@cdimauro

Quote:

coder76 wrote: Yes, though the quality of texturemapping is better with SAGA, PS1 uses affine texturemapping, and fixed point arithmetic for vertex calculations, the texture mapping quality is indeed poor. Gunnar said he aims for something like PS2 quality, but I'm not sure how performance of SAGA compares to PS2. For me, PS2 quality/performance 3D in a 68k Amiga would be more than enough. Quote: cdimauro wrote: That's not possibile with the current, limited resources of the FPGA being used, and the clock frequencies that it can reach. PS2 specs: https://en.wikipedia.org/wiki/PlayStation_2_technical_specifications

Yes, perhaps with future iterations SAGA gets faster, this 3D demo with a single object already runs in less than 50/60 FPS according to frame time.

Quote:

cdimauro wrote: I've also programmed the AGA at the time, and it was good only because it was cheap and builtin on my A1200. You can see yourself, from what you've reported above, that it's a very bad, and the only useful thing is the bank switching that on other platforms is done and works much better. I've written another series of articles which better covers this topic. This is the last one: Missed opportunities to improve the Amiga chipset At the bottom you'll find the list of all previous articles. The #5 covers AGA (and all chips/chipset that was produced by Commodore for the Amiga) and you can read my technical analysis of why it was too little, too late, too bad.

Ok nice, this sort of information is rare to find on net, how to actually extend the Amiga chipset in the right way. It will take me a longer time to read everything though.

Quote:

coder76 wrote: Regarding the SAGA, you can look at SAGA docs how Gunnar added his new hardware registers into $dff000, one can choose to extend old registers, or just make entirely new registers, like he did for e.g. 16x16 bit audio channels. Quote: cdimauro wrote: I already know it, but Gunnar follows the same "principle" of the Commodore engineers that substituted the original ones: adding patches to the chipset to quickly get what they need. So, the design is very bad. The audio channels example that you've reported is showing exactly this: a bad patch over the existing chip (just duplicating the audio registers bank), instead of a proper designing. You can see on the above series how audio could be improved (article #1) in a better way, also taking advantage of the increased frequencies (#3) and increased memory bus width (#4), going specific details for 16, 32, and 64+ memory bus widths. In a short sentence: a much cleaner and future-proof evolution of the chipset.

Yes, you might be right, there is also limited space in the $dff000 space, but Gunnar just wanted something that works fast, and that's not so bad for overall design with the audio, if you want compatibility with old system, just use the old registers and otherwise use new registers. He also added a new DMACON2 register that controls new audio channels and ethernet, and a second INTREQ2 for interrupts. I'm assuming that there will not be a new version of the chipset by someone else, so he can do what he wants to. There is also some ATARI hardware regs he added recently.

One of the important points is that blitter can run in parallell with CPU, and that there is enough bandwidth for both to do so. And maybe enhance DMA fetches to 128/256 bit, I'm not sure if this implemented in SAGA. I also suggested an improvement to SAGA sprites as they still use only 16 colors each, with an attach bit to get 2 sprites together for 256 colors, and a separate 256 color bank for each maybe.

@coder76

Quote:

coder76 wrote: @cdimauro Quote: cdimauro wrote: That's not possibile with the current, limited resources of the FPGA being used, and the clock frequencies that it can reach. PS2 specs: https://en.wikipedia.org/wiki/PlayStation_2_technical_specifications Yes, perhaps with future iterations SAGA gets faster, this 3D demo with a single object already runs in less than 50/60 FPS according to frame time.

Because it's a simple demo. SAGA could improve, but its intrinsic limits can't be overcome.

A much better FPGA is needed, but the platform is already quite expensive and then it would be out of market.

The game changer would be the 68080 in ASIC, which could likely reach 1-2Ghz, but it requires a consistent investment which isn't worth only looking at the Amiga market (is should expand to the embedded market, to justify financing it).
Quote:

Quote: cdimauro wrote: I already know it, but Gunnar follows the same "principle" of the Commodore engineers that substituted the original ones: adding patches to the chipset to quickly get what they need. So, the design is very bad. The audio channels example that you've reported is showing exactly this: a bad patch over the existing chip (just duplicating the audio registers bank), instead of a proper designing. You can see on the above series how audio could be improved (article #1) in a better way, also taking advantage of the increased frequencies (#3) and increased memory bus width (#4), going specific details for 16, 32, and 64+ memory bus widths. In a short sentence: a much cleaner and future-proof evolution of the chipset.

Yes, you might be right, there is also limited space in the $dff000 space,[/quote]
4KB only for the chipset registers is a very good amount... if you know how to use it. On my series there's also a memory map proposed, using the full 4KB space, for both the old / 16-bit and the new / 32-bit registers.
Quote:

but Gunnar just wanted something that works fast, and that's not so bad for overall design with the audio,

Well, he just wanted to achieve the goal of adding more channels, but with the minimal effort on side. So, a quick solution which, unfortunately, is just a patchwork (like AGA).
Quote:

if you want compatibility with old system, just use the old registers and otherwise use new registers.

It's possible to have both.

Take a look at my articles, and you'll find the solutions for evolving (A LOT) the original platform, while keeping 100% backward-compatibility.
Quote:

He also added a new DMACON2 register that controls new audio channels and ethernet, and a second INTREQ2 for interrupts.

Yes, those are needed.
Quote:

I'm assuming that there will not be a new version of the chipset by someone else, so he can do what he wants to.

There are already other projects with a slightly better Amiga chipset, but Gunnar does always what he likes (which is OK: it's his project. Nothing to say here).
Quote:

There is also some ATARI hardware regs he added recently.

No, it's already since some years. It made sense, because the Atari ST is also strong, and they always have seek for solutions to improve the original platform.

The processor is the same, but the chipsets are different. However, the ST chipset is a joke compared to the Amiga one, so supporting the ST wasn't a big burden.
Quote:

One of the important points is that blitter can run in parallell with CPU, and that there is enough bandwidth for both to do so. And maybe enhance DMA fetches to 128/256 bit, I'm not sure if this implemented in SAGA. I also suggested an improvement to SAGA sprites as they still use only 16 colors each, with an attach bit to get 2 sprites together for 256 colors, and a separate 256 color bank for each maybe.

You can find all of that, and much more, on my articles.

@cdimauro

Quote:

Of course: who's interested on PowerPCs in our community? Amiga was, and is, only 68k. PowerPC machines aren't Amigas, and they'll never be.

They were never part of Commodore's vision, no. However, the Amiga was changing hands left right and centre towards the end and with a future devoid of any new 68K silicon I don't think PPC was a bad choice at the time. It was a migration path that Apple had shown was possible. And to be fair to the NG systems, that migration was actually a lot better than Apple managed originally.

I was an early PPC adopter, getting one of the first revision BlizzardPPC (I had to send it back and it came back pretty extensively reworked). The kernel wars were annoying but by the time 3.5 came out it was all pretty much settled.

They were interesting times and it was fun to work with. I don't have an axe to grind about that path except the fact that it came to an end - the same dead end as 68K silicon - years ago.

_________________
Doing stupid things for fun...

cdimauro Quote:

Because it's a simple demo. SAGA could improve, but its intrinsic limits can't be overcome. A much better FPGA is needed, but the platform is already quite expensive and then it would be out of market. The game changer would be the 68080 in ASIC, which could likely reach 1-2Ghz, but it requires a consistent investment which isn't worth only looking at the Amiga market (is should expand to the embedded market, to justify financing it).

The SAGA chipset could remain in FPGA as a 2D chipset, at least for compatibility. The CPU and GPU cores need more area and higher clock speeds than FPGA can provide. With an ASIC, the AC68080 CPU ISA is not particularly attractive for embedded use as it aims for desktop like performance with late 1990s features and misses the mark there too. Someone serious about a 68k ASIC would likely prefer licensing original 68k cores for max compatibility, easier retro and embedded marketing and more professionally developed HDL code. Licensing a professionally developed 3D GPU would also likely be easier than further developing the SAGA 3D for an ASIC. Gunnar is doing what he can for an affordable FPGA but I believe it is far removed from what needs to be done for a professionally developed ASIC. Targeting ASIC development would require more expensive FPGA hardware used for ASIC development despite reduced sales that may not break even. Gunnar may go for one of the cheap FPGA to ASIC conversions to try to keep up with higher performance CPU and GPU options of the competition but that is likely the limit anyone would invest in his precious due to poor design decisions like optimizing for a FPGA. ASIC CPU cores blow away affordable FPGA CPU cores and even ASIC CPU cores with a major emulation handicap outperform affordable FPGA CPU cores yet his precious is designed and optimized for FPGA. If someone on the Apollo Team had the foresight to tell him ahead of time and stand up to him, he would no longer be on the Apollo Team as Gunnar is the expert in all things.

Karlos Quote:

They were never part of Commodore's vision, no. However, the Amiga was changing hands left right and centre towards the end and with a future devoid of any new 68K silicon I don't think PPC was a bad choice at the time. It was a migration path that Apple had shown was possible. And to be fair to the NG systems, that migration was actually a lot better than Apple managed originally. I was an early PPC adopter, getting one of the first revision BlizzardPPC (I had to send it back and it came back pretty extensively reworked). The kernel wars were annoying but by the time 3.5 came out it was all pretty much settled. They were interesting times and it was fun to work with. I don't have an axe to grind about that path except the fact that it came to an end - the same dead end as 68K silicon - years ago.

The Apple migration from 68k to PPC was a "mass" migration with 150,000 pre-orders and 1 million PPC Macs sold in the first year.

https://en.wikipedia.org/wiki/Power_Macintosh#Release_and_reception_(1994–1995) Quote:

The Power Macintosh was formally introduced at the Lincoln Center for the Performing Arts in Manhattan on March 14. Pre-orders for the new Power Macintosh models were brisk, with an announced 150,000 machines already having been sold by that date. MacWorld's review of the 6100/60 noted that "Not only has Apple finally regained the performance lead it lost about eight years ago when PCs appeared using Intel's 80386 CPU, but it has pushed far ahead." The performance of 680x0 software is slower due to the emulation layer, but MacWorld's benchmarks showed noticeably faster CPU, disk, video, and floating-point performance than the Quadra 610 it replaced. By January 1995, Apple had sold 1 million Power Macintosh systems.

Apple users were more into software, the UI and eye candy while Amiga users were more tech savvy, understood and liked the hardware more and valued value and compatibility more. The PPC AmigaNOne primarily delivered eye candy with a minor upgrade to performance. PPC is further behind in performance now compared to the competition than the 68k was then and PPC performance/price was poor then and much worse today. How easy migration is is not as important as how large migration is. How large was the migration from 68k Amiga to PPC AmigaNOne in comparison to PPC Macs? How does this compare to 68k Amiga users buying RPi Amiga emulation hardware, V4SA hardware, THEA500 Mini and likely to buy THEA1200? What kind of 68k Amiga stand alone hardware is required to retain 68k Amiga users and can it be mass produced for mass migration?

@matthey

Quote:

matthey wrote: The SAGA chipset could remain in FPGA as a 2D chipset, at least for compatibility. The CPU and GPU cores need more area and higher clock speeds than FPGA can provide.

Why?

Quote:

the AC68080 CPU ISA is not particularly attractive for embedded use as it aims for desktop like performance with late 1990s features and misses the mark there too.

The AC68080 wasn't designed for embedded so it's hardly surprising that it isn't attractive for that. But why are we even talking about embedded?

I only have a V2 Vampire in my A600 but even that is much more powerful than the A3000 with 060 and RTG that I had in 1997. V4 also has 'Super' AGA at blistering speed, high speed IDE, 100Mbit Ethernet, 16 channel sound, 1GB RAM and 3D hardware texturing. It's every 90's Amiga fan's dream and more. Saying it 'misses the mark there too' is just plain wrong. It's not cheap, but period-accurate alternatives are even more expensive without offering nearly as much.

The only thing that makes V4 less attractive to me is that it's too powerful. If I develop software for it I will be limiting my audience to other V4 owners, and I will not be achieving my goal of proving what can be done on real 1990's era Amigas. My A600 lies idle most of the time for the same reason.

Quote:

Gunnar may go for one of the cheap FPGA to ASIC conversions to try to keep up with higher performance CPU and GPU options of the competition but that is likely the limit anyone would invest in his precious due to poor design decisions like optimizing for a FPGA.

Gunnar may try doing that if he wants. Getting enough investment to finance it is a marketing issue, not hardware.

You say he made 'poor design decisions like optimizing for a FPGA', but I bet it wouldn't take much to modify it for ASIC. The real question is where is the break-even point? To justify going ASIC you need to get the unit price down below FPGA in a quantity that is saleable. Once again that is mostly a marketing problem.

If I was Gunnar I would offer the bare chip (possibly on a carrier for ease of DIY installation) for people to put in their own Amiga designs. This would make the cost of entry much lower and encourage diversity, making his 'Amiga Chipset' a popular part that makers would go for. It might even attract OEMs who want to reach the mass market.

Quote:

If someone on the Apollo Team had the foresight to tell him ahead of time and stand up to him, he would no longer be on the Apollo Team as Gunnar is the expert in all things.

Too many cooks spoil the broth. After getting rid of Gunnar they would fight amongst themselves over what to do, or even worse agree on an unachievable goal like Commodore's engineers did with AAA - and we wouldn't get a finished product.

Quote:

Apple users were more into software, the UI and eye candy while Amiga users were more tech savvy, understood and liked the hardware more and valued value and compatibility more.

Mac users were willing to put up with under-performing hardware because they had been trained to do so from the start. Or were they? The truth is Apple was already losing market share in 1995 despite no longer having competition from Atari and Commodore. Sales dropped by 10% in 1996 and 30% in 1997. By 1998 they only had 2.7% market share. Mac sales didn't take off again until well into the 21st century when they switched to x86.

@Karlos

Quote:

Karlos wrote: @cdimauro Quote: Of course: who's interested on PowerPCs in our community? Amiga was, and is, only 68k. PowerPC machines aren't Amigas, and they'll never be. They were never part of Commodore's vision, no. However, the Amiga was changing hands left right and centre towards the end and with a future devoid of any new 68K silicon I don't think PPC was a bad choice at the time. It was a migration path that Apple had shown was possible. And to be fair to the NG systems, that migration was actually a lot better than Apple managed originally.

Well, "NG" (SICH!) systems arrived when Apple already transitioned to a completely new platform (MacOS X).

By the time (on 2000, to be more precise), PowerPC weren't competitive anymore and Apple was already moving to Intel's x86 (only IBM's stopped this process, with the promised G5).

Moving to PowerPCs after that time frame is already for sure a lack of understanding how the technology was evolving. Hence, a very bad decision (from Amiga Inc.).
Quote:

I was an early PPC adopter, getting one of the first revision BlizzardPPC (I had to send it back and it came back pretty extensively reworked). The kernel wars were annoying but by the time 3.5 came out it was all pretty much settled.

But that wasn't the "NG" time: those were accelerator cards developed by third-parties.

Amiga had other accelerators even well before that (with the notable ones using the famous Transputers).
Quote:

They were interesting times and it was fun to work with. I don't have an axe to grind about that path except the fact that it came to an end - the same dead end as 68K silicon - years ago.

Tinkering is (usually) always fun.

But, out of curiosity: did you play with PowerPC assembly?

@matthey

Quote:

matthey wrote: cdimauro Quote: Because it's a simple demo. SAGA could improve, but its intrinsic limits can't be overcome. A much better FPGA is needed, but the platform is already quite expensive and then it would be out of market. The game changer would be the 68080 in ASIC, which could likely reach 1-2Ghz, but it requires a consistent investment which isn't worth only looking at the Amiga market (is should expand to the embedded market, to justify financing it). The SAGA chipset could remain in FPGA as a 2D chipset, at least for compatibility.

That's exactly what I've proposed time ago: chipset -> FPGA, CPU -> ASIC.
Quote:

The CPU and GPU cores need more area and higher clock speeds than FPGA can provide. With an ASIC, the AC68080 CPU ISA is not particularly attractive for embedded use as it aims for desktop like performance with late 1990s features and misses the mark there too. Someone serious about a 68k ASIC would likely prefer licensing original 68k cores for max compatibility, easier retro and embedded marketing and more professionally developed HDL code. Licensing a professionally developed 3D GPU would also likely be easier than further developing the SAGA 3D for an ASIC. Gunnar is doing what he can for an affordable FPGA but I believe it is far removed from what needs to be done for a professionally developed ASIC. Targeting ASIC development would require more expensive FPGA hardware used for ASIC development despite reduced sales that may not break even. Gunnar may go for one of the cheap FPGA to ASIC conversions to try to keep up with higher performance CPU and GPU options of the competition but that is likely the limit anyone would invest in his precious due to poor design decisions like optimizing for a FPGA. ASIC CPU cores blow away affordable FPGA CPU cores and even ASIC CPU cores with a major emulation handicap outperform affordable FPGA CPU cores yet his precious is designed and optimized for FPGA. If someone on the Apollo Team had the foresight to tell him ahead of time and stand up to him, he would no longer be on the Apollo Team as Gunnar is the expert in all things.

Hum. I don't see a clear goal here, so exactly which markets should be supported.

If you need to support retrogaming then embedding a cycle-exact 68000 is mandatory, of course. An asymmetrical multicore is perfectly fine (a 68000 for such old system, or used as an I/O-companion core. And a 680x0 for high-performance). The very difficult part is getting a synthesised 68000.

A dual issue, in order core, is quite common for the embedded market, and a 680x0 (I prefer to don't put 68080 here, because Gunnar's extensions are better to be removed and replaced with more useful ones) can surely fit. This design can also be castrated to reach different embedded markers (e.g.: a single pipeline, for example. No FPU. No SIMD/Vector. FPU using data registers. SIMD/Vector using data registers. Crypto extensions. And so on).

The GPU isn't always needed on the embedded market. Many times not even a simple 2D chipset is needed.

A more modern GPU, with 3D and perhaps RT is required only on high-end markets.

So, the question is: which kind of markets do you think that can be addressable by a new 680x0 platform?
Quote:

Apple users were more into software, the UI and eye candy while Amiga users were more tech savvy, understood and liked the hardware more and valued value and compatibility more.

Platforms were also very different.

Apple's high-level, abstracted OS design allowed to very quickly evolve the poor hardware platform over the time, with little changes from the application side. Customers were used to the eye-candy UI, but it was also very functional and that's why the UI's primary goal.

Amiga users were on very limited part tech savvy. Most of them were just gamers, and the UI wasn't important.
However, even focusing only on tech savvies, the primary problem of the platform is that it was totally open and without the required abstraction to let the hardware evolve in better ways.
It was fascinating and enjoyable for those who liked to tinker and get their hands directly on the hardware, but also the cause of the platform's evolutionary problems, which were bound hand and foot to the original specifications.
Quote:

The PPC AmigaNOne primarily delivered eye candy with a minor upgrade to performance.

As I've said on my previous comment, PowerPCs were already at their twilight when the AmigaOne platform was defined.

And, BTW, the first post-Amiga PowerPC systems weren't even label as "AmigaOne". They were just PCs with the x86 CPU replaced by a PowerPC one. And they remained the same even with such label attached.

So, there's really no added value for such platforms.

Paradoxically, Amithlon was much more palatable (as well as technically interesting), and x86 assembly coding was way more easier and fun (compared to PowerPC, of course).
Quote:

PPC is further behind in performance now compared to the competition than the 68k was then and PPC performance/price was poor then and much worse today. How easy migration is is not as important as how large migration is. How large was the migration from 68k Amiga to PPC AmigaNOne in comparison to PPC Macs? How does this compare to 68k Amiga users buying RPi Amiga emulation hardware, V4SA hardware, THEA500 Mini and likely to buy THEA1200?

Indeed. It's a minor thing, basically irrelevant.

The Vampire market, alone, is greater than all post-Amiga PowerPC systems put together.

It's a clear signal. Well, it should be...
Quote:

What kind of 68k Amiga stand alone hardware is required to retain 68k Amiga users and can it be mass produced for mass migration?

There are basically two, completely different markets for the Amiga: games and applications.

Each market needs a completely different hardware platform.

Which means that if you want to provide a single platform for covering both, then it will be "fat" and expensive (e.g. an ASIC is certainly required for the second one).

@cdimauro

Quote:

Paradoxically, Amithlon was much more palatable (as well as technically interesting), and x86 assembly coding was way more easier and fun (compared to PowerPC, of course).

25 years later I still keep holding out for Bernie to GPL the Amithlon and/or Umilator source. We can but dream! 😊

_________________

@IntuitionAmiga: besides the GPL license (I'm totally against such viral licenses. MPL/APL is a much better choice, if the goal is that contributions should be given back), I fully agree.

However and AFAIR, sources got lost, right?

@cdimauro

You’re the opposite extreme to me then. Everything i release is “GPL v3 or later�. Admittedly this is mostly only to piss off the type of people that go on unrelated groups to evangelise Rust etc because for some reason none of them like GPL. 🤣

I thought the same about the source being lost but i could have sworn i saw Bernie say that wasn’t true a couple of years ago, just that it’s in storage and he’s thousands of miles away from it.

_________________

@cdimauro

We Amiga users use PowerPC Amiga because it works like 68k Amiga.
PowerPC Amiga is better than 68k Amiga because is at least as good as cheap pc from win95 era.
Especially in graphics. Last good Amiag chipset was ECS.
68k followers never provide something as good as ECS at rational price.
Radeon R200 cards are great. It is what C= should provide in middle of 90

Switch to pc. It was obvious in 2001 that Amiga should to switch to unix like Apple do.
But x86 followers instead of hard work on Amiga graphics and gui waste time on attacks on PowerPC.
So after almost 25 still no reason to use Amiga like solution on PC.
Windows is just better.

bhabbott Quote:

Why?

A FPGA Amiga chipset would save time and allow bug fixes until it is mature enough for an ASIC. Retro 16-bit chipsets fit in a $5-$15 FPGA and performance is usually adequate. Different chipsets can be swapped in and out of a FPGA to give universal 68k hardware to reach a larger 68k market than just the 68k Amiga market. An always built in ASIC chipset would be useful for control and a GUI where the Amiga chipset would be a good choice so it would depend on how development progresses.

bhabbott Quote:

The AC68080 wasn't designed for embedded so it's hardly surprising that it isn't attractive for that. But why are we even talking about embedded?

Big endian 68k, ColdFire and SuperH hardware were very popular in the embedded market and new 68k hardware with ColdFire enhancements is an easier replacement than ARM hardware with deprecated big endian support. Big endian hardware was more popular for industrial embedded use than little endian hardware and there is still a lot of it out there that has not been replaced. About half of RPi sales are into the embedded market and the idea again is to increase volumes for mass production. The AC68080 was not designed for embedded use because it was planned to be a large FPGA core with desktop performance. Gunnar added back some ColdFire support that he had removed but the encodings may have moved. The SoC price, power and performance likely need to improve to attract embedded customers but an ASIC improves all these areas.

bhabbott Quote:

I only have a V2 Vampire in my A600 but even that is much more powerful than the A3000 with 060 and RTG that I had in 1997. V4 also has 'Super' AGA at blistering speed, high speed IDE, 100Mbit Ethernet, 16 channel sound, 1GB RAM and 3D hardware texturing. It's every 90's Amiga fan's dream and more. Saying it 'misses the mark there too' is just plain wrong. It's not cheap, but period-accurate alternatives are even more expensive without offering nearly as much. The only thing that makes V4 less attractive to me is that it's too powerful. If I develop software for it I will be limiting my audience to other V4 owners, and I will not be achieving my goal of proving what can be done on real 1990's era Amigas. My A600 lies idle most of the time for the same reason.

Most so called Amiga standalone hardware is replacing ancient 68k Amiga hardware with ARM hardware. This ARM virtual Amiga hardware is winning the battle against the Vamp/AC hardware largely due to better CPU performance and a lower price (ASIC value is better than FPGA value). Even with higher performance and lower priced 68k Amiga hardware, there would still be developers targeting the original Amiga hardware. Right now, developers are developing and testing their 68k Amiga software on non 68k Amiga hardware. Amiga users are using their 68k Amiga software on non 68k Amiga hardware. The 68k Amiga hardware is going extinct instead of proliferating. Modern 68k Amiga hardware could come very close in compatibility with the original hardware. The logic would mostly be the same and clock speeds could be reduced, caches disabled, memory wait states added, etc. A 68000 core could be provided for max CPU compatibility as an iComp accelerator provides in FPGA with a real 68040/68060. Older AmigaOS versions would still work on the hardware and WHDLoad and other patches can fix buggy software.

bhabbott Quote:

Gunnar may try doing that if he wants. Getting enough investment to finance it is a marketing issue, not hardware. You say he made 'poor design decisions like optimizing for a FPGA', but I bet it wouldn't take much to modify it for ASIC. The real question is where is the break-even point? To justify going ASIC you need to get the unit price down below FPGA in a quantity that is saleable. Once again that is mostly a marketing problem.

A quick and easy FPGA to ASIC conversion may give an AC68080@500MHz. A 1GHz-2GHz core would likely require substantial rework. Optimizing for a FPGA is about using standard FPGA blocks efficiently where designing for an ASIC may require building basic blocks and it may be better not to use some FPGA blocks. Multi-level caches are needed and power becomes a concern which is not the case for an AC68080@100MHz in FPGA. Does the AC68080 design use power gating and is it a fully static core design?

https://en.wikipedia.org/wiki/Power_gating
https://en.wikipedia.org/wiki/Static_core

The 68060 design is a full static design, uses power gating of units extensively, uses a decoupled instruction fetch pipeline and execution pipelines with an instruction buffer to decrease instruction fetch power, etc. There are full static 68000 core designs as well. ColdFire designs are 100% synthesizable using auto layout tools for routing. The 68k and ColdFire cores are designed and prepared for an ASIC with low power features. They have undergone extensive verification and testing with some like the 68060 being MC qualified. I doubt the AC68080 core is as ASIC ready as most 68k and ColdFire cores.

bhabbott Quote:

Too many cooks spoil the broth. After getting rid of Gunnar they would fight amongst themselves over what to do, or even worse agree on an unachievable goal like Commodore's engineers did with AAA - and we wouldn't get a finished product.

In my experience, Gunnar is not professional. Jens is more qualified, professional and created the N68050 but Gunnar took over the CPU project as his. Thomas is more talented, professional and created the SAGA chipset but Gunnar took over the SAGA chipset project as his. A bad cook is more likely to spoil the broth than too many cooks.

bhabbott Quote:

Mac users were willing to put up with under-performing hardware because they had been trained to do so from the start. Or were they? The truth is Apple was already losing market share in 1995 despite no longer having competition from Atari and Commodore. Sales dropped by 10% in 1996 and 30% in 1997. By 1998 they only had 2.7% market share. Mac sales didn't take off again until well into the 21st century when they switched to x86.

The initial enthusiasm for PPC from Apple was short lived. High end PPC hardware was expensive and low end hardware for the masses was poor performance. Apple almost went bankrupt in 1997 with perhaps 90 days of cash remaining before they were bailed out by Microsoft.

https://techstartups.com/2023/04/14/in-1997-apple-was-on-verge-of-bankruptcy-with-a-1-billion-loss-now-the-iphone-giant-makes-1-billion-every-3-days/

Apple CEO John Sculley almost killed what became the most valuable public business in the world with PPC. PPC was bad for Motorola and Amiga businesses too. Apple, Motorola and the Amiga should have stayed with the 68k. The Pentium and PPC killer 8-stage 68060 was locked in the Motorola basement at 50MHz instead.

@matthey

Quote:

matthey wrote: . Big endian hardware was more popular for industrial embedded use than little endian hardware and there is still a lot of it out there that has not been replaced.

So markets/products that were low end 20-30 years ago and haven't till now not seen the need for an HW upgrade?

Yeah, those are gonna jump on an untested new chip from a dubious company and off course they will be paying top $ for the privilege.

Might have worked 20 years ago.

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

@matthey

Quote:

matthey wrote: bhabbott Quote: The AC68080 wasn't designed for embedded so it's hardly surprising that it isn't attractive for that. But why are we even talking about embedded? Big endian 68k, ColdFire and SuperH hardware were very popular in the embedded market... About half of RPi sales are into the embedded market and the idea again is to increase volumes for mass production.

The Raspberry Pi uses a Broadcom BCM2835 SoC designed for multimedia devices. The CPU inside it is largely irrelevant - all the other stuff is what matters. The Raspberry Pi foundation figured it could be used to make a small low cost computing computing device for the education/hobbyist market - like Arduino but much more powerful. They managed to convince Broadcom to sell chips to them, and the rest is history.

I watched the RPI's popularity rise with amusement. At first it wasn't even CE approved because they weren't expecting it to be used in commercial products. But as with the Arduino it inspired hobbyists and small-time engineers to do things that were otherwise out of their reach, and then they were tied to the platform. In many cases it was technically a poor choice for their application, but that didn't matter. The important thing was that they could get it to do what they wanted without the knowledge, skills, and infrastructure that was normally required.

No serious full-scale manufacturer would use a RPi in their products, but there are many small-scale cases where a custom design isn't worth the effort. The Raspberry Pi company realized this and developed other products that were more suited to small-scale embedded applications, including the cut down Pi Zero and the Pi Pico (which uses their own chip design with dual ARM CPUs and peripherals similar to other MCUs - a completely different architecture).

The only reason we are interested in RPi is for enhancing our retro Amiga computers. PiStorm has been a hit because it provides extreme acceleration and RTG for about half the price of a Vampire, with the potential advantage of open source code. The actual CPU inside it is irrelevant. What's important is that it's available now and in active development at a relatively low price (about the same as a 50MHz 030 board).

Could a similar chip with embedded 68k CPU be better for us? Sure, but that doesn't exist and won't exist because there isn't a big enough market for it. Nobody (apart from fans of retro 68k machines) is interested in a 68k SoC because there are already other chips that do the job. The Pi 5 now has a 2.4 GHz quad-core 64-bit ARM Cortex-A76 CPU and VideoCore VII GPU making it twice as powerful as the Pi 4. Pi users aren't wanting for performance. 68k is completely irrelevant to the embedded world.

Quote:

Most so called Amiga standalone hardware is replacing ancient 68k Amiga hardware with ARM hardware. This ARM virtual Amiga hardware is winning the battle against the Vamp/AC hardware largely due to better CPU performance and a lower price (ASIC value is better than FPGA value).

By many fans are not so concerned about getting the highest possible CPU performance. Vampire offers things you can't get in PiStorm that are more important to us. The reasons for buying a PiStorm (apart from novelty) are 1. the lower price and 2. the possibility (however remote) of getting what you want, not what Gunnar wants.

Quote:

Even with higher performance and lower priced 68k Amiga hardware, there would still be developers targeting the original Amiga hardware.

Yep. People like me. Nothing less (or more) than the real thing is good enough!

Quote:

Right now, developers are developing and testing their 68k Amiga software on non 68k Amiga hardware. Amiga users are using their 68k Amiga software on non 68k Amiga hardware. The 68k Amiga hardware is going extinct instead of proliferating.

Cross-developing on a 'modern' PC has numerous advantages. You can use inefficient tools that would run far too slow on original Amiga hardware, in a more comfortable environment. You can debug your code in an Emulator which provides more insight into what the hardware is doing. I have done this for the Amstrad CPC and Mattel Aquarius and it was great. However I enjoy developing stuff on my A1200 just like I used to in the 90's, so it enhances the retro experience whereas doing it on a PC wouldn't.

68k Amiga hardware isn't going extinct. It is getting more expensive as hoarders collect more than they need, but eventually they will die or move on and their hoards will be released. Meanwhile replacement hardware is being manufactured - just need the price of original hardware to increase a bit more to make it viable. What we need now are replacement custom chips, as most of the spares seem to have been hoovered up.

Or do we? I am quite happy with the Amigas I have now. I passed over several opportunities to acquire other models because I knew they wouldn't get used. What's really going extinct isn't Amiga hardware, it's us.

@IntuitionAmiga

Quote:

IntuitionAmiga wrote: @cdimauro You’re the opposite extreme to me then. Everything i release is “GPL v3 or later�. Admittedly this is mostly only to piss off the type of people that go on unrelated groups to evangelise Rust etc because for some reason none of them like GPL. 🤣

ROFL Unfortunately and very likely, he's one of the "new wave" of blind Rust fanaticals. I've met some of them and they are really like Talibans ("they saw the light").

To me it's different. Usually I don't share my sources, even the old ones. If I decide to do it, then I prefer the BSD license, because I prefer that anyone can enjoy my code as it wants. But if recognize that some code deserves to have any changes given back, then I chose the MPL.

Being totally transparent and honest (as usual!), I don't like to force people to share their sources just because they use some library and/or some piece of code of someone else. That's the reason why I'm against viral license like GPL.
Quote:

I thought the same about the source being lost but i could have sworn i saw Bernie say that wasn’t true a couple of years ago, just that it’s in storage and he’s thousands of miles away from it.

Looks like the pirate's treasure.

Strange that he has no chance to go back there for any reason.

Anyway, at least we know that it's not lost.

@ppcamiga1

Quote:

ppcamiga1 wrote: @cdimauro We Amiga users

No, YOU and SOME other Amiga users.
Quote:

use PowerPC Amiga

Again, there's no PowerPC Amiga. BY DEFINITION.

You can dream whatever you like, but the reality is totally different: those are PCs (yes, PCs!) with the x86 CPU replaced by a PowerPC one. Dot.

Sorry for breaking your wet dreams...
Quote:

because it works like 68k Amiga.

First of all, Amiga is using a 68k (from Motorola). BY DEFINITION.

Second, how? Is there any custom chip like the Amiga? No. Dot.

Sorry breaking again your wet dreams?
Quote:

PowerPC Amiga is better than 68k Amiga

See above: you can't compared things which don't exists with the ones which exists. BY ELEMENTARY LOGIC.
Quote:

because is at least as good as cheap pc from win95 era.

Which means which is supercrap, since Amiga had accelerators and/or additional cards which did better than a PC from 1995.
Quote:

Especially in graphics. Last good Amiag chipset was ECS. 68k followers never provide something as good as ECS at rational price.

Quote:

Radeon R200 cards are great. It is what C= should provide in middle of 90

I reveal you a secret: Commodore went bankrupt BEFORE that...
Quote:

ppcamiga1 wrote: Switch to pc.

Already did on 1996 (when my A1200 broke all video outputs).
Quote:

It was obvious in 2001 that Amiga should to switch to unix like Apple do.

Then... why OS4?
Quote:

But x86 followers instead of hard work on Amiga graphics and gui

I reveal you another secret: x86 followers don't do that. BY LOGIC, since x86 == PC and Amiga == Motorola 68k + Commodore's custom chipset + Commodore's Kickstart ROM.
Quote:

waste time on attacks on PowerPC.

Why attacking a dead corpse?
Quote:

So after almost 25 still no reason to use Amiga like solution on PC.

Well, there are strong reasons for doing it: it's simple that PowerPC fanatics can't accept it.
Quote:

Windows is just better.

Overall, yes. But not as lightweight like AROS and it's missing some features which were unique to the Amiga.

So, that's the reason why people are still using Amiga-like solutions on PCs.

And I reveal you another secret: this platform is evolving to 64-bit (and partially SMP): something which is simply impossible for the current PowerPC solutions.

@matthey

Quote:

matthey wrote: Most so called Amiga standalone hardware is replacing ancient 68k Amiga hardware with ARM hardware. This ARM virtual Amiga hardware is winning the battle against the Vamp/AC hardware largely due to better CPU performance and a lower price (ASIC value is better than FPGA value). Even with higher performance and lower priced 68k Amiga hardware, there would still be developers targeting the original Amiga hardware. Right now, developers are developing and testing their 68k Amiga software on non 68k Amiga hardware.

That's not a bad idea, because it's much more productive developing on another platform (like a PC). Not only for the more advanced tools, but also because it's faster for testing.

You can still develop on an Amiga, but there aren't so many good tools ported. And "freeze/Guru and reboot" isn't the best experience for a developer.

Many platforms (especially consoles) already have/use "alien" platforms for the development of their software, for the same reasons.

The important thing is... developing! It doesn't matter where/how.

@bhabbott

Quote:

bhabbott wrote: The only reason we are interested in RPi is for enhancing our retro Amiga computers. PiStorm has been a hit because it provides extreme acceleration and RTG for about half the price of a Vampire, with the potential advantage of open source code.

There's no open source, and you should know it: only promises since SEVERAL YEARS.
Quote:

The actual CPU inside it is irrelevant. What's important is that it's available now and in active development at a relatively low price (about the same as a 50MHz 030 board). Could a similar chip with embedded 68k CPU be better for us? Sure, but that doesn't exist and won't exist because there isn't a big enough market for it. Nobody (apart from fans of retro 68k machines) is interested in a 68k SoC because there are already other chips that do the job. The Pi 5 now has a 2.4 GHz quad-core 64-bit ARM Cortex-A76 CPU and VideoCore VII GPU making it twice as powerful as the Pi 4. Pi users aren't wanting for performance. 68k is completely irrelevant to the embedded world.

There's market, even on the embedded world.

It's the product (68k. Or other), which is missing.

@cdimauro

Quote:

There's no open source, and you should know it: only promises since SEVERAL YEARS.

My controversial take, bhabbott was talking about Pistorm, not vampire.

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga