This time we talk about innovations to the graphics in this article of the new mini-series dedicated to the improvements that would have been possible to the Amiga's chips in order to make the platform evolve better and make it more competitive.

English: https://www.appuntidigitali.it/22547/missed-opportunities-to-improve-the-amiga-chipset-2-graphics/

Italian: https://www.appuntidigitali.it/22440/le-occasioni-mancate-per-migliorare-il-chipset-dellamiga-2-la-grafica/

@cdimauro
Where are people like Lou and Hammer saying how crap the Amiga chipset was to be missing so many features as they compare to later and more expensive hardware? People forget how early and primitive chipset hardware and CPUs were still in 1985. It wasn't that the 68000 Amiga was bad but rather that the 68k Amiga was good, revolutionized the market, was copied and then was replaced while CBM stood by waiting for the tiny chipset to drop in cost enough to turn into the next C64. They didn't even try to cost reduce the 68k Amiga chipset and memory. Such simple chipset changes could have saved time and money. Innovation at Amiga Corporation became necessity only at CBM. It's too bad the Ranger chipset specs never resurfaced to see if Jay slipped in a few ideas despite the oppressive CBM management and directives.

@matthey: Indeed.

The Amiga was a revolution when it was introduced, despite all limitations (and budget) of the time. It literally changed the world and how personal computing was.

However, after that:

"We made Amiga, They fucked it up"

They = ALL people which were in charge after that the original team left.

Only one thing: the Ranger doesn't look a great improvement to me. As a developer, I would have liked more the innovations which Im talking about : very cost effective and which helped on my real needs. VRAM = more bandwidth was were good to have, but too much expensive.

cdimauro Quote:

Indeed. The Amiga was a revolution when it was introduced, despite all limitations (and budget) of the time. It literally changed the world and how personal computing was.

Jay Miner and the Amiga get pretty much zero recognition too. Some of it is the embarrassing CBM as Chuck Peddle and the CBM Pet are not recognized much either. The C64 is more recognized based on being the highest selling single computer model, perhaps until one of the RPi models? Computer history is all about Steve Jobs and Bill Gates even though they were no where near the innovators and pioneers. Steve Wozniak was more of a pioneer than either of them.

cdimauro Quote:

However, after that: "We made Amiga, They fucked it up" They = ALL people which were in charge after that the original team left.

CBM hadn't even done much to f it up by that time. The philosophies at Amiga Corporation and CBM certainly didn't mesh and the friction grew worse due to the Easter egg in ROM that caused CBM to pull Amigas from shelves slowing the already slow launch.

cdimauro Quote:

Only one thing: the Ranger doesn't look a great improvement to me. As a developer, I would have liked more the innovations which Im talking about : very cost effective and which helped on my real needs. VRAM = more bandwidth was were good to have, but too much expensive.

There should have been more of the simple enhancements that increase value like you suggest. They would have ended up in ECS with the CBM time line which wouldn't be so bad if ECS had been out in 1987-1989. There is still no chip mem bandwidth increase or color upgrades though. I understand the cost factor. That is why I suggested a high end chipset and a low end chipset which is eventually upgraded to the high end chipset specs. Upgraded chipsets are going to be developed anyway and developers would have been able to test their software on high end models before the low end models get it, improving compatibility. The Amiga needed a higher end chipset for high end Amigas to stay relevant and, in the late 1980s and early 1990s, there was a great divide between high end graphics with VRAM and low end graphics with DRAM. Also, VRAM is a great fit for the Amiga as the CPU and chipset each get their own port to the dual ported memory. I believe VRAM was affordable enough for the low end by the early 1990s although there may have been better options to increase chip memory bandwidth by that time. One thing is for sure, the Amiga didn't really have a high end system available in the 1990s. ECS in the Amiga 3000 was far from high end and AGA arrived too late to ever be high end.

@matthey

Quote:

matthey wrote: Quote: cdimauro However, after that: "We made Amiga, They fucked it up" They = ALL people which were in charge after that the original team left. CBM hadn't even done much to f it up by that time. The philosophies at Amiga Corporation and CBM certainly didn't mesh and the friction grew worse due to the Easter egg in ROM that caused CBM to pull Amigas from shelves slowing the already slow launch.

You're right. That was a bad move. Even because Commodore saved them and the Amiga project, and they got a lot of money due to acquisition.
Quote:

cdimauro Quote: Only one thing: the Ranger doesn't look a great improvement to me. As a developer, I would have liked more the innovations which Im talking about : very cost effective and which helped on my real needs. VRAM = more bandwidth was were good to have, but too much expensive. There should have been more of the simple enhancements that increase value like you suggest. They would have ended up in ECS with the CBM time line which wouldn't be so bad if ECS had been out in 1987-1989. There is still no chip mem bandwidth increase or color upgrades though.

Yes, more was needed for sure.

On another article I'll write a time line of what could have been possible in different periods of time (1987, 1990, 1992 at least), taking into account the technology progresses and the Amiga market(s).
Quote:

I understand the cost factor. That is why I suggested a high end chipset and a low end chipset which is eventually upgraded to the high end chipset specs. Upgraded chipsets are going to be developed anyway and developers would have been able to test their software on high end models before the low end models get it, improving compatibility. The Amiga needed a higher end chipset for high end Amigas to stay relevant and, in the late 1980s and early 1990s, there was a great divide between high end graphics with VRAM and low end graphics with DRAM.

The problem of high-end chipsets is that you've to duplicate the projects and the number of chips to be released. Even a simple revision of a chips costed some millions at the time AFAIR.

For this reason Commodore had developed just a single set of chips to be used on all its machines.

You can see it with the ECS version used for Amiga 3000, which had a 16 bit bus despite the system being 32-bit and, what's even worse, the chipset was still castrated forcing the faster CPUs to wait for 4 cycles to access the Chip mem, even if it was possible doing it in two cycles. Here, it could have been possible to do have a single chip working well on both 16 and 32-bit machines if engineers were competent (which wasn't the case as we know).
Quote:

Also, VRAM is a great fit for the Amiga as the CPU and chipset each get their own port to the dual ported memory. I believe VRAM was affordable enough for the low end by the early 1990s although there may have been better options to increase chip memory bandwidth by that time.

I still see VRAM as a too much expensive solution, whereas 140ns (7Mhz) DRAM (instead of the standard 280ns -> 3.5Mhz used on OCS) for a 14Mhz chipset was much cheaper, and paired with a 14Mhz 68000 it could have boosted A LOT the entire platform make it competitive with the consoles & PCs of the time (especially if some of my suggestions would have been implemented as well).
Quote:

One thing is for sure, the Amiga didn't really have a high end system available in the 1990s. ECS in the Amiga 3000 was far from high end and AGA arrived too late to ever be high end.

Absolutely. Too much incompetence.

At the AGA time it could have been possibile to use 70ns DRAM -> 14Mhz DRAM for a 28Mhz chipset, which paired with the 32 bit bus would have boosted 8 x (compared to the OCS) the overall system bandwidth, which would have allowed to reach a new level of performance (especially if paired with a 28Mhz 68EC020).

Last edited by cdimauro on 29-Jul-2024 at 05:52 AM.

@cdimauro

Quote:

cdimauro wrote: Absolutely. Too much incompetence.[quote] If only Commodore had employed you to design their new chipset, eh? [quote]At the AGA time it could have been possibile to use 70ns DRAM -> 14Mhz DRAM for a 28Mhz chipset, which paired with the 32 bit bus would have boosted 8 x (compared to the OCS) the overall system bandwidth,

'Possible', but would involve changing ChipRAM timing which would risk breaking compatibility. Also the DRAMs were the most expensive chips in the A1200. Requiring the fastest chips would increase the price and could have supply issues. By the time Commodore produced the A1200 their financial position was so bad that suppliers were asking for cash up front - another reason to avoid using the most expensive ships.

Quote:

which would have allowed to reach a new level of performance (especially if paired with a 28Mhz 68EC020).

No such thing as a 28MHz 68EC020.

Hey, I get it. With the benefit of hindsight we can all imagine ways the A1200 could have been made better. But we weren't there, trying to make something that worked well enough to invest millions of dollars into when the future was far from certain. When you think about everything that had to come together to produce the A1200, it's a miracle that they made it as good as they did (so many ways they could have screwed it up).

IMO we couldn't ask for more. A more 'competent' company would have avoided the Amiga and gone all-in on PCs - and the World would be a poorer place for it. There would be no jealous Amiga fans moaning about 'missed opportunities' and 'too much incompetence' - just sad PC fans arguing about which x86 CPU or graphics card was best - and we wouldn't know what we were missing. Nobody would even dream of an alternative system so awesome that fans would still be using it 30 years later.

