@cdimauro

Quote:

Exactly, and we already had coprocessors with a precise design for performing various tasks. The proposed DSP is completely different -> alien to the existing ecosystem.

"Fat Paula" was in the "works", but it was low on the priority list.

For the Amiga 2000, Commodore was busy competing against 3rd party add-on vendors and did not focus on core graphics improvements.

R&D resources are diverted from Amiga's core chipset R&D.

Furthermore, Henri Rubin has a weak leadership style.

DSP3210 is a quick fix for the running "out of time" problem.

"Commodore The Final Years" by Brian Bagnall will fill in the blanks with the "read my lips, no new chips" directive for the A3000 project.

For example,
Quote:

From "Commodore The Final Years" by Brian Bagnall, The Gail Problem Jeff Porter had laid the groundwork for the C65 marketing push, including a plan to attract a large number of launch titles. “That’s marketing 101 on how to make the C65 successful,� he says. “Get the third party software developers on your side. And how do you do that? By getting the people who work for Commodore on your side to talk to the third party developers.� Porter needed to attract some of the top C64 developers in the US over to the C65 platform. At the time there were many software houses who had made their name on the C64, including EA, Activision, Broderbund, Epyx, Origin, and Access Software. In the latter part of 1990, these companies started embracing the PC world as new video and sound cards made games more exciting. Games such as Wing Commander came out that turned the heads of video gamers. Paul Lassa was excited to see what software developers would create for his new machine. “We’ve got this dynamite product that's got these great features and everything but it's going to need this support community and especially the developers to develop software and games for it,� he explains.

For the US market, Commodore pushed C65 with US developers when PC VGA's Wing Commander's 1990 release was the nuke.

Paul Lassa is obsessed with his C65 machine and it had drawn company resources from the Amiga.

The second nuke is Mode X's publishing in July 1991 which gave baseline VGA optimizations for many PC game developers.

Last edited by Hammer on 23-Jul-2024 at 03:14 AM.
Last edited by Hammer on 23-Jul-2024 at 03:13 AM.
Last edited by Hammer on 23-Jul-2024 at 03:11 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: @cdimauro Only for the high-end. It's too bad $108 68EC040-25 wasn't useable for the Amiga. Apple's 68LC040-25 Macs reached $1000 USD range in 1993 which is not high end. Commodore's A3640 cost is about $400 which includes the full 68040-25. Due to weak partitioned design, end users couldn't assemble CD32's $399 and A3640 $400. 3rd party Amiga accelerator vendors don't have Commodore's economics of scale.

Those were for high-end systems. Commodore's primary market was low-end: it should have provided more value on this segment.
Quote:

Quote: @cdimauro Amiga's primary market was very different. And, what's worse, the Amiga chipset could have been extended / evolved WITHOUT requiring an alien and very expensive (compared to the chipset cost) component like such DSP. Ex-Amiga team 3DO's geometry math co-processor was CPU PIO driven i.e. CPU writes values into the geometry math co-processor's registers and the CPU polls the geometry math co-processor's completion state. This should sound similar to CPU writes into Akiko registers.

Unfortunately, yes: it worked like the crappy Akiko (which had no DMA for the C2P conversion: "only Amiga engineers made it possible!").
Quote:

This is a quick hack to obtain higher math power.

SNES has already shown it.
Quote:

Amiga Blitter and DSP3210 have setup costs.

Better to have a more powerful Blitter, since it already had some technology built to go beyond the simple 2D area blitting (read: drawing lines).

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: Exactly, and we already had coprocessors with a precise design for performing various tasks. The proposed DSP is completely different -> alien to the existing ecosystem. "Fat Paula" was in the "works", but it was low on the priority list.

I can imagine...
Quote:

For the Amiga 2000, Commodore was busy competing against 3rd party add-on vendors and did not focus on core graphics improvements. R&D resources are diverted from Amiga's core chipset R&D.

There was still development on the Amiga side, as Lew reported.
Quote:

Furthermore, Henri Rubin has a weak leadership style. DSP3210 is a quick fix for the running "out of time" problem.

A quick fix for their incompetence.
Quote:

"Commodore The Final Years" by Brian Bagnall will fill in the blanks with the "read my lips, no new chips" directive for the A3000 project.

This is the excuse which incompetent people have invented to auto-absolve themselves

In fact, new chips were developed...
Quote:

For example, Quote: From "Commodore The Final Years" by Brian Bagnall, The Gail Problem Jeff Porter had laid the groundwork for the C65 marketing push, including a plan to attract a large number of launch titles. “That’s marketing 101 on how to make the C65 successful,� he says. “Get the third party software developers on your side. And how do you do that? By getting the people who work for Commodore on your side to talk to the third party developers.� Porter needed to attract some of the top C64 developers in the US over to the C65 platform. At the time there were many software houses who had made their name on the C64, including EA, Activision, Broderbund, Epyx, Origin, and Access Software. In the latter part of 1990, these companies started embracing the PC world as new video and sound cards made games more exciting. Games such as Wing Commander came out that turned the heads of video gamers. Paul Lassa was excited to see what software developers would create for his new machine. “We’ve got this dynamite product that's got these great features and everything but it's going to need this support community and especially the developers to develop software and games for it,� he explains. For the US market, Commodore pushed C65 with US developers when PC VGA's Wing Commander's 1990 release was the nuke. Paul Lassa is obsessed with his C65 machine and it had drawn company resources from the Amiga. The second nuke is Mode X's publishing in July 1991 which gave baseline VGA optimizations for many PC game developers.

Yes, many other mistakes from management.

Which don't obliterated the ones from the engineers...

@kolla

Quote:

kolla wrote: Anyone interested in DSP could just get a Delfina card or whatnot…

Depends on what they were interested in using it for. The DSP in the Falcon is a bit better suited to general purpose use than just audio tasks.

_________________
Doing stupid things for fun...

@cdimauro

Quote:

In fact, new chips were developed...

ECS Denise is not new since has existed from late 1987. There was a delay with monochrome vs color hi-res Denise drama. This is Commodore's typical #metoo based on competitor's product release.

After Amiga Ranger R&D was canceled in mid-1986, the rest of 1986 and 1987 were wasted on this monochrome vs color hi-res Denise drama.

Quote:

From "Commodore The Final Years" by Brian Bagnall, On February 4, 1987, LSI head Ted Lenthe produced a development schedule with prototype samples expected in May and the first 1000 production units in July. He assigned Hi-Res Denise the chip number 8369. Commodore engineer Bob Raible would perform the layout for Hi-Res Denise, with assistance from Amiga engineers Glenn Keller and Mark Shieu. Engineer Victor Andrade became lead designer on another chip, dubbed Hi-Res Fat Agnus, which received chip number 8372. A new chip designer named Bill Gardei would provide simulation and testing support. (skip section) Chipset Crisis Commodore’s LSI group labored through 1987 to complete the Hi-Res chipset, Denise and Agnus. The team ended up blowing past the original schedule, which called for samples in May and the first 1000 production chips in July 1987. In fact, the first samples arrived in August, and by September they were still working out bugs. George Robbins had been put in charge of overseeing the new chipset, along with the LSI engineers. Robbins felt they had done a good job taking over from the Amiga designers, especially in light of integrating new features, such as something called “high speed shifting logic�. (skip section) It began to dawn on the engineers that the monochrome hi-res Denise chip developed by the Los Gatos engineers was not worth developing anymore. The Hi-Res chip design had begun when Unix workstations, Mac, and IBM all had few or no colors, and that meant the system was a business machine. By 1987, not even the business world wanted monochrome anymore. The playfield had changed too much. Bob Welland and George Robbins looked at all this and began to consider two new goals for the chipset. First, they wanted a quick modification to the hi-res Denise chip to allow 640 x 480 noninterlaced with color. And second, they wanted to output the video signal to multisync VGA monitors. (skip section) This view contrasted with Bob Welland and George Robbins, who wanted the designers to add four color registers to the existing mono 8369 Denise in order to produce a color Denise (plus four additional color registers to handle color in the sprites). Robbins in particular believed there was little risk to continuing development on the Hi-Res chipset, considering there were no immediate computer systems in design to take advantage of them, and nothing would be released until late 1988. (skip section) The new chip, tentatively called Color Hi-Res Denise 8373, began to take shape. The team expected to have full tapeout in two months, meaning samples could arrive as early as December

This is December 1987 for Color Hi-Res Denise 8373 .

The "Commodore The Final Years" book by Brian Bagnall is a recommended read for timelines and drama.

Quote:

Continue from "Commodore The Final Years" by Brian Bagnall, By November 23, the logic design for the new Color Hi-Res Denise was complete, and layout work began. It looked like the engineers might have samples before the Christmas shutdown of CSG. Ted Lenthe planned to work out the color bugs in the first version of 8373, followed by another revision to perfect the color output. He wanted this version ready by the March 1, 1988 Hanover show.

This is November 1987.

March 1988 for Rev 1 8373 ECS Denise.

Lesson: If your product R&D pace is based on competitor's product releases, it's too late.

Time was wasted on minor upgrade high-resolution Denise.


Super Buster, Fat Gary, Ramsey are the 32-bit version from their 16-bit counterparts.

Released in 1988, C= A2620's 020 accelerator card has 32-bit memory controller functions via multiple chips and uses ZIP memory modules like on the A3000.

C= A2620's multiple chips for the 32-bit memory controller function were combined into Ramsey.

Meanwhile, SVGA ET4000's R&D was after ET3000's December 1987 release.

Dave Haynie's "read my lips, no new chips" is correct since the A3000 is a collection of various A2000 improvements.

For SVGA camp:
Tseng Labs' goal for the ET4000 project is to cost reduce and improve upon IBM's VGA and 8514 standards.

ATI's goal for the Mach8 project is to cost reduce and improve upon IBM's VGA and 8514 standards.

Last edited by Hammer on 26-Jul-2024 at 08:25 AM.
Last edited by Hammer on 26-Jul-2024 at 08:09 AM.
Last edited by Hammer on 26-Jul-2024 at 08:02 AM.
Last edited by Hammer on 26-Jul-2024 at 07:57 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:

Those were for high-end systems. Commodore's primary market was low-end: it should have provided more value on this segment.

$1000 USD in 1993 is about near mid-range.

$699 in 1987 is $ 888.68 in 1993.

$1000 USD is about $1500 AUD.

Reference
https://www.minneapolisfed.org/about-us/monetary-policy/inflation-calculator


Quote:

Better to have a more powerful Blitter, since it already had some technology built to go beyond the simple 2D area blitting (read: drawing lines).

The problem is backward compatibility with timing issues. AAA is only register-compatible

Apollo-Core has "turtle mode".

Extra effort was invested in Emu68's "turtle mode" which just large CPU power increase and unchanged Amiga chipset. A large CPU power increase can cause graphics corruption in backwards compatibility.

PC world has feature freeze the VGA standard while moving forward with RTG capable Windows 2D acceleration and VBE.

_________________
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: In fact, new chips were developed... ECS Denise is not new since has existed from late 1987. There was a delay with monochrome vs color hi-res Denise drama. This is Commodore's typical #metoo based on competitor's product release. After Amiga Ranger R&D was canceled in mid-1986, the rest of 1986 and 1987 were wasted on this monochrome vs color hi-res Denise drama. Quote: From "Commodore The Final Years" by Brian Bagnall, On February 4, 1987, LSI head Ted Lenthe produced a development schedule with prototype samples expected in May and the first 1000 production units in July. He assigned Hi-Res Denise the chip number 8369. Commodore engineer Bob Raible would perform the layout for Hi-Res Denise, with assistance from Amiga engineers Glenn Keller and Mark Shieu. Engineer Victor Andrade became lead designer on another chip, dubbed Hi-Res Fat Agnus, which received chip number 8372. A new chip designer named Bill Gardei would provide simulation and testing support. (skip section) Chipset Crisis Commodore’s LSI group labored through 1987 to complete the Hi-Res chipset, Denise and Agnus. The team ended up blowing past the original schedule, which called for samples in May and the first 1000 production chips in July 1987. In fact, the first samples arrived in August, and by September they were still working out bugs. George Robbins had been put in charge of overseeing the new chipset, along with the LSI engineers. Robbins felt they had done a good job taking over from the Amiga designers, especially in light of integrating new features, such as something called “high speed shifting logic�. (skip section) It began to dawn on the engineers that the monochrome hi-res Denise chip developed by the Los Gatos engineers was not worth developing anymore. The Hi-Res chip design had begun when Unix workstations, Mac, and IBM all had few or no colors, and that meant the system was a business machine. By 1987, not even the business world wanted monochrome anymore. The playfield had changed too much. Bob Welland and George Robbins looked at all this and began to consider two new goals for the chipset. First, they wanted a quick modification to the hi-res Denise chip to allow 640 x 480 noninterlaced with color. And second, they wanted to output the video signal to multisync VGA monitors. (skip section) This view contrasted with Bob Welland and George Robbins, who wanted the designers to add four color registers to the existing mono 8369 Denise in order to produce a color Denise (plus four additional color registers to handle color in the sprites). Robbins in particular believed there was little risk to continuing development on the Hi-Res chipset, considering there were no immediate computer systems in design to take advantage of them, and nothing would be released until late 1988. (skip section) The new chip, tentatively called Color Hi-Res Denise 8373, began to take shape. The team expected to have full tapeout in two months, meaning samples could arrive as early as December This is December 1987 for Color Hi-Res Denise 8373 . The "Commodore The Final Years" book by Brian Bagnall is a recommended read for timelines and drama. Quote: Continue from "Commodore The Final Years" by Brian Bagnall, By November 23, the logic design for the new Color Hi-Res Denise was complete, and layout work began. It looked like the engineers might have samples before the Christmas shutdown of CSG. Ted Lenthe planned to work out the color bugs in the first version of 8373, followed by another revision to perfect the color output. He wanted this version ready by the March 1, 1988 Hanover show. This is November 1987. March 1988 for Rev 1 8373 ECS Denise.

That wasn't the final ECS Design. The above is talking about supporting VGA mode, and nothing else.

ECS Design (I'm talking about the one introduced with the Amiga 3000) had programmable video modes support (like the Motorola 6845 of... 1980!) AND Super-HighRes (1280 horizontal) support.
Quote:

Lesson: If your product R&D pace is based on competitor's product releases, it's too late. Time was wasted on minor upgrade high-resolution Denise.

The monochrome Denise should have been the infamous Ultra-Highred mode with a single monochrome bitplane to be shown at something like 1024x1024 pixels (and an UHRes sprite as well). A completely dumb decision which have wasted precious resources on the chip, instead of introducing some more useful features.
Quote:

Super Buster, Fat Gary, Ramsey are the 32-bit version from their 16-bit counterparts. Released in 1988, C= A2620's 020 accelerator card has 32-bit memory controller functions via multiple chips and uses ZIP memory modules like on the A3000. C= A2620's multiple chips for the 32-bit memory controller function were combined into Ramsey.

Amiga's memory controllers and chipset's memory controllers are completely different things, as I've already reported in the past.

It's correct, however, that the chipset should have been extended to a 32-bit bus both externally AND internally (another great failure of Commodore engineers).
Quote:

Meanwhile, SVGA ET4000's R&D was after ET3000's December 1987 release.

Yup. I was expecting a bit more on 1987 from the new Amigas. I'll talk about this on another article of the series (first I want to complete the list of improvements).
Quote:

Dave Haynie's "read my lips, no new chips" is correct since the A3000 is a collection of various A2000 improvements.

It isn't, because final ECS Denice had more features introduced. And it's a new chip.

Plus... you continue to "forget" Amber its very expensive memory chips.
Quote:

For SVGA camp: Tseng Labs' goal for the ET4000 project is to cost reduce and improve upon IBM's VGA and 8514 standards. ATI's goal for the Mach8 project is to cost reduce and improve upon IBM's VGA and 8514 standards.

But those were much more expensive compared to the Amiga chipset. That's something which needs to be factored as well.
Quote:

Hammer wrote: @cdimauro Quote: Those were for high-end systems. Commodore's primary market was low-end: it should have provided more value on this segment. $1000 USD in 1993 is about near mid-range. $699 in 1987 is $ 888.68 in 1993. $1000 USD is about $1500 AUD. Reference https://www.minneapolisfed.org/about-us/monetary-policy/inflation-calculator

The Amiga 1200 and CD32 had proper costs for their respective market segments: you don't needed to spend much more.

E.g.: Nintendo's consoles always had a $199 introductory price and she never sold her console at loss.

You cannot compete at all with a $1000 price for the base Amiga system...
Quote:

Quote: Better to have a more powerful Blitter, since it already had some technology built to go beyond the simple 2D area blitting (read: drawing lines). The problem is backward compatibility with timing issues.

It can be easily addressed. You can see an example on my older article which talked about the 14Mhz chipset / system, and you'll see soon another example on my next article which I'm going to publish.
Quote:

AAA is only register-compatible

Which was GOOD!
Quote:

PC world has feature freeze the VGA standard while moving forward with RTG capable Windows 2D acceleration and VBE.

Well, VGA haven't defined strict timings since the beginning. And you know it very well, since accessing its VRAM from the CPU was quite slow, whereas competitors proposed much faster clones from this PoV.

@Karlos

Quote:

I don't know if Arnie is just a register compatible implementation that does it all digitally into a linear DAC, but I can't see an FPGA implementation having any of the analogue side behaviours. A "programmed like Paula but sounds like any modern soundcard" isn't an exciting prospect to me. Id much rather have it the other way around: a MIDI sampler with all the special characteristics of Paula's unique sound.

I looked it up recently. They've added new registers. Just like for RTG/VGA. I would have expected it to be layered over AUD0-AUD3. There's at least a couple of free register slots they could have used for extra flags and bank switches. Ok bank switching looked bad on AGA, but I think it would neatly expand Paula here, Also, they called it SuperAGA, not SuperAAA or anything so I expect them to follow AGA standards set. It would have also meant tracker software could be hacked to use double or more tracks more easily. Possibly.

As to the output, Arnie isn't Paula Perfect (TM). I recall reading something about a 24 bit mixer. So it's like an average DAC. Given the FPGA is digital this makes sense and is practical. Will it be lifeless and detectable? Well, it wouldn't be a perfect replica, but for most people they wouldn't notice.

UAE has had this criticism that it doesn't render the audio correctly. I am unaware if UAE has an audio plugin that attempts to emulate Paula exact as possible. But it also is emulating analogue circuits as well. So, it needs to modulate some sine waves with PCM values and apply PWM for volume. This would also apply to a new Amiga audio. A DSP/DAC is a practical solution. We have to consider that Paula was a means to an end and providing four channels of real sounds in two track stereo was a simple way of achieving that. Given the time with AM/FM synths and 16 channels becoming common, playing real sounds with less channels looks like a simple but effective way of doing it. So a DAC replacing it is practical. However, aside from Paula "warmth" it does rely on the PCM for proper output. So a very small set of audio data, with samples of sine, must also generate a proper sine on the output without any digital choppiness.

@Hypex

Quote:

They've added new registers.

You just summed up the entire apollo core project.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@Hammer

Quote:

For the Amiga 2000, Commodore was busy competing against 3rd party add-on vendor

Aside from the A1500… what do you mean? The number of expansion cards for A2000 isn’t exactly long, and most of them seem to exist because of A2500UX configurations.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

Just going to say it again. Fully programmable, resonant lowpass filters, post DAC, per channel.

Trust me, that would have slayed everything and everyone in 1985. The reason the Emulator was so successful was that its sampler - which wasn't massively better than Paula for playback (the model 1 had a sample rate of about 27kHz, but with a form of companded delta encoding for better dynamics than linear 8 bit) routed the output through a traditional analogue filter stage familiar to subtractive synth nerds.

Even if the digital control for cutoff and resonance were only the same 6 bit precision as the volume, it would have made a big difference. Sample based synthesis on the cheap, would have been lapped up by musicians in 1985.

_________________
Doing stupid things for fun...

Quote:

@cdimauro That wasn't the final ECS Design. The above is talking about supporting VGA mode, and nothing else.

"Read my lips, no new chips" was in the context of Commodore's internal directive.

Quote:

From Continue from "Commodore The Final Years" by Brian Bagnall, By November 23, the logic design for the new Color Hi-Res Denise was complete, and layout work began. It looked like the engineers might have samples before the Christmas shutdown of CSG. Ted Lenthe planned to work out the color bugs in the first version of 8373, followed by another revision to perfect the color output. He wanted this version ready by the March 1, 1988 Hanover show. That was Phase I. Phase II would then incorporate the changes required for genlock. These features were especially important because genlock was the only major feature not available in other competing chipsets on the market.

ECS Denise 8373 Phase 2 (R2?) incorporates the changes required for genlock.

ECS Denise 8373R3 was available for the public. I have ECS Denise 8373R4.

For A3000 project known as A2001 is an extension from A2000.
Quote:

From Continue from "Commodore The Final Years" by Brian Bagnall, Porter expected to see a prototype Super A500 in January 1988. Robbins went to work designing the system while awaiting the prototype hi-res Amiga chipset. Commodore had released the A2000-CR in the same case Commodore Germany had used for the A2500 (later renamed A2000). Porter wanted something sleeker and cost reduced for the Amiga 2000. “We needed something that could satisfy the business marketplace, and I didn't want it being so big fat and ugly,� he explains. Porter assigned mechanical designer Herb Mosteller to work on a new case which he dubbed the A2001. He planned to place the new cost-reduced A2000 motherboard in the new case. After Dave Haynie completed the A2000-CR project, it was time for him to get to work on the next iteration of Commodore’s high-end line of Amigas. The new project, which the engineers dubbed the A3000, would also reside in the new A2001 case. And much like the Super A500, it would use the hi-res Amiga chipset, a Motorola 68020 chip running at 14.2 MHz (double the clock speed of the A2000), high density floppy disks storing up to 1760 KB, 2 MB of VRAM, and a SCSI interface.

The VRAM-enabled Amiga high-resolution chipset was canceled for the existing ECS with a 1985 era 260 ns read/write cycle DRAM.

1988-era VRAM had 40 ns serial (sequential) access mode e.g. IBM 8514.

Commodore sabotaged their VRAM project with a 2MB VRAM requirement and the PC world wasn't crazy enough for 2 MB of VRAM.

3DO has lessons from Super A500/A3000's 2MB VRAM debacle i.e. MADAM (Agnus counterpart) can access both 2MB FP DRAM and 1MB VRAM. Amiga Hombre's Nathaniel has access to both 32-bit system memory and 32-bit/64-bit VRAM. The solution is the hybrid design.

PSX has 2MB EDO system memory and 1 MB VRAM (or 1MB SGRAM in later models).

2MB VRAM as Chip RAM is crazy.

The interpretation of "read my lips, no new chips" statement must come from the source, not from an external source.

Last edited by Hammer on 29-Jul-2024 at 05:35 AM.
Last edited by Hammer on 29-Jul-2024 at 05:29 AM.
Last edited by Hammer on 29-Jul-2024 at 05:14 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: Quote: @cdimauro That wasn't the final ECS Design. The above is talking about supporting VGA mode, and nothing else. "Read my lips, no new chips" was in the context of Commodore's internal directive.

Which is proven false since new chips were developed...
Quote:

Quote: From Continue from "Commodore The Final Years" by Brian Bagnall, By November 23, the logic design for the new Color Hi-Res Denise was complete, and layout work began. It looked like the engineers might have samples before the Christmas shutdown of CSG. Ted Lenthe planned to work out the color bugs in the first version of 8373, followed by another revision to perfect the color output. He wanted this version ready by the March 1, 1988 Hanover show. That was Phase I. Phase II would then incorporate the changes required for genlock. These features were especially important because genlock was the only major feature not available in other competing chipsets on the market. ECS Denise 8373 Phase 2 (R2?) incorporates the changes required for genlock. ECS Denise 8373R3 was available for the public. I have ECS Denise 8373R4.

So, 4 chips (new revision -> new mask -> new chip. That's how it goes in the silicon industry), as you can see.

How many other chips were produced? And I'm not referring here only to Denise. So, overall.
Quote:

For A3000 project known as A2001 is an extension from A2000. Quote: From Continue from "Commodore The Final Years" by Brian Bagnall, Porter expected to see a prototype Super A500 in January 1988. Robbins went to work designing the system while awaiting the prototype hi-res Amiga chipset. Commodore had released the A2000-CR in the same case Commodore Germany had used for the A2500 (later renamed A2000). Porter wanted something sleeker and cost reduced for the Amiga 2000. “We needed something that could satisfy the business marketplace, and I didn't want it being so big fat and ugly,� he explains. Porter assigned mechanical designer Herb Mosteller to work on a new case which he dubbed the A2001. He planned to place the new cost-reduced A2000 motherboard in the new case. After Dave Haynie completed the A2000-CR project, it was time for him to get to work on the next iteration of Commodore’s high-end line of Amigas. The new project, which the engineers dubbed the A3000, would also reside in the new A2001 case. And much like the Super A500, it would use the hi-res Amiga chipset, a Motorola 68020 chip running at 14.2 MHz (double the clock speed of the A2000), high density floppy disks storing up to 1760 KB, 2 MB of VRAM, and a SCSI interface. The VRAM-enabled Amiga high-resolution chipset was canceled for the existing ECS.

OK, and this still proves that new chips were developed.
Quote:

The interpretation of "read my lips, no new chips" statement must come from the source, not from an external source.

You continue to trust the "internal source" despite the clear evidence that the statement is a lie, looking at the FACTs that even you've reported.

It's obvious that this sentence was stated to offload all responsibilities to the management, and avoiding taking the responsibilities for his incompetence.

EDIT: you've added other stuff after that I've already replied. You should avoid ADDING NEW THINGS to already written comments. If you've something more to write, then write a NEW COMMENT, please.

Last edited by cdimauro on 29-Jul-2024 at 05:33 AM.

@cdimauro

Quote:

Which is proven false since new chips were developed...

ECS wasn't "new" from Commodore's engineers' POV.

Bug fixing on existing chip design is not considered "new".

http://www.bambi-amiga.co.uk/amigahistory/mikesinz.html

Commodore engineer Mike Sinz's view point
Quote:

The A3000 was a new motherboard design with new bus designs and chips to support and integrate all of this. The A3000 had a number of radical departures from earlier designs *other than* the ECS chipset. I won't go into all of these here but they were significant and many leading edge. (Again, the chipset did not change so that part was not "new")

1989 released A500 rev6 PCB has built-in full ECS support which includes jumpers for 2MB Chip RAM configuration.

Quote:

So, 4 chips (new revision -> new mask -> new chip. That's how it goes in the silicon industry), as you can see.

Commodore engineers Mike Sinz and Dave Haynie did not consider ECS to be new.

Quote:

You continue to trust the "internal source" despite the clear evidence that the statement is a lie, looking at the FACTs that even you've reported.

Who are you? You're not from Commodore. Each company has its own culture.

Last edited by Hammer on 29-Jul-2024 at 05:45 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: Which is proven false since new chips were developed... ECS wasn't "new" from Commodore's engineers' POV. Bug fixing on existing chip design is not considered "new".

Again, bug fixing of chips required a set of new MASKS, which are EXPENSIVE to produce. And this brought to NEW chips.

If you don't add new things on such chips you're wasting a lot of money without taking advantage of this great opportunity.
Quote:

http://www.bambi-amiga.co.uk/amigahistory/mikesinz.html Commodore engineer Mike Sinz's view point Quote: The A3000 was a new motherboard design with new bus designs and chips to support and integrate all of this. The A3000 had a number of radical departures from earlier designs *other than* the ECS chipset. I won't go into all of these here but they were significant and many leading edge. (Again, the chipset did not change so that part was not "new") 1989 released A500 rev6 PCB has built-in full ECS support which includes jumpers for 2MB Chip RAM configuration.

Again #2, I don't care about motherboard revisions.

How many chip revisions were done? That's the most important thing on this context. New revision -> new masks (expensive!) -> new chips. It's very simple to understand.
Quote:

Quote: So, 4 chips (new revision -> new mask -> new chip. That's how it goes in the silicon industry), as you can see. Commodore engineers Mike Sinz and Dave Haynie did not consider ECS to be new.

Again #3: what they don't consider it is irrelevant. The reality and how the industry worked is how I've reported.

On top of that, what about Amber? Isn't it a new chip? How was it possible to produce it with this "read my lips, no new chips" directive? It magically appeared? I don't think so...
Quote:

Quote: You continue to trust the "internal source" despite the clear evidence that the statement is a lie, looking at the FACTs that even you've reported. Who are you? You're not from Commodore. Each company has its own culture.

And people which are lying to hide their complete failures...

You should have been grow enough to understand it.

In Italy we're used to say something which translates roughly like this:

"Innkeeper, how is the wine?"

and he replies:

"Good!"

What do you expect from the innkeeper, when you're asking about how is his wine? Obviously he says that his wine is good, right?

Here is the same. And there are plenty of FACTs which are proving that this engineer was lying to defend his incompetence.

@cdimauro

Quote:

It's obvious that this sentence was stated to offload all responsibilities to the management, and avoiding taking the responsibilities for his incompetence.

The management is the problem.

When you have similarly ranked engineers are heading in different directions.

Quote:

From Continue from "Commodore The Final Years" by Brian Bagnall, The A3000 hinged on the engineers creating a new chipset for the machine, but the engineers arguments continued since September with no one able to agree on a spec for the new chipset. At the time, Porter told Bucas and Rubin, “Can you say ‘can of worms’? … Welland and Hedley will still be arguing by February about the next video chips.� The problem was, there were at least three different proposals for a hi-res chipset by three different engineers. Bob Welland wanted to begin fresh with a new architecture. Hedley Davis wanted to revise the existing Agnus/Denise architecture. Similarly, George Robbins wanted to revise the Agnus/Denise architecture based on a 32-bit architecture. No one could agree. With this situation, it looked like the next generation chipset had every possibility of being bogged down for months. Porter had previously hoped to show a prototype of an Amiga 3000 at the Hanover show in March 1988, but now it looked like that timeline was overly optimistic.

Gould himself had occupied the position of CEO and President of Commodore International since he had terminated Thomas Rattigan’s contract.


Meanwhile, PC's SVGA cloners have IBM's VGA and 8514 standards to be cost-reduced and improved upon. IBM has established a strong 640x480p, double scan ~31 kHz and 256 colors use case since 1984 PGC.

_________________
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:

Again, bug fixing of chips required a set of new MASKS, which are EXPENSIVE to produce. And this brought to NEW chips. If you don't add new things on such chips you're wasting a lot of money without taking advantage of this great opportunity.

That's a scope creep.

ECS Denise 8373R1 Phase I has the fundamental 4-color double scan high-resolution mode.
ECS Denise 8373R2 Phase II has genlocking functions for ECS. OCS Denise already has genlocking functions.
ECS Denise 8373R3 1st public release.
ECS Denise 8373R4 revision.

Quote:

Again #2, I don't care about motherboard revisions.

A500 rev6A motherboard revisions align with the intended ECS release. You argue as a consumer instead of being inside Commodore.

I have both A500 rev5 and A500 rev6A. I prefer A500 rev6A due to inherent full ECS support without hacking the PCB.

Full ECS has been tested for A500 rev6A's maximum configuration supported by the jumper switches i.e. ECS Denise, ECS Agnus AB/B, and 2 MB Chip RAM.

https://forum.amiga.org/index.php?topic=55384.0
Different ECS Denise revisions on different Amiga models with different results.

My ECS Denise with "390433-02 8373R4PD 3691 24" works on my A500 Rev6A without complication. A500 Rev6A didn't have complications with using newer Kickstart ROMs.

My 1988-era A500 Rev5 needs extra mods.

Quote:

How many chip revisions were done? That's the most important thing on this context. New revision -> new masks (expensive!) -> new chips. It's very simple to understand.

The context is Commodore's, not from you which is outside of Commodore.

The A3000 is nothing more than an A2000 with the A2091 SCSI, A2320 Amber, and A2630 boards built in. Both A2000 with A2630 and A3000 still have 16-bit Denise.

-----------------
The Pentium FDIV bug needs corrective design as a new stepping e.g. the FDIV bug affects the 60 and 66 MHz Pentium P5 in stepping levels prior to D1, and the 75, 90, and 100 MHz Pentium P54C in steppings prior to B5. The 120 MHz P54C and P54CQS CPUs are unaffected.

Last edited by Hammer on 29-Jul-2024 at 07:11 AM.
Last edited by Hammer on 29-Jul-2024 at 06:56 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:

Again, bug fixing of chips required a set of new MASKS, which are EXPENSIVE to produce. [quote] Today masks are enormously expensive, $millions or more. Back then chips were made in a very different way so the masks themselves may not have been a big deal. Creating them was the complex bit. [quote]And this brought to NEW chips. If you don't add new things on such chips you're wasting a lot of money without taking advantage of this great opportunity.

A bug fix won't be that big a change so wont be that expensive to implement.

An enhancement would have been a different story. That involves creating the design, testing it in software and hardware, and finally putting it into the chip design.
That could mean changing the entire layout of the chip, which back then was a major piece of work. Adding even small things would have involved months of work and cost a lot of money.

It's much better to make big changes all at once.

_________________
Whyzzat?

@you will never guess who

Quote:

The Pentium FDIV bug needs corrective design as a new stepping e.g. the FDIV bug affects the 60 and 66 MHz Pentium P5 in stepping levels prior to D1, and the 75, 90, and 100 MHz Pentium P54C in steppings prior to B5. The 120 MHz P54C and P54CQS CPUs are unaffected.

On the subject of how the Amiga audio could be better, this insightful pearl of wisdom could not be more relavent.

No wait, that other thing. Irrelevant.

Anyway, in case you guys didn't hear it.... Filters. That's what we lacked.

I've had the pleasure of playing Paula through an external filter, controlled via MIDI, and while the result is technically paraphonic (4 voices, one filter), the result is massive, especially when using all 4 channels like a single stacked oscillator. A fatter sound has never been heard.

_________________
Doing stupid things for fun...

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: It's obvious that this sentence was stated to offload all responsibilities to the management, and avoiding taking the responsibilities for his incompetence. The management is the problem. When you have similarly ranked engineers are heading in different directions.

And you should know that engineers aren't there left alone: there's always a TECHNICAL lead that they belong to and which decides which direction to take.
Quote:

Quote: From Continue from "Commodore The Final Years" by Brian Bagnall, The A3000 hinged on the engineers creating a new chipset for the machine, but the engineers arguments continued since September with no one able to agree on a spec for the new chipset. At the time, Porter told Bucas and Rubin, “Can you say ‘can of worms’? … Welland and Hedley will still be arguing by February about the next video chips.� The problem was, there were at least three different proposals for a hi-res chipset by three different engineers. Bob Welland wanted to begin fresh with a new architecture. Hedley Davis wanted to revise the existing Agnus/Denise architecture. Similarly, George Robbins wanted to revise the Agnus/Denise architecture based on a 32-bit architecture. No one could agree. With this situation, it looked like the next generation chipset had every possibility of being bogged down for months. Porter had previously hoped to show a prototype of an Amiga 3000 at the Hanover show in March 1988, but now it looked like that timeline was overly optimistic.

As you can see yourself (I've highlighted some parts), problems were coming from engineers: they didn't know what to do.

However, they were paid every single month, even when doing nothing concretely. Eggebrecht told something about that...
Quote:

Hammer wrote: @cdimauro Quote: Again, bug fixing of chips required a set of new MASKS, which are EXPENSIVE to produce. And this brought to NEW chips. If you don't add new things on such chips you're wasting a lot of money without taking advantage of this great opportunity. That's a scope creep. ECS Denise 8373R1 Phase I has the fundamental 4-color double scan high-resolution mode. ECS Denise 8373R2 Phase II has genlocking functions for ECS. OCS Denise already has genlocking functions. ECS Denise 8373R3 1st public release. ECS Denise 8373R4 revision.

You see yourself: NEW revisions -> NEW chips...
Quote:

Quote: Again #2, I don't care about motherboard revisions. A500 rev6A motherboard revisions align with the intended ECS release. You argue as a consumer instead of being inside Commodore. I have both A500 rev5 and A500 rev6A. I prefer A500 rev6A due to inherent full ECS support without hacking the PCB. Full ECS has been tested for A500 rev6A's maximum configuration supported by the jumper switches i.e. ECS Denise, ECS Agnus AB/B, and 2 MB Chip RAM. https://forum.amiga.org/index.php?topic=55384.0 Different ECS Denise revisions on different Amiga models with different results. My ECS Denise with "390433-02 8373R4PD 3691 24" works on my A500 Rev6A without complication. A500 Rev6A didn't have complications with using newer Kickstart ROMs. My 1988-era A500 Rev5 needs extra mods.

Again, the important thing is about the NEW chips: that was the discussion about.

Of course, if there are changes to the chips then motherboard changes might be required, but that's another thing (not part of the discussion).
Quote:

Quote: How many chip revisions were done? That's the most important thing on this context. New revision -> new masks (expensive!) -> new chips. It's very simple to understand. The context is Commodore's, not from you which is outside of Commodore.

I, even from the outside, can see FACTs like... NEW chips being developed. As even YOU proved.

In this context, as I've said, what's important is how many of them were developed. You reported at leat FOUR revisions for ECS, and I bet that there are other revisions, which may involve other chips.
Quote:

The A3000 is nothing more than an A2000 with the A2091 SCSI, A2320 Amber, and A2630 boards built in. Both A2000 with A2630 and A3000 still have 16-bit Denise.

Amber was developed for the Amiga 3000.

But this doesn't change the thing: it's a COMPLETELY (I hope that you like it now) NEW chip which was developed under the "read my lips, no new chips" directive, which is clearly a big lie used to justify their incompetence.


@minator

Quote:

minator wrote: @cdimauro Quote: Again, bug fixing of chips required a set of new MASKS, which are EXPENSIVE to produce. Today masks are enormously expensive, $millions or more.

They were expensive all the time. However, at least nowadays you have masks costing as low as 100k, which was simply a dream decades ago.
Quote:

Back then chips were made in a very different way so the masks themselves may not have been a big deal.

They were expensive: that's the point.
Quote:

Creating them was the complex bit.

Ehm... how many engineers worked on such masks?

And how many are working nowadays on similar chips ("scaled" to our complexity, of course)?
Quote:

Quote: And this brought to NEW chips. If you don't add new things on such chips you're wasting a lot of money without taking advantage of this great opportunity. A bug fix won't be that big a change so wont be that expensive to implement. An enhancement would have been a different story. That involves creating the design, testing it in software and hardware, and finally putting it into the chip design. That could mean changing the entire layout of the chip, which back then was a major piece of work. Adding even small things would have involved months of work and cost a lot of money.

Correct. However, ECS required a lot of changes on several important parts of the chip: they have even added an Ultra-hires bitmap + sprite which should have been interfaced to... VRAM!
Quote:

It's much better to make big changes all at once.

Sure.


@Karlos

Quote:

Karlos wrote: Anyway, in case you guys didn't hear it.... Filters. That's what we lacked. I've had the pleasure of playing Paula through an external filter, controlled via MIDI, and while the result is technically paraphonic (4 voices, one filter), the result is massive, especially when using all 4 channels like a single stacked oscillator. A fatter sound has never been heard.

I don't know of which filters are you talking about (sorry, music/audio isn't my domain).

Any example of what was required? How to use them as the registers level?