@kolla

Quote:

kolla wrote: @cdimauro I don’t see what number of audio channels have to do with a DSP?

Missed opportunities to improve the Amiga chipset – 1: audio


@vox

Quote:

vox wrote: @cdimauro Surely chipset evolution beyond AGA would be great (mythical AAA now Gunnars SAGA) but was not possible that quick.

They (engineers) started discussing the project by end of 1987...
Quote:

So earlier push of AGA could make them live a bit more.

Who them? Commodore?
Quote:

Looking at Falcon DSP plus 030 and faster RAM was quite a success making machine run 040 class software, making good audio use etc.

Falcon was a failure, not a success...
Quote:

So at that time, and with no evolved sound chip anywhere near, yes, looks like nice stopgap solution. Not as good as real Paula update.

Well, Paula NEVER changed.

When they decided to finally give it an upgrade, was too late (AAA has never seen the light, because Commodore broke).


@matthey

Quote:

matthey wrote: @all I primarily blame upper management while cdimauro places much of the blame with engineers.

No, I equally blame both.

I'm only tired of always seen the management being blamed and never the engineers. With the latters which since years are finger pointing only the management, miserably trying to hide their responsibilities, due to their gigantic ego and narcissism (here are the Amigans which should also be blamed, for having put them in the supreme chair and let them become like gods to be idolised).
Quote:

My perspective is that the system engineers are like general practitioner doctors. They could see problems first and addressed them with patched up solutions like a flicker fixer and DSP. They may not understand the big picture though. The chipset engineers are more like surgeons. They can go in and fix even major problems but they need very detailed information of what to fix. We don't have as much info from the chip engineers but the limited info makes it sound like CBM upper management restricted what they could do. A surgeon can have a bigger impact than a general practitioner but they need planning and very specific details on what to do. After the "no new chips" mandate, AGA was rushed and it is not a good idea to rush meticulous surgeons. They tried to save the patient with major changes but it was too little too late.

Unfortunately they also missed the head physician: the doctor which has the most expertise and the correct vision on how things should go.

And the "no new chips" was just a miserable excuse of some engineer to hide his responsibilities, as I've said, since new chips were developed.


@vox

Quote:

vox wrote: @matthey For what I have read and seen since CBM demise, my personal conclusion was that engineers had ideas how to go forward, but were always limited by menagment decisions.

Engineers lived in the land of confusion: they didn't know how to evolve the platform, and lost time in discussions.

There was no one which had the right technical leadership and vision to take the right decisions, planning and designing the new platform.

Once the original team left, who remained wasn't able to fill the gap and give a decent advance for our beloved computers.
Quote:

As an example, I liked missed opportunities with ECS (cdmauros analysis) but would not blame engs for lack of innovation in ECS.

Who should be blamed then? Have you seen what "enhancements" were brought with this chipset? Could you please tell me how much useful were the Ultra hires bitmap & sprite that they've inserted there?
Quote:

Seems managment needed quick and minor update, a drop in replacement, to go along with AmigaOS 2.0 rather then (r)evolution.

Well, they haven't added even the required minor fixes to the platform (e.g.: not even 8 bitplanes = 256 colours)...
Quote:

Similar goes for CBM65, A3000 AGA DSP etc. engs did prototypes, managment did not green light it.

Eh? Management was aware of all this.

@cdimauro

Quote:

cdimauro wrote: Well, Paula NEVER changed. When they decided to finally give it an upgrade, was too late (AAA has never seen the light, because Commodore broke).

And the things they were going to put in it (HD floppy data rate, 16-bit sound, audio input) were boring.

Meanwhile PC users were enjoying awesome OPL synth music that didn't get ported to the Amiga because it required too much programming effort and a ton of memory to store the converted samples.

Quote:

Quote: matthey wrote: @all I primarily blame upper management while cdimauro places much of the blame with engineers. No, I equally blame both.

I blame neither. They did pretty good under the circumstances. The things they did wrong, most Amiga fans would have done wrong too (or similar things). It would not have been better under Jack Tramiel (look at the C16 and ST to see why).

We are lucky to have gotten the Amiga at all, since it was only by luck that Commodore bought it, and only management's 'fault' that they ran with it (a sensible company would have just continued the C64 and PC lines). If it wasn't for Commodore's 'incompetence' the Amiga would have sunk without trace and none of us would know what we were missing.

This thread is just more boring 'what if' hindsight, unless we can actually realize some it to experience the 'what if'. That is why my next project will be a useful sound upgrade that Commodore could easily have been added at low cost and can still be added today without resorting to exotic modern hardware. I am going to make an OPL3 synth board that plugs into the parallel port or clock port.

I am also interested in other ideas for simple upgrades that can be cheaply realized using technology that was available to amateurs 'back in the day' (ie. off the shelf parts or repurposed PC components, not custom chips or FPGAs).

@cdimauro

Quote:

cdimauro wrote: @bhabbott: IBM compatibility doesn't matter. Other systems survived despite the PCs.

The PC-8201A certainly didn't. Nor did its siblings.

TRS-80 model 200
Quote:

Announced: 1984 The TRS-80 model 200 portable computer is an enhanced version of, and is mostly software compatible with, the earlier, and more popular, TRS-80 model 100 from 1983... The TRS-80 model 200 was not a top seller, nowhere near that of the earlier model 100. This is probably due to MS-DOS computers becoming all-the-rage, and the model 200 was not one of them.

The story goes that Marshal Smith cancelled the Commodore LCD in 1985 because Radio Shack's CEO told him there was no future in LCD portables - the implication being that Smith was suckered into dropping it by the competition. But if that was so then why was the Model 200 a poor seller? It had been introduced in 1984, so the CEO probably had a good idea of how sales were going.

In 1985 Radio Shack introduced their last non-IBM compatible portable, the Model 600. It had a similar size display to the Commodore LCD, plus an internal 3.5" floppy drive and a modem. It was a very poor seller - only 20,000 were made. So it seems that Radio Shack's CEO was telling the truth (at least from his perspective).

Commodore was coming in at the tail end of the market, thinking they would clean up because they had a better LCD display. But that's not how computer markets generally work. The most popular platform has the advantage, and few will jump to an incompatible platform just for some minor hardware improvement - especially when it's a maverick with unknown future.

The LCD portable was a shot in the dark, and Commodore didn't have the money to produce it and the Amiga - which had cost a lot more to develop than expected. Brian Bagnall's book "Commodore the Amiga Years" quotes Andy Finkel (who developed the LCD portable's software) as saying:-
Quote:

"we would have had to buy tooling for the case... and we just didn't have the several hundred thousand dollars... We said no, the Amiga needs that money to finish."

Was that a good decision? I know which machine I would rather have...

@cdimauro

Quote:

cdimauro wrote: @kolla Quote: kolla wrote: @cdimauro I don’t see what number of audio channels have to do with a DSP? Missed opportunities to improve the Amiga chipset – 1: audio

Right, and...? A DSP is so much more than just added channels and 16bit audio, they could have improved Paula your way and still have good use for a DSP for other things.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@bhabbott

Cheap and easy midi and sound digitizer would be a boom back then (esp. if were part of standard models) and fit your plan.

@kolla @cdmauro

Falcon was a failure but for reason Atari alalready abandoning home computing and gambling it all on Jaguar. On hw aspects its what Amiga could be on every level, even TOS was upgraded.
Looking at what small niche Falcon community archieved as well as few companies that supported it like Silmaris its great machine
On gaming level gfx wise is same or better then aga, on sound level its better. So those few titles surpass Amiga in Amiga Atari war :)
On sound pro software it beats Amiga and was long time studio standard, last batches being rebranded as sound studio equipment

DSP was used as coprocessor for math, for voice synthesis etc. and made games like Doom or Duke possible on lower clock 030, extending m68k abilitie.

On ECS - major use was vga producitivity mode and super hi res for certain use case, but cbm failed to supply standard vga connector to exploit it
Plus more chip ram, surely and no more slow chip ram
Deinterlacer as in a3000 should be part of every Amiga, as interlaced modes were eye killers and usable only on paper.

But what if ...

Managment did kill things that had market potential, like c65 or a3000 plus aga dsp, to a500+ and a600, which is a bad choice
A600 with numpad makes sense in bring ide pcmcia to amiga before aga, but a500 plus was waste in every aspect.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

vox Quote:

DSP was used as coprocessor for math, for voice synthesis etc. and made games like Doom or Duke possible on lower clock 030, extending m68k abilitie.

With a lower clocked 68030, it would be better to add a faster clocked 68030 than a DSP (or fast memory on the Amiga). High clocked 68030s may not have offered good value for mass production but at least mid-clocked 68030s would have offered better value than a DSP. The jump to a 68040 was a large cost difference and a 68030 with DSP would have been better performance than a 68040 in some cases but the DSP is much more difficult to program for games like Doom and Duke and DSPs usually had no data cache practically limiting them to streaming data (also one of the reasons why they are difficult to program). The cost of the 68040 was likely to drop faster than the cost of a 68030+DSP. If the 68040 was out on time, the 68040 price would have had more time to drop and would have been more competitive. Motorola should have considered a reduced cache version of the 68040 without FPU too. A DSP to offload dedicated tasks of the highest performance 68k CPU makes some sense as used by A/V Macs. Trying to use a DSP to improve general purpose performance leads to requiring a DSP which increases hardware costs and is best avoided. Codecs/libraries for dedicated tasks can use non-DSP algorithms for lower end lower cost hardware.

@bhabbott

Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: Well, Paula NEVER changed. When they decided to finally give it an upgrade, was too late (AAA has never seen the light, because Commodore broke). And the things they were going to put in it (HD floppy data rate, 16-bit sound, audio input) were boring.

Bruce, repeating like a parrot your rhetoric only shows that you're badly aging.

No, those were NOT boring stuff. Those are things that were missing from the platform, that everybody know it very well, and you mark as boring only because Commodore wasn't able to provide them, to protect the company.

That's always the case with you, blind Commodore zealot: you repeat the same mantra every time that the platform needs to be protected against the "enemies".
Quote:

Meanwhile PC users were enjoying awesome OPL synth music that didn't get ported to the Amiga because it required too much programming effort and a ton of memory to store the converted samples.

Sure. They were enjoying even this stuff: https://en.wikipedia.org/wiki/Roland_MT-32

Release date: 1987

The original MT-32 comes with a preset library of 128 synth and 30 rhythm sounds, playable on 8 melodic channels and one rhythm channel.
[...]
Sounds are created from up to 4 partials which can be combined in various ways (including ring modulation). With 32 partials available overall, polyphony depends on the tonal complexity of the music, and 8 to 32 notes can be played simultaneously.

right? Oh, poor PC users...
Quote:

Quote: No, I equally blame both. I blame neither. They did pretty good under the circumstances. The things they did wrong, most Amiga fans would have done wrong too (or similar things). It would not have been better under Jack Tramiel (look at the C16 and ST to see why). We are lucky to have gotten the Amiga at all, since it was only by luck that Commodore bought it, and only management's 'fault' that they ran with it (a sensible company would have just continued the C64 and PC lines). If it wasn't for Commodore's 'incompetence' the Amiga would have sunk without trace and none of us would know what we were missing.

Nevertheless, there were a lot of mistakes and there have been a lot of incompetent people. At the end, Commodore broke because of them.

That's, again, your rhetoric that you continue to report like a broken record, to blindly defend them.
Quote:

This thread is just more boring 'what if' hindsight, unless we can actually realize some it to experience the 'what if'. That is why my next project will be a useful sound upgrade that Commodore could easily have been added at low cost and can still be added today without resorting to exotic modern hardware. I am going to make an OPL3 synth board that plugs into the parallel port or clock port. I am also interested in other ideas for simple upgrades that can be cheaply realized using technology that was available to amateurs 'back in the day' (ie. off the shelf parts or repurposed PC components, not custom chips or FPGAs).

You like? Enjoy! But I don't care.
Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: @bhabbott: IBM compatibility doesn't matter. Other systems survived despite the PCs. The PC-8201A certainly didn't. Nor did its siblings. TRS-80 model 200 Quote: Announced: 1984 The TRS-80 model 200 portable computer is an enhanced version of, and is mostly software compatible with, the earlier, and more popular, TRS-80 model 100 from 1983... The TRS-80 model 200 was not a top seller, nowhere near that of the earlier model 100. This is probably due to MS-DOS computers becoming all-the-rage, and the model 200 was not one of them. The story goes that Marshal Smith cancelled the Commodore LCD in 1985 because Radio Shack's CEO told him there was no future in LCD portables - the implication being that Smith was suckered into dropping it by the competition. But if that was so then why was the Model 200 a poor seller? It had been introduced in 1984, so the CEO probably had a good idea of how sales were going. In 1985 Radio Shack introduced their last non-IBM compatible portable, the Model 600. It had a similar size display to the Commodore LCD, plus an internal 3.5" floppy drive and a modem. It was a very poor seller - only 20,000 were made. So it seems that Radio Shack's CEO was telling the truth (at least from his perspective). Commodore was coming in at the tail end of the market, thinking they would clean up because they had a better LCD display. But that's not how computer markets generally work. The most popular platform has the advantage, and few will jump to an incompatible platform just for some minor hardware improvement - especially when it's a maverick with unknown future. The LCD portable was a shot in the dark, and Commodore didn't have the money to produce it and the Amiga - which had cost a lot more to develop than expected. Brian Bagnall's book "Commodore the Amiga Years" quotes Andy Finkel (who developed the LCD portable's software) as saying:- Quote: "we would have had to buy tooling for the case... and we just didn't have the several hundred thousand dollars... We said no, the Amiga needs that money to finish." Was that a good decision? I know which machine I would rather have...

Irrelevant. And you're trying to distract from what was the original context in this part of the discussion.

Your point is always the same: since IBM started to sold the PCs, then everything else had no future.

No, dear zealot: even with the PCs taking the lead of the market, new, completely different, platforms were born and being commercialized.

And again, I'm not talking only of Apple (which, BTW, created the II GS after... the Amiga).

You should be old enough to know the history, but the problem with you is that you do NOT want to see them, because they contrast your ridiculous rhetoric that "PC destroying everything else" (to justify the Commodore failure, of course).

@kolla

Quote:

kolla wrote: @cdimauro Quote: cdimauro wrote: @kolla [quote] kolla wrote: @cdimauro Missed opportunities to improve the Amiga chipset – 1: audio Right, and...? A DSP is so much more than just added channels and 16bit audio, they could have improved Paula your way and still have good use for a DSP for other things.

The primary reason was