What if we collaborated with the SID chip’s creator on new SID chips, or with the Music Maker creators on a new version? (Just examples.) What if we helped reassemble the ROMs into Commodore? What if we even merged with Amiga? These are all goals. At some point along that path though, it stops being a reboot and starts being Commodore again. That’s the vision I began outlining in part 1. I want Commodore back.

"What if we even merged with Amiga?" Which Amiga? Amiga Corporation, Amiga Inc, Amiga Technologies? Oh, probably the ones with the Commodore ROMs. I would want to see the Commodore 64x sales numbers first. I expect THEA500 Mini sales numbers are higher and a merger between Amiga Corporation and RGL sounds more appealing to me so far. Peri has charisma for marketing but he also comes off as shady and weird.

Quote:

Q: How can you make money from Commodore without over-licensing or diluting the brand? A: By focusing on quality. Which ties back to the earlier question about deciding who can use the logo. If we only license carefully curated products that honour the brand’s legacy. Instead of mass licensing, we build flagship products in-house and selectively partner with creators who get it right. And there are a lot of them. Fewer, better releases mean stronger long-term value. Of course, this all only happens if we complete an acquisition at all.

Is the Commodore 64x one of the "carefully curated products that honour the brand’s legacy"? Will the Commodore A500x with a different facade but similar guts be also? Is a Commodore case with x86-64 hardware and Linux OS a deserving Commodore inspired successor or assimilation to make a quick buck with Commodore branding and cases?

#6 Quote:

Quote: I asked Retro Recipes if they bought Commodore, could they fix this and they said yes they think they can. Source: Legal dispute: Hyperion can be held liable for copyright infringements

With the mighty Commodore reassembled, mere mortals would be forced to capitulate in awe.

OneTimer1 Quote:

The "Let's Buy Commodore" project is not about reviving the 68060, it is not even about building a new 6510 or a new C64, it is about someone interested in the brand name, so it might be usable for Commodore related products.

Yes, I know Peri and My Retro Computer Ltd would like to do practically the opposite of what I propose. The Commodore 64x overshadows the MEGA65 which is almost a recreation of a C65 as Commodore was planning for the future of the C64. The Commodore Amiga plans for the Amiga were AAA and later a more practical AA+ chipset, which has documentation and may partially exist in logic. AGA was rushed, left much to be desired and still used NMOS chips. Many Amiga fans have never experienced AGA and it should have been the much better AA+. Commodore planned a full CMOS 68k AA+ SoC ASIC that internal Commodore docs suggested would reduce the Amiga price by $100 USD, likely saving the business if they had produced it. The 68060 and Amiga chipsets exist, if not lost, and could be used to build the 68k SoC ASIC that Commodore was working very slowly toward. Is Commodore OS Vision Linux and x86-64 hardware with Commodore branding and a Commodore quality seal of approval Commodore inspired enough for the Amiga instead? Was PPC AmigaNOne Amiga inspired enough without the 68k and Amiga chipset? Was AmigaNOwhere Amiga inspired enough without the 68k, Amiga chipset and AmigaOS? How far does a brand alone with a lot of marketing propaganda get a product?

Last edited by matthey on 16-Jun-2025 at 12:28 AM.

@matthey

Quote:

I thought the original MiniMig used a 68000 CPU.

ASIC 68SEC000 and chipset in FPGA yes.

Quote:

About 30 years of silicon improvements should do it.

It's not just a matter of material, there's more to it.

Quote:

Everything on one SoC ASIC die.

And exactly what is this "everything"? What is this System that you wish to put on a Chip"?

Last edited by kolla on 16-Jun-2025 at 03:57 AM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@OneTimer1

Quote:

OneTimer1 wrote: @cdimauro Quote: cdimauro wrote: A new, modernized C64 should be built with 100% compatibility ... I wouldn't spend money on a 100% compatible C64, I have plenty of them lying around. I can even get compatible C64 replacement boards, with or without improvements for an acceptable price. But I might spend some Euros on a small C64 compatible mobile device, with TFT, micro keyboard and some Euros more if it has USB 2.0 for Joystick, Keyboard and USB-Stick. You might realize how different expectations can be Quote: cdimauro wrote: I want to see the existing turbo tapes/disks running on it: plenty of the software was releases using them, which can't run on the C65! If you want to have a compatible tape device, you must use an 6502 running exactly the same clock frequency like the C64, timing is critical and was done using the CPU only, you will even run into troubles reading BASIC programs from a PET on a C64 or a VIC20

This was exactly what I was talking about, and shows why your "different" expectations that you expressed at the beginning of your post, in reality match (because what's primarily important with the C64 is... running its software --> the CPU speed is crucial).

@matthey

Quote:

matthey wrote: cdimauro Quote: @matthey: even the Commodore 64 deserves much better. The C65 was created by engineers that had no idea of the software which was already running on the C64, and created a quite incompatible product. In fact, I want to see the existing turbo tapes/disks running on it: plenty of the software was releases using them, which can't run on the C65! If the MEGA65 works like the C65, then the situation will be the same. The MEGA65 does work like the Commodore C65 when using the C65 core. The C65 has a C64 compatibility mode but it is less compatible than the C128 C64 compatibility mode. Primitive hardware that has minimal abstraction and relies on exact timing does not allow much upgrading with full compatibility. The MEGA65 uses a FPGA which can use a C64 core though. C64 for MEGA65 Version 5 Release Trailer https://www.youtube.com/watch?v=n3ke0alwjds Sy Twothousandandtwo 1.4K views 1 year ago 16 Comments Quote: Experience the Commodore 64 with great accuracy and sublime compatibility on your MEGA65! This core is based on the MiSTer Commodore 64 core which itself is based on the work of many others. MJoergen and sy2002 ported the core to the MEGA65 in 2022 and 2023. With our Release 5, we are striving for a retro C64 PAL experience: The core turns your MEGA65 into a Commodore 64 with a C1541 drive (you can mount *.d64) images. It supports the following hardware ports of the MEGA65: Joystick port for joysticks, mice and paddles Expansion port for C64 cartridges: Games, Freezers, Fast loader cartridges, REUs, Multi-Function Flash Cartridges, etc. IEC port so that you can attach real 1541 & 1581 drives as well as printers, plotters or modern devices such as the SD2IEC and the Ultimate-II+ Additionally, the C64 for MEGA65 core can simulate a 1750 REU with 512KB of RAM, it can simulate cartridges (by loading *.crt files) and it offers a Dual SID / Stereo SID experience. The C64 runs the original Commodore KERNAL and the C1541 runs the original Commodore DOS, which leads to authentic loading speeds. You will be surprised, how slowly the C64/C1541 were loading... You can optionally use JiffyDOS or use fast loader cartridges to speed up loading. And you will be amazed by the 99.9% compatibility that this core has when it comes to games, demos and other demanding C64 software. Some demos are even recognizing this core as genuine C64 hardware. And even things like using a fast loader cartridge while connecting a genuine 1541 via IEC are working flawlessly. DOWNLOAD THIS CORE FOR YOUR MEGA65 HERE: https://tinyurl.com/4cdfb4za THE CORE IS 100% OPEN SOURCE - VISIT US HERE: https://github.com/MJoergen/C64MEGA65 Unlike closed hardware like the Vamp/Apollo Core hardware, MiSTer FPGA cores can be flashed into the MEGA65. The CPU clock speed is adjustable between "1, 2, 3.5 and 40.5MHz" and the FPGA "Allows bigger third-party cores". https://mega65.org/ A MiSTer is capable of much of the same but the I/O and ports are customized for the C65 which also works well enough for the C64.

But the problem here is that either you flash the MEGA65 or the C64 core.

You have no chance of using them very quickly, like I did on my C128 (power-on -> native/C128. GO64 or reset + keypress -> C64).

It would have been much more interesting to disable & enable some new features on the fly (clock and instruction speed above all), so as to improve (even considerably) the existing software when it makes the most sense to do so.
Quote:

The Amiga 1200 MiSTer+MiSTress combo is somewhat similar where the MiSTress board contains Amiga specific I/O. The MEGA65 is about as far as faithful C64/C65 hardware can go and a FPGA allows to retain better compatibility than the C65 would have. The price is high and the market very niche as the video views and comments show.

Indeed. There was another project which kolla shared some time ago, which was cheaper.
Quote:

The 68k Amiga can advance much further with a 68k SoC ASIC while FPGA technology for chipsets allows to expand the market beyond the Amiga to a much larger retro 68k market.

Yup.
Quote:

cdimauro Quote: A new, modernized C64 should be built with 100% compatibility at its foundations. On top of this a lot of things can be added. So, similar to the C128, but without the alien Z80 and the super-crappy VDC. Replacing the 6510 with the Z80 was an upgrade though. The Z80 was handicapped in performance with only a 4-bit ALU but the ISA is nicer and the code density much better. The C128 was one of the few computers to retain a 6502 family CPU and it did not survive into the Amiga although was deeply embedded into the keyboard and some peripherals. The Z80 survived into some 68k hardware like the Radio Shack TRS-80 Model 16 and Sega Genesis for compatibility and as an I/O processor.

But it has nothing to do with the Commodore ecosystem.
Quote:

The 68000 itself survived for compatibility and I/O into the Atari Jaguar and Sega Saturn. The 68k was used in the Amiga CD32, Atari Jaguar, Neo Geo, Sega Genesis/Mega Drive and Sega Saturn consoles while the Commodore Amiga, Sharp X68000 and Atari ST were gaming computers. The 6502 is 50 years old but the 68000 is only 4 years younger at 46 years old. https://sonicstate.com/news/2024/09/19/the-68000-cpu-turns-45/ Quote: I love this architecture - there are 7 such chips in my (sound) studio, so here's to the next 45 years, 68k! The 6502 changed the world by being a very affordable MPU while the 68k changed the world by being a very good MPU. Then Motorola through out their beautiful 68k baby for an adopted ugly PPC baby and never looked back. The 68k should be brought back because not only was it good 46 years ago but it is still good today.

*

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: cdimauro wrote: @matthey: even the Commodore 64 deserves much better. The C65 was created by engineers that had no idea of the software which was already running on the C64, and created a quite incompatible product. In fact, I want to see the existing turbo tapes/disks running on it: plenty of the software was releases using them, which can't run on the C65! If the MEGA65 works like the C65, then the situation will be the same. A new, modernized C64 should be built with 100% compatibility at its foundations. On top of this a lot of things can be added. So, similar to the C128, but without the alien Z80 and the super-crappy VDC. C65 was a stealth R&D rebellion against Henri Rubin's bet the farm on ECS with "read my lips, no new chips".

Then I assume that all those engineers were immediately fired once the C65 was shown to the management, right?

"no new chips" -> they have wasted (a lot of) money of the company going AGAINST such directive.

IF this directive exists, of course, which I STRONGLY doubt. And that's not only because of the C65 existence, but because of the many chips that Commodore produced AND the general "working model" (!) of Commodore engineers (read Eggebrecht's declaration on the same book AND his interview AND the declaration of Hedley Davis on the same book).
Quote:

C65 project allowed the CSG LSI group to learn about Amiga's bitplane graphics architecture with 8 bitplane improvement. This is important after Commodore management kicked out Amiga's original graphics engineers. In 1989, CSG's C65 surprise on corporate politics forced the system integration group to quickly start AGA (Pandora) R&D with a go-ahead permission from Henri Rubin. With a go-ahead permission from Henri Rubin, the same CSG LSI group designed the AA Lisa chip with 8 bitplanes and ported some improvements from AAA, e.g. 16.7 million color palette. A single engineer was allocated to debug C65, ECS Agnus B, and AAA Andrea, which created an R&D bottleneck. Commodore HR and Herni Rubin would rather hire PC system integration and AMIX software engineers (from Coherent) instead of ASIC engineers. From Commodore - The Final Years CeBIT 1988 As usual, Commodore was the highlight of CeBIT when the doors opened on March 16, 1988. CEO Irving Gould, along with US president Max Toy and COO Henri Rubin in tow, gave a press conference in front of 500 attendees and journalists. Gould announced the positive financial results from the previous year. The bright spot was that 40% of the revenues were generated by Amiga products. The Amiga 500 and Amiga 2000 were truly taking over revenues from the C64. Gould was also remarkably open, or perhaps undisciplined, and revealed all the upcoming products, including the redesigned PC line, the A2286 bridgeboard, the A2500AT, the A2500UX, the A2620 accelerator, the A2300 Genlock for the Amiga 2000, the Transputer, and even the Amiga 3000 (much to the chagrin of Commodore’s engineers). Commodore Germany’s emphasis on PC-compatible products became understandable when Gould revealed Commodore was the leader in the West German microcomputer industry, with a market share of over 50%. In the business marketplace, Commodore ranked second, only behind IBM itself. Commodore Germany’s leadership, including Harald Speyer and Winfried Hoffmann, were ecstatic about the new PC40-III with an ATI graphics card. In fact, a demo running on the PC40 put the noninterlaced scan converter running on the Amiga 500 to shame. (Rubin made Porter turn off the A500 demo in response.) The new PC would also include a Commodore-made 1352 mouse. Although there was much to be excited for at the show, overall the German Commodore employees were upset that there were no new Amiga computers. In 1987, Porter had intended to have early samples of the new chipset ready by now, and incorporated into an Amiga 3000, but CSG had just missed the deadline for the chips by a few weeks. Make ATI Great by Commodore. Commodore gave ATI's 1st major OEM contract.

Guess what: the LSI group was able to produce a much better chipset, whereas the Amiga engineers group were shocked by it...

@minator

Quote:

minator wrote: @OneTimer1 Other nice to have, I don't see the logic of doing an ASIC either. But without the equivalent 30 years of architectural improvement. So, it'll run slow and hot. Without out of order execution, modern prefetch or branch prediction it'll be constantly stalling. Without the power saving tech, you wont be able to save power while the processor is stalled, and it'll stay hot.

OoOE isn't really important. There are plenty of in-order cores (ARM, RISC-V) which are still sold like pancakes, because they do their job very well and consuming just a fraction of power-hungry OoO cores (which are also fatter -> more expensive).

But yes, other things are important to have.
Quote:

This on top of CISC processors being inherently inefficient because of the big decoders.

That's the main concern that I've when Matt shared that the 68060 used a very big 64k x 16-bit LUT for decoding the 68k instructions: I don't know how much impact it could have on the operating frequencies.

@OneTimer1 and @thread

Quote:

The "Let's Buy Commodore" project is not about reviving the 68060, it is not even about building a new 6510 or a new C64, it is about someone interested in the brand name, so it might be usable for Commodore related products.

This question of "what is the point of this" keeps swirling and for good reason - it is the key question. But when you think about it, there are only 2 valid candidate answers:

1) to cater to us retro-computing enthusiasts, or
2) to do for the next generation what Commodore did for us (which I posted about here years ago)

(NOTE: You might say I'm missing an option - to build a new, modern computing platform. But I don't think anyone really considers that a realistic goal, and only leads back to people repeating the question "what is the point?".)

Just my 2 cents, but I think the point of resurrecting Commodore as a brand shouldn't be about particular tech (e.g. 68060) or faithful reproductions - it could (or should) be about #2 above - to give back a kid-friendly computing platform that can inspire a passion for computing in kids in the same way it did for us back in the 80's and 90's. When you think about it, that is what Commodore did for us 50-somethings, and that is why so many of us feel nostalgic for the brand today.

Why should we expect that Commodore must grow up, simply because we did? Commodore as a brand should remain about kid-friendly computing, where the point is to inspire a passion for computing in the next generation. Because that is the ONLY thing Commodore ever really achieved, the only thing it was really good at. And why shouldn't that be good enough? It's a narrowing of the focus, and focus is good. There are already plenty of brands out there fighting for the adult and business computing market. But who is really focused only delivering a computing platform aimed squarely at children?

_________________
Vic-20, C-64, C-128
Amiga 1000, 3000
AmigaOne X1000

@jingof

Quote:

jingof wrote: there are only 2 valid candidate answers: 1) to cater to us retro-computing enthusiasts, or 2) to do for the next generation what Commodore did for us (which I posted about here years ago) (NOTE: You might say I'm missing an option - to build a new, modern computing platform. But I don't think anyone really considers that a realistic goal, and only leads back to people repeating the question "what is the point?".) ... But who is really focused only delivering a computing platform aimed squarely at children?

1. Retro computing enthusiasts are a clear target, according to the videos presented by Peri. he has presented a wide variety of potential products reaching from simple C64 recreations over FPGA based systems up to C64x that are x86 based PCs with Linux.

2. Really next generations that where not listed under 1 could also be a target, but I don't thing the RaspberryPI needs a C= sticker, they where targeted as educational computers and could be used to learn Python (that's what the PI stands for). Python is somehow the BASIC of today with options for building high level applications.

What is missing on Python (in my opinion) are books explaining simple programming but today most of it is done via online tutorials (in text or as video on YT).

But educational platforms doesn't seem to be the target for Peri.'s initiative, they run with or without a C= sticker

Last edited by OneTimer1 on 16-Jun-2025 at 08:40 AM.

@OneTimer1

Quote:

I don't thing the RaspberryPI needs a C= sticker, they where targeted as educational computers and could be used to learn Python

I haven't spent very much time looking into RaspberryPI's use as an educational platform, so I could be wrong here. But it seems to me that parents don't see RaspberryPI as a turn-key solution for an educational platform. More a project for them first, to make it available and approachable by their children. An internet search turns up a lot of examples of parents asking, "how would I setup a RaspberryPI as a good kid-friendly computing platform". IOW, it has potential to be a turn-key product aimed at children, but it is not that today.

And even if it is true, that RaspberryPI already meets this need, no one ever said to only pursue opportunities with zero existing players. In fact, that's usually a bad indicator.

Quote:

doesn't seem to be the target for Peri.'s initiative

I agree with you there. That's why I say, why should we expect Commodore would be something it never was before. It was a kids platform back then, to learn how to code, to build games on etc. Why should Commodore become an adult-oriented brand/platform, just because we grew up? Keep to its original purpose.

Last edited by jingof on 16-Jun-2025 at 09:41 AM.
Last edited by jingof on 16-Jun-2025 at 09:12 AM.
Last edited by jingof on 16-Jun-2025 at 09:00 AM.

_________________
Vic-20, C-64, C-128
Amiga 1000, 3000
AmigaOne X1000

kolla Quote:

ASIC 68SEC000 and chipset in FPGA yes.

But not a SoC.

kolla Quote:

It's not just a matter of material, there's more to it.

Most of the gains from the 1970-1990s Amiga chip silicon would come from using a more modern fab process. Yes, there is major work required to create a 68k SoC ASIC. However, much of the logic exists and could be licensed.

Motorola 68k CPU cores
Amiga Corporation Amiga chipsets
SiFive SoC I/O

At least the later 68k cores and SiFive SoC I/O units are designed to be modular like ARM a la carte IP. Combining the tested cores and units should be much quicker and easier than starting from scratch.

kolla Quote:

And exactly what is this "everything"? What is this System that you wish to put on a Chip"?

68k SoC ASIC
multiple 68060 cores
a 68000 core
3D GPU core
high level cache and memory controller
modern I/O support
legacy I/O support (most on headers or through GPIO)

Vortex86 SoCs with modern I/O and legacy I/O support for x86 hardware are similar to what I want for a 68k SoC.

https://en.wikipedia.org/wiki/Vortex86

The Vamp/AC SoC FPGA is also similar but very limited by the affordable FPGA and it has other issues. Moore's law is slowing down so it is not necessary to move to a new fab process as often and it is more important to have good designs and improve existing designs.

cdimauro Quote:

But the problem here is that either you flash the MEGA65 or the C64 core. You have no chance of using them very quickly, like I did on my C128 (power-on -> native/C128. GO64 or reset + keypress -> C64).

It is fast to flash a core, at least compared to 1541 drive times that most C64 users are used to and the timing of which is required for maximum compatibility. The C65 has a GO64 mode too for basic C64 compatibility but full compatibility is not possible without including duplicate C64 hardware. Duplicate hardware including multiple CPUs and video hardware meant the C128 was not cheap. There are rumors that the Amiga 500 ended up cheaper to produce than the C128. The 68k Amiga SoC that Commodore planned would have dropped the Amiga price by $100 USD according to internal Commodore docs so the Amiga 1200 and CD32 may have ended up cheaper to produce than the C128 too. Even the FPGA based MiSTer+MiSTress A1200 is cheaper than the MEGA65 and can use a C64 or C65 FPGA core which is likely adequate for most Commodore fans.

cdimauro Quote:

It would have been much more interesting to disable & enable some new features on the fly (clock and instruction speed above all), so as to improve (even considerably) the existing software when it makes the most sense to do so. [quote] FPGAs usually has multiple clocks/PLLs but some are used for other purposes. Many clocks are desirable for retro compatibility. Even my 68060 CSMK3 accelerator has 3 oscillator locations and jumpers for clock routing. The Amiga AAA chipset would have also required many clocks. cdimauro [quote] OoOE isn't really important. There are plenty of in-order cores (ARM, RISC-V) which are still sold like pancakes, because they do their job very well and consuming just a fraction of power-hungry OoO cores (which are also fatter -> more expensive). But yes, other things are important to have.

When the instructions and data are cached in L1 (no L1 cache miss stalls) and when there are no load-to-use stalls (68060 design), OoO cores are indeed expensive and wasteful. Most RISC cores have fat code increasing the L1 cache misses and load-to-use stalls which is why they benefit from OoOE removing the excessive stalling. The 68060 has small code with fewer L1 instruction cache misses and avoids load-to-use stalls.

cdimauro Quote:

That's the main concern that I've when Matt shared that the 68060 used a very big 64k x 16-bit LUT for decoding the 68k instructions: I don't know how much impact it could have on the operating frequencies.

LUTs are expensive in FPGA because the data is programmable. There is a big difference between what I expect is SRAM for LUTs in a FPGA and read only ROM data in an ASIC. ROM bits in an array can use less than 1 transistor per bit on average. The most common SRAM for CMOS uses 6 transistors/bit (6T SRAM) while 8 transistors/bit (8T SRAM) can reduce power.

https://en.wikichip.org/wiki/static_random-access_memory

It would be possible to have CPU cores with programmable decoders similar to FPGA programmable logic. This could allow to change ISAs as well as fix bugs which would be convenient but it would likely use more active transistors increasing power. The increased active transistors of CISC decoders are partially offset by the reduced instruction cache and DRAM active transistors. The reduced instruction fetch from a good code density also reduces power and energy used.

https://www.cast-inc.com/blog/consider-code-density-when-choosing-embedded-processors Quote:

Code Density’s Effect On Power Consumption A denser instruction set yields smaller code, which in turn requires fewer instruction fetches. This has a major impact on efficiency, because the energy used to fetch an instruction from on-chip memory is much greater than that required for the CPU to execute that instruction. How much greater? The 2010 Stanford dissertation study by James Balfour on Efficient Embedded Computing showed that for a typical 32-bit RISC processor, an instruction fetch from a 2 Kbyte on-chip cache consumes 6.8 times more energy than a 32-bit add operation (in a typical 45nm CMOS process). Another study by Dally et al published in IEEE Computer found that 42% of a typical embedded processor’s energy consumption comes from instruction fetching, and only 6% is consumed by performing the actual arithmetic operations. Both these studies used a small, on-chip cache. When instructions more realistically need to be accessed from larger, on-chip or even worse off-chip memory, the impact is magnified, and the factor of 6.8x can go to 50x or more. The conclusion here is clear: using an ISA that requires fewer instructions is a much bigger energy saver than simply selecting a processor that executes more efficiently.

The 68060 decoders are much different than for x86 yet it does not appear to be an issue in a comparison with the P54C Pentium.

Metric | 68060 | PentiumP54C
Power (max) 5.5W@75MHz 9.5W@75MHz (+73% for Pentium)
Performance (ByteMark-int) 1.20 0.82 (+46% for 68060)
Area (transistors) 2.5million 3.3million (+32% for Pentium)

The 68060 dominated the Pentium in PPA.

Last edited by matthey on 16-Jun-2025 at 07:29 PM.
Last edited by matthey on 16-Jun-2025 at 07:26 PM.

@Thread

Commodore both brought to market, cost reduced and made successful and mismarketed and killed the Amiga as well as themselves! I think we are doing alright without this cursed brand! Let Amiga Corp have control of the IP that they already own. This is done by dealing with the Amiga IP squatters and not wasting time on a dead man's life's work that he lost control of in his own lifetime! It was Gould's pension by the end! Drained of any life force like a Gelfling in the Dark Crystal! RIP C=!

Last edited by BigD on 16-Jun-2025 at 07:44 PM.

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

jingof Quote:

I haven't spent very much time looking into RaspberryPI's use as an educational platform, so I could be wrong here. But it seems to me that parents don't see RaspberryPI as a turn-key solution for an educational platform. More a project for them first, to make it available and approachable by their children. An internet search turns up a lot of examples of parents asking, "how would I setup a RaspberryPI as a good kid-friendly computing platform". IOW, it has potential to be a turn-key product aimed at children, but it is not that today. And even if it is true, that RaspberryPI already meets this need, no one ever said to only pursue opportunities with zero existing players. In fact, that's usually a bad indicator.

Original and low end RPi SBCs could have been more complete and easier to setup and use but the newer RPi 400/500 are more complete.

jingof Quote:

I agree with you there. That's why I say, why should we expect Commodore would be something it never was before. It was a kids platform back then, to learn how to code, to build games on etc. Why should Commodore become an adult-oriented brand/platform, just because we grew up? Keep to its original purpose.

The RPi is kid friendly because of the price. It is the new poor mans platform for the masses like Commodore used to be. The Commodore 64x will never be a replacement for the original Commodore because of the hardware price. The AmigaNOne will never be a replacement for the original Commodore because of the price. How many people less than 20 years old do you think own an AmiagNOne computer? My guess is zero. RPi produces affordable hardware for the masses like the original Commodore while new Commodore and Amiga branded hardware is expensive hardware for the classes. Commodore dinosaurs may think current offerings are cheap enough for the masses as the hardware is cheaper than when they were a kid saving up for a computer by working all summer. Modern kids look at available hardware and they see the RPi, Chinese handheld gaming devices and emulators for smart phones that fit in their allowance for lunch in some cases. The RPi success came from aggressively reducing the price!

https://www.forbes.com/sites/michaelwolf/2013/02/10/one-million-raspberry-pis-later-the-story-of-how-a-mobile-phone-chip-helped-deliver-a-vision/ Quote:

The project originated in 2006, when a former University of Cambridge professor named Eben Upton and his colleagues noticed a lack of real computer knowledge among incoming computer science students. Figuring the iPhone generation didn't really have access to the type of tinkerer's machines of yore like the BBC Micro or Commodore 64, they set out to see if they could create one themselves, all for the price of a text book. And here's the amazing part of the story: They did just that. On Leap Day 2012, over five years after the conception of the original vision, the first Raspberry Pi's went on sale for the price of a text book (or $35 for the first models, and $25 for newer, lower-cost units). A Very Aggressive Price Point But to say the journey was an easy one would be to minimize over five years of effort, much of which was spent haggling over pennies as Upton and others tried to source the roughly 80 or so parts for the machine at the bargain-basement prices needed. According to Upton, whom I spoke to recently for the NextMarket podcast, there were times when he wondered if they'd been too aggressive in thinking they could make a computer for $35. "It's a very aggressive price point," said Upton. "We did try a bunch of routes to get there. We ended up with prototypes that were for one reason or another that were inadequate." Perhaps the biggest challenge of all was finding a computer processor that was both powerful and low-cost enough for their machine. After first trying some low-power microcontrollers, they found they were not getting the performance they wanted at the price they needed.

Value is very important too as they looked at performance/price. All current Commodore and Amiga hardware is nowhere close to being on the radar of kids. Even THEA500 Mini likely did not have a low enough price and/or good enough value for kids. Some parents likely bought Minis for their kids. Some parents likely bought the Commodore 64x for their kids. No parents likely bought AmigaNOnes for their kids because the price and value are ridiculous. The way to proliferate the Amiga by introducing it to the younger generations and kids is to aggressively reduce the price and increase the value. The way to do this is with a 68k SoC ASIC to leverage the 68k Amigas small footprint cost advantage but there may be too many Amiga dinosaurs accepting extinction. The retro games resonate with kids and present an opportunity if hardware can be aggressively priced reduce down to fit on their radar.

How important was a SoC for Eben Upton's goals?

https://www.forbes.com/sites/michaelwolf/2013/02/10/one-million-raspberry-pis-later-the-story-of-how-a-mobile-phone-chip-helped-deliver-a-vision/ Quote:

This Chip Came Along... Eventually Upton's patience and front row seat paid off. It was 2009, and a promising new system on chip with the decidedly chip-like name of BCM2835 came along. The BCM2835 was packed with functionality: it had a powerful ARM processor engine built inside, as well as a powerful video engine called VideoCore, which is the same video processing engine used in some Samsung mobile phones and iPods, all at a low enough cost to meet the requirements of cost-conscious mobile phone manufacturers. In short, everything Raspberry Pi needed. Upton knew he had found his chip. "I was on design team at Broadcom for the chip we used. You wake up one day and you realize you're not just designing a great chip, but you're designing exactly the chip you need for the project you were planning." One million Raspberry Pi's and possibly a new generation of tinkerers later, it appears Upton found just the chip he was looking for.

The original RPi SoC is not particularly powerful or high tech. It uses a scalar ARM11 CPU core where the 68060 is superscalar and has better performance efficiency (performance/MHz). The advantage of a SoC is price efficiency (performance/price) for the whole system though.

Last edited by matthey on 16-Jun-2025 at 09:20 PM.
Last edited by matthey on 16-Jun-2025 at 09:06 PM.
Last edited by matthey on 16-Jun-2025 at 08:22 PM.
Last edited by matthey on 16-Jun-2025 at 08:21 PM.

@BigD

Quote:

... I think we are doing alright without this cursed brand! Let Amiga Corp have control of the IP that they already own. ...

The whole "Let's Buy Commodore" project is without the Amiga, Amiga is its own brand and some companies have proven they can sell Amiga compatible retro systems, without any brand name of the past.

Last edited by OneTimer1 on 16-Jun-2025 at 10:16 PM.
Last edited by OneTimer1 on 16-Jun-2025 at 10:10 PM.
Last edited by OneTimer1 on 16-Jun-2025 at 10:09 PM.

@OneTimer1

There have been several mentions of Amiga in the comments. I think many comments exist to simply gauge interest, as well as potentially gaining new contacts. It all might help him determine how far he can go with this.

Waiting for Part 2.....

#6

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@matthey

Quote:

matthey wrote: kolla Quote: ASIC 68SEC000 and chipset in FPGA yes. But not a SoC.

No, luckily not.

Quote:

Motorola 68k CPU cores

Where would you get this?

Quote:

Amiga Corporation Amiga chipsets

Where would you get this?

Quote:

SiFive SoC I/O

Got details about this?
Which SiFive product or equivalent are you suggesting?

Quote:

68k SoC ASIC multiple 68060 cores

For SMP?
For management?

Quote:

a 68000 core

Huh, for what? For Legacy? Running separately, with its own RAM? Why just one?

Quote:

3D GPU core

What 3D core?

Quote:

high level cache and memory controller modern I/O support legacy I/O support (most on headers or through GPIO)

What modern I/O?
What legacy I/O?

Quote:

Vortex86 SoCs with modern I/O and legacy I/O support for x86 hardware are similar to what I want for a 68k SoC.

Hm, modern I/O? I see ISA, ATA, USB 2.0 and 100 Mb ethernet...


Anyways, good luck - better start designing, no time to lose here!

Last edited by kolla on 17-Jun-2025 at 11:44 AM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@jingof

Quote:

jingof wrote: @OneTimer1 and @thread Quote: The "Let's Buy Commodore" project is not about reviving the 68060, it is not even about building a new 6510 or a new C64, it is about someone interested in the brand name, so it might be usable for Commodore related products. This question of "what is the point of this" keeps swirling and for good reason - it is the key question. But when you think about it, there are only 2 valid candidate answers: 1) to cater to us retro-computing enthusiasts, or 2) to do for the next generation what Commodore did for us (which I posted about here years ago) (NOTE: You might say I'm missing an option - to build a new, modern computing platform. But I don't think anyone really considers that a realistic goal, and only leads back to people repeating the question "what is the point?".) Just my 2 cents, but I think the point of resurrecting Commodore as a brand shouldn't be about particular tech (e.g. 68060) or faithful reproductions - it could (or should) be about #2 above - to give back a kid-friendly computing platform that can inspire a passion for computing in kids in the same way it did for us back in the 80's and 90's. When you think about it, that is what Commodore did for us 50-somethings, and that is why so many of us feel nostalgic for the brand today.

You can cover both #1 and #2 with a single product.
Quote:

Why should we expect that Commodore must grow up, simply because we did? Commodore as a brand should remain about kid-friendly computing, where the point is to inspire a passion for computing in the next generation. Because that is the ONLY thing Commodore ever really achieved, the only thing it was really good at. And why shouldn't that be good enough? It's a narrowing of the focus, and focus is good. There are already plenty of brands out there fighting for the adult and business computing market. But who is really focused only delivering a computing platform aimed squarely at children?

That's really difficult to define nowadays.

I can tell you that I find a lot of fun by using Python, because it allows me to quickly tinker about the idea which popped on my mind.

But that's very very distant from what I did when I started programming, 43 years. Besides the BASIC experience, then it was... machine language, which was not that much fun. -_-