@matthey

Quote:

matthey wrote: @cdimauro Where are people like Lou and Hammer saying how crap the Amiga chipset was to be missing so many features as they compare to later and more expensive hardware? People forget how early and primitive chipset hardware and CPUs were still in 1985. It wasn't that the 68000 Amiga was bad but rather that the 68k Amiga was good, revolutionized the market, was copied and then was replaced while CBM stood by waiting for the tiny chipset to drop in cost enough to turn into the next C64. They didn't even try to cost reduce the 68k Amiga chipset and memory. Such simple chipset changes could have saved time and money. Innovation at Amiga Corporation became necessity only at CBM. It's too bad the Ranger chipset specs never resurfaced to see if Jay slipped in a few ideas despite the oppressive CBM management and directives.

1. That's a false narrative about me.

2. A500 was competitive from 1987 to 1990. The problem is in the 1990s.

3. For $1000 USD or $1500 AUD in 1993, the Amiga solution didn't deliver 486SX-33 level performance when the Amiga AGA solution was the slower 68030-50 based. 68EC040-25 is brain-dead for the desktop 68K with DMA.

Amiga's 3rd party accelerator doesn't have economies of scale.

Motorola didn't factor in price-disruptive X86 clones.

Should I post more Data Quest reports on how Motorola's pricing policies were caught off guard?

4. AGA can be sufficient, IF the "game ready" optimized C2P software example is part of Commodore's SDK in 1992-1993 instead of giving up!

https://bigbookofamigahardware.com/bboah/product.aspx?id=1604
From Commodore's engineers,
Quote:

The “chunky to planar� logic was thought out in a lunchtime conversation between Beth Richard (system chip design), Chris Coley (board design), and Ken Dyke (software) over Subway sandwiches on a picnic table in a nearby park one day, because Ken was telling us how much of a pain it was to shuffle bits in software to port games from other platforms to the Amiga planar system. We took the idea to Hedley Davis, who was the system chip team manager and lead engineer on Akiko and he said we could go ahead with it. I showed him the “napkin sketch� of how I thought the logic would work and was planning on getting to it the next day as it was already late afternoon by that point. I came in the next morning and Hedley had completed it already, just from the sketch!

John Carmack repeated similar statements in 1994!

Regardless of AGA or AAA, 68EC020-14 is not performance-competitive for a texture-mapped 3D game experience.

$4 difference separates 68EC020-14 and 68EC020-25.



CD32's FMV module has the following:

1. 24-bit DAC (STM's STV8438CV) for 16.7 million colors display.

2. MPEG-1 decoder from C-Cube CL450, 352 x 240 pixels @ 30hz, 352 x 288 pixels at 25 Hz, pixel interpolation and frame duplication to produce output formats of 704 x 240 pixels at 60 Hz or 704 x 288 pixels at 50 Hz.

https://websrv.cecs.uci.edu/~papers/mpr/MPR/ARTICLES/060803.PDF

CL450 has about 398K transistors with up to 40 MHz. CL450 includes a licensed MIPS-X RISC processor with semi-custom extensions. In quantities of 100K or more per year, the price is less than $50 in 1992.
CL450's MIPS-X RISC processor still has the usual RISC instruction set.

3. LSI l64111qc (Digital Audio Decoder, 16-bit DAC),

4. 512 KB local RAM, NEC 423260 DRAM 4Mbit (512 KB) with 80 ns.

5. Lattice ispLSI 1024-60LJ CPLD.

Commodore is willing to spend on this non-core business by following the failed CDI.

Commodore says NO to MIPS RISC processor @ 40 Mhz for Amiga's general-purpose games.

Commodore says NO to 512 KB Fast RAM for Amiga's general-purpose games.

Commodore says NO to 24-bit color 704 x 288p for Amiga's general-purpose games.

Commodore says NO to 16-bit stereo audio for Amiga's general-purpose games.

The argument against $20 DSP3210 is implemented on CD32's higher cost FMV module with a very narrow functionality!

My Dad purchased an ex-corporate A3000 in early 1992 and A1200 is largely missing in action in Australia's Xmas 1992 market! A4000/040 is expensive like a no-name Pentium 60 PC clone. A4000/030 missed Xmas 1992 sales.

For Xmas 1992, my Dad purchased an i386DX-33/ET4000-based gaming PC clone instead.

For 256-color gaming, the A3000 is a lesson on dead-end graphics evolution with unpartitioned Amiga graphics architecture. My Dad has banned all future Commodore purchases in our family. There is dead-end ECS bitterness. My A1200 purchase is my decision with my money.

Quote:

matthey wrote: There should have been more of the simple enhancements that increase value like you suggest. They would have ended up in ECS with the CBM time line which wouldn't be so bad if ECS had been out in 1987-1989. There is still no chip mem bandwidth increase or color upgrades though. I understand the cost factor. That is why I suggested a high end chipset and a low end chipset which is eventually upgraded to the high end chipset specs. Upgraded chipsets are going to be developed anyway and developers would have been able to test their software on high end models before the low end models get it, improving compatibility. The Amiga needed a higher end chipset for high end Amigas to stay relevant and, in the late 1980s and early 1990s, there was a great divide between high end graphics with VRAM and low end graphics with DRAM. Also, VRAM is a great fit for the Amiga as the CPU and chipset each get their own port to the dual ported memory. I believe VRAM was affordable enough for the low end by the early 1990s although there may have been better options to increase chip memory bandwidth by that time. One thing is for sure, the Amiga didn't really have a high end system available in the 1990s. ECS in the Amiga 3000 was far from high end and AGA arrived too late to ever be high end.

3DO's MADAM (Agnus counterpart with a matrix math co-processor and texture mapper known as CEL) has a hybrid design with can access both 2MB FP DRAM and 1MB VRAM memory pools. Matrix math co-processor is CPU PIO driven.

"Super A500" has up to 1 MB VRAM and the intended A3000 has up to 2MB VRAM. These are expensive VRAM configurations from 1987-1990.

3DO's MADAM has lessons from the Super A500's VRAM or bust mindset.

Around 1987-1988, 68020-16 had a $77 asking price. MIPS R3000 is winning Unix design wins during 1988. Intel 486DX was released in 1989 to counter it. 68040 was late and it was released in 1990. Many 68K Unix vendors are jumping ship to RISC CPU designs.

IBM 8514 has 512 KB (40 ns serial access) VRAM.

Tseng Labs ET4000AX used interleaved memory controllers with 80 ns access FP DRAM (e.g. Sanyo LC324256AP-80) to deliver "VRAM-like" performance.

Smart interleaved memory controller wasn't Commodore's forte.

Last edited by Hammer on 30-Jul-2024 at 12:59 PM.
Last edited by Hammer on 30-Jul-2024 at 12:56 PM.
Last edited by Hammer on 30-Jul-2024 at 12:55 PM.
Last edited by Hammer on 30-Jul-2024 at 12:54 PM.
Last edited by Hammer on 30-Jul-2024 at 12:51 PM.
Last edited by Hammer on 30-Jul-2024 at 12:24 PM.
Last edited by Hammer on 30-Jul-2024 at 12:23 PM.
Last edited by Hammer on 30-Jul-2024 at 12:20 PM.
Last edited by Hammer on 30-Jul-2024 at 12:16 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@bhabbott

Quote:

No such thing as a 28MHz 68EC020. Hey, I get it. With the benefit of hindsight we can all imagine ways the A1200 could have been made better. But we weren't there, trying to make something that worked well enough to invest millions of dollars into when the future was far from certain. When you think about everything that had to come together to produce the A1200, it's a miracle that they made it as good as they did (so many ways they could have screwed it up).

Amiga accelerators with 28 Mhz 68020 are slightly overclocked 68020-25 e.g. http://amiga.resource.cx/exp/apollo1220

Apollo 1220 Turbo's 28Mhz 68020 used a 68020-25 model.

Quote:

'Possible', but would involve changing ChipRAM timing which would risk breaking compatibility. Also the DRAMs were the most expensive chips in the A1200. Requiring the fastest chips would increase the price and could have supply issues. By the time Commodore produced the A1200 their financial position was so bad that suppliers were asking for cash up front - another reason to avoid using the most expensive ships.

A1200 has a "healthy profit margin", but it wouldn't matter anyway with CD32's reduced profit margin with a lower asking price.

Since Commodore has thrown away A1200's "healthy profit margin" for a lower profit margin with a lower asking price of CD32, A1200 could have supported an extra $50 BOM cost components.

Commodore offered FMV module for about USD $250 or 199 UKP with $50 MIPS-X @40Mhz based SoC with 512 KB Fast RAM (80 ns access FP DRAM, about $13) + 24-bit display DAC + 16-bit stereo audio DAC equipped

Last edited by Hammer on 30-Jul-2024 at 01:16 PM.
Last edited by Hammer on 30-Jul-2024 at 01:14 PM.
Last edited by Hammer on 30-Jul-2024 at 01:11 PM.
Last edited by Hammer on 30-Jul-2024 at 01:05 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@cdimauro

Quote:

You can see it with the ECS version used for Amiga 3000, which had a 16 bit bus despite the system being 32-bit and, what's even worse, the chipset was still castrated forcing the faster CPUs to wait for 4 cycles to access the Chip mem, even if it was possible doing it in two cycles. Here, it could have been possible to do have a single chip working well on both 16 and 32-bit machines if engineers were competent (which wasn't the case as we know).

FYI, A3000's Chip RAM is 32-bit wide for the CPU i.e. 7.1 MB/s limit like on AGA Amigas

https://aminet.net/package/util/moni/bustest
A3000's bustest chip ram writel reached 7MB/s which is 2X faster than A500's.

Only AGA's Lisa can make use of A3000's faster 32-bit Chip RAM.

A3000's Chip RAM used 120 ns access DRAM chips i.e. eight 44256k 120 ns DRAM DIPs for 1 MB Chip RAM.

A3000's ZIP RAM can be 80 ns access. There's no graphics IP on A3000's Fast RAM.

A3000's 32-bit Chip RAM led to AA3000+ evolution.

Last edited by Hammer on 30-Jul-2024 at 12:38 PM.
Last edited by Hammer on 30-Jul-2024 at 12:36 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Hammer wrote: 1. That's a false narrative about me. 2. A500 was competitive from 1987 to 1990. The problem is in the 1990s. 3. For $1000 USD or $1500 AUD in 1993, the Amiga solution didn't deliver 486SX-33 level performance when the Amiga AGA solution was the slower 68030-50 based. 68EC040-25 is brain-dead for the desktop 68K with DMA.

To be fair we have to compare like for like. In 1993 the A1200 was much cheaper than a 486SX-33 system. So take stuff out of that PC until the price matches - now which one does the better job?

No, that wasn't the problem. The problem was 90% of the market was PCs. They were everywhere - so everybody wanted one. And everybody wanted software that ran on them, so everybody was developing software for them. If by some miracle Commodore had managed to make the A1200 more powerful than a 486SX-33 for half the price it wouldn't have mattered. Of course in reality this was impossible - the best they could hope for was making it for the same price. Now customers have a choice - an industry standard PC, or some weird incompatible home computer. Of course they take the PC every time.

We can pontificate about how the Amiga chipset could have been made better, but this is just an academic exercise that has no bearing on how 'competitive' it would have been. The die was cast in 1981 when IBM introduced the PC, and it soon became obvious that no other platform would survive.

@bhabbott

Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: Absolutely. Too much incompetence.[quote] If only Commodore had employed you to design their new chipset, eh? As I've reported several times, I'm NOT an hardware engineer. But I'm skilled developer, I know how the hardware worked, and I also know how to evolve it to have a better system. Quote: [quote]At the AGA time it could have been possibile to use 70ns DRAM -> 14Mhz DRAM for a 28Mhz chipset, which paired with the 32 bit bus would have boosted 8 x (compared to the OCS) the overall system bandwidth, 'Possible', but would involve changing ChipRAM timing which would risk breaking compatibility.

Everything which I'm writing on my articles are 100% backward-compatible.
Quote:

Also the DRAMs were the most expensive chips in the A1200. Requiring the fastest chips would increase the price and could have supply issues. By the time Commodore produced the A1200 their financial position was so bad that suppliers were asking for cash up front - another reason to avoid using the most expensive ships.

That's why I've talked about "AGA" before and not about "A1200".

If 70ns DRAMs on 1992 were too much expensive, then those could have been used on high-end systems (A4000).
Quote:

Quote: which would have allowed to reach a new level of performance (especially if paired with a 28Mhz 68EC020). No such thing as a 28MHz 68EC020.

Do you mean produced by Motorola?

Worst case, you can take a 25Mhz one and slightly overclock. The Fab process for those old chips were very mature, at the time.
Quote:

Hey, I get it. With the benefit of hindsight we can all imagine ways the A1200 could have been made better. But we weren't there, trying to make something that worked well enough to invest millions of dollars into when the future was far from certain.

No, the changes which I've proposed are minimal.
Quote:

When you think about everything that had to come together to produce the A1200, it's a miracle that they made it as good as they did (so many ways they could have screwed it up).

Well, they HAVE screwed it in several ways. That's why it's a big patchwork with little value.

I loved it, because it was very cheap, affordable for my pockets, and allowed me to switch to something better. But it's also very limited (less money, less features) and crippled by our marvelous engineers...
Quote:

IMO we couldn't ask for more.

IMO yes. A lot more, with minimal changes.
Quote:

A more 'competent' company would have avoided the Amiga and gone all-in on PCs - and the World would be a poorer place for it. There would be no jealous Amiga fans moaning about 'missed opportunities' and 'too much incompetence' - just sad PC fans arguing about which x86 CPU or graphics card was best - and we wouldn't know what we were missing. Nobody would even dream of an alternative system so awesome that fans would still be using it 30 years later.

Stop this dummy PC rethoric that you're continuously bringing to just the failure of Commodore: other companies survived.

Diversity was NOT a showstopper for other systems. You can handle it.

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: You can see it with the ECS version used for Amiga 3000, which had a 16 bit bus despite the system being 32-bit and, what's even worse, the chipset was still castrated forcing the faster CPUs to wait for 4 cycles to access the Chip mem, even if it was possible doing it in two cycles. Here, it could have been possible to do have a single chip working well on both 16 and 32-bit machines if engineers were competent (which wasn't the case as we know). FYI, A3000's Chip RAM is 32-bit wide for the CPU i.e. 7.1 MB/s limit like on AGA Amigas https://aminet.net/package/util/moni/bustest A3000's bustest chip ram writel reached 7MB/s which is 2X faster than A500's.

It could have been DOUBLE only if we had some normal (not even good are required here) engineers that allowed the processor to access Chip mem every two cycles instead of every four.

Only Amiga engineers made it possible...
Quote:

Only AGA's Lisa can make use of A3000's faster 32-bit Chip RAM.

At least...
Quote:

A3000's Chip RAM used 120 ns access DRAM chips i.e. eight 44256k 120 ns DRAM DIPs for 1 MB Chip RAM. A3000's ZIP RAM can be 80 ns access. There's no graphics IP on A3000's Fast RAM.

Interesting, and that was on 1990. How about 70ns DRAMs on 1992: were they so much expensive?
Quote:

A3000's 32-bit Chip RAM led to AA3000+ evolution.

With the same crippled Chip mem access...

@bhabbott

Quote:

bhabbott wrote: @Hammer Quote: Hammer wrote: 1. That's a false narrative about me. 2. A500 was competitive from 1987 to 1990. The problem is in the 1990s. 3. For $1000 USD or $1500 AUD in 1993, the Amiga solution didn't deliver 486SX-33 level performance when the Amiga AGA solution was the slower 68030-50 based. 68EC040-25 is brain-dead for the desktop 68K with DMA. To be fair we have to compare like for like. In 1993 the A1200 was much cheaper than a 486SX-33 system. So take stuff out of that PC until the price matches - now which one does the better job? No, that wasn't the problem. The problem was 90% of the market was PCs. They were everywhere - so everybody wanted one. And everybody wanted software that ran on them, so everybody was developing software for them. If by some miracle Commodore had managed to make the A1200 more powerful than a 486SX-33 for half the price it wouldn't have mattered. Of course in reality this was impossible - the best they could hope for was making it for the same price. Now customers have a choice - an industry standard PC, or some weird incompatible home computer. Of course they take the PC every time. We can pontificate about how the Amiga chipset could have been made better, but this is just an academic exercise that has no bearing on how 'competitive' it would have been. The die was cast in 1981 when IBM introduced the PC, and it soon became obvious that no other platform would survive.

Are you PC envy?

@bhabbott
Quote:

To be fair we have to compare like for like. In 1993 the A1200 was much cheaper than a 486SX-33 system. So take stuff out of that PC until the price matches - now which one does the better job?

For games, SNES dominated the US low-end gaming market and killed the C64.

For the delivered "strong 2D gaming experience", SNES dominated the US low-end priced gaming market.

A gaming PC would lose against SNES if it delivered a similar 2D gaming experience as SNES, but a gaming PC's higher price has delivered a superior texture-mapped 3D gaming experience.

A1200 has the potential to deliver SNES level 2D gaming experience, but A1200 has a higher price.

The higher price for A1200's desktop computing freedom is not enough to negate SNES's strong 2D gaming experience and gaming PC's higher price with superior texture-mapped 3D gaming experience.

Amiga's non-gaming user base is very weak i.e. it's not Mac userbase scale.

The Amiga was squeezed out of the mainstream market while Sony found a place for PlayStation.


For most end users, the type of delivered gaming experience is an important factor over raw hardware specifications.


"What if" games like Dread or Grind could have delivered a good texture-mapped 3D gaming experience for stock A500 and A1200, but this requires extra programming effort.

Last edited by Hammer on 31-Jul-2024 at 07:29 AM.
Last edited by Hammer on 31-Jul-2024 at 07:01 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@cdimauro

Quote:

It could have been DOUBLE only if we had some normal (not even good are required here) engineers that allowed the processor to access Chip mem every two cycles instead of every four. Only Amiga engineers made it possible...

A3000 has bundled with some nice new RAM chips... where's the new graphics IP?

I upgraded my A3000 with 4MB static column ZIP RAM for 030's burst mode and Amiga graphics IP is largely frozen in time.

GVP's EGS RTG graphics cards were expensive when compared to PC SVGA counterparts, hence A3000's inactive inline ISA slots' BOM cost is wasted. The work from PCMCIA could bridge the ISA slots for Super Buster.

Later 3rd party busboard has active PCI slots, but they are too late.

Commodore issued their FUD PR against GVP's EGS RTG with promises of Commodore RTG and it worked.

Phase 5's CyberGraphicsX RTG API was allowed to live without Commodore's PR FUD.

Another problem is Zorro II and Zorro III markets didn't access A500/A1200's economies of scale.

Last edited by Hammer on 31-Jul-2024 at 07:16 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: It could have been DOUBLE only if we had some normal (not even good are required here) engineers that allowed the processor to access Chip mem every two cycles instead of every four. Only Amiga engineers made it possible... A3000 has bundled with some nice new RAM chips...

Sure, and the processor was using HALF of the available bandwidth because those geniuses weren't able to check the CPU memory requests every two cycles instead of every four: what a gargantuan endeavour required it...
Quote:

where's the new graphics IP?

Well, this you've to ask to the above geniuses.
Quote:

I upgraded my A3000 with 4MB static column ZIP RAM for 030's burst mode and Amiga graphics IP is largely frozen in time. GVP's EGS RTG graphics cards were expensive when compared to PC SVGA counterparts, hence A3000's inactive inline ISA slots' BOM cost is wasted. The work from PCMCIA could bridge the ISA slots for Super Buster. Later 3rd party busboard has active PCI slots, but they are too late. Commodore issued their FUD PR against GVP's EGS RTG with promises of Commodore RTG and it worked. Phase 5's CyberGraphicsX RTG API was allowed to live without Commodore's PR FUD. Another problem is Zorro II and Zorro III markets didn't access A500/A1200's economies of scale.

Not even counting the bugs of Zorro III which crippled the memory bandwidth.

How long do you still want to defend those incompetents?

@cdimauro

Quote:

cdimauro wrote: You can see it with the ECS version used for Amiga 3000, which had a 16 bit bus despite the system being 32-bit and, what's even worse, the chipset was still castrated forcing the faster CPUs to wait for 4 cycles to access the Chip mem, even if it was possible doing it in two cycles. It could have been DOUBLE only if we had some normal (not even good are required here) engineers that allowed the processor to access Chip mem every two cycles instead of every four.

So Commodore's engineers were below 'normal'? Right.

Please explain how this double speed access would work.

@cdimauro

Quote:

cdimauro wrote: Not even counting the bugs of Zorro III which crippled the memory bandwidth. How long do you still want to defend those incompetents?

Dave Haytnie: Zorro III vs PCI
Quote:

"Part of the problem with so many armchair quarterbacks in this business is that they don't know the Amiga as a SYSTEM. A system engineer understands both the software and hardware, because in fact, until you get to implementation details, they're the same exact thing. Hackers rarely learn this approach, it's even rare among professionals. In fact, at my first job, I was one of about 10 new hires at General Electric, the only people in the company even ALLOWED to touch both hardware and software at the same time." "Of course you didn't forget that I did Buster. What you seem to forget is that I did all of Zorro III -- the specification, the first controllers, the first cards. Of course the Buster chip was a problem, though no worse, really, than Intel's first crack at a PCI controller. Of course, they've had seven or so architectural generations to improve on the controller. Zorro III got just the one, though the Level II chips did at least finally implement the full set of Zorro III features, with only one known bug left..." "There have been some Zorro III bugs, but that's not a gigantic surprise. And in fact, one of the problems with Zorro III leads to this -- it's an asynchronous bus protocol. For those of you who haven't designed custom chips, back in 1989 when I did the first Fat Buster I prototype, there was no simulation technology available to properly exercise such a design. When I started the Fast Buster II design in 1990, there was no simulation technology available for this. Today, there still isn't. I wrote about a third of the software used to verify the Buster designs, but some of the problems really could only be found in practice." "The basic architecture is all wrong, but that's actually intentional -- Zorro III wasn't designed to be a great match to the '030, or to Commodore's 1989 gate array technology." "Don't get me wrong, I like what I did in Zorro III. There were plenty of good reasons to build things that way, in those times. The whole of the PC industry came up with EISA, which was very much their answer to Zorro III. And in as much as the boys at Amiga, Los Gatos, gave me a better foundation to work from with Zorro I, I made Zorro III better than The Industry made EISA."

But Dave Haynie was totally incompetent, not even up to the level a 'normal' engineer, right?

@bhabbott

Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: You can see it with the ECS version used for Amiga 3000, which had a 16 bit bus despite the system being 32-bit and, what's even worse, the chipset was still castrated forcing the faster CPUs to wait for 4 cycles to access the Chip mem, even if it was possible doing it in two cycles. It could have been DOUBLE only if we had some normal (not even good are required here) engineers that allowed the processor to access Chip mem every two cycles instead of every four. So Commodore's engineers were below 'normal'? Right.

Right. See below.
Quote:

Please explain how this double speed access would work.

It's clearly reported above. I copy & paste for YOUR convenience:

the chipset was still castrated forcing the faster CPUs to wait for 4 cycles to access the Chip mem, even if it was possible doing it in two cycles

Do you know how a 68000 worked? Here's how the Hardware Manual explains it:

The 68000 uses only the even-numbered memory access cycles. The 68000 spends about half of a complete processor instruction time doing internal operations and the other half accessing memory. Therefore, the allocation of alternate memory cycles to the 68000 makes it appear to the 68000 that it has the memory all of the time, and it will run at full speed.

This does NOT apply to 68020 processors, but still checking if the CPU requires memory access is done in the second (odd) slot / color clock.

There's a thread on EAB where Toni Wilen (AFAIR) better explains is.
Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: Not even counting the bugs of Zorro III which crippled the memory bandwidth. How long do you still want to defend those incompetents? Dave Haytnie: Zorro III vs PCI Quote: "Part of the problem with so many armchair quarterbacks in this business is that they don't know the Amiga as a SYSTEM. A system engineer understands both the software and hardware, because in fact, until you get to implementation details, they're the same exact thing. Hackers rarely learn this approach, it's even rare among professionals. In fact, at my first job, I was one of about 10 new hires at General Electric, the only people in the company even ALLOWED to touch both hardware and software at the same time." "Of course you didn't forget that I did Buster. What you seem to forget is that I did all of Zorro III -- the specification, the first controllers, the first cards. Of course the Buster chip was a problem, though no worse, really, than Intel's first crack at a PCI controller. Of course, they've had seven or so architectural generations to improve on the controller. Zorro III got just the one, though the Level II chips did at least finally implement the full set of Zorro III features, with only one known bug left..." "There have been some Zorro III bugs, but that's not a gigantic surprise. And in fact, one of the problems with Zorro III leads to this -- it's an asynchronous bus protocol. For those of you who haven't designed custom chips, back in 1989 when I did the first Fat Buster I prototype, there was no simulation technology available to properly exercise such a design. When I started the Fast Buster II design in 1990, there was no simulation technology available for this. Today, there still isn't. I wrote about a third of the software used to verify the Buster designs, but some of the problems really could only be found in practice." "The basic architecture is all wrong, but that's actually intentional -- Zorro III wasn't designed to be a great match to the '030, or to Commodore's 1989 gate array technology." "Don't get me wrong, I like what I did in Zorro III. There were plenty of good reasons to build things that way, in those times. The whole of the PC industry came up with EISA, which was very much their answer to Zorro III. And in as much as the boys at Amiga, Los Gatos, gave me a better foundation to work from with Zorro I, I made Zorro III better than The Industry made EISA." But Dave Haynie was totally incompetent, not even up to the level a 'normal' engineer, right?

And a liar (as per "Read my lips, no new chips"):
https://en.wikipedia.org/wiki/Extended_Industry_Standard_Architecture

Year created 1988; 36 years ago
Intel introduced their first EISA chipset (and also their first chipset in the modern sense of the word) as the 82350 in September 1989.
The first EISA computer announced was the HP Vectra 486 in October 1989.

So, EISA was a answer to Zorro III when the former was released BEFORE the latter, eh?

@bhabbott
Quote:

bhabbott wrote: @cdimauro But Dave Haynie was totally incompetent, not even up to the level a 'normal' engineer, right?

If you recall Commodore The Final Years by Brian Bagnall, Amiga's R&D budget is behind Apple and Atari. AMD's and Intel's R&D budgets are monstrous.

It's challenging to execute leading-edge R&D with a meager R&D budget.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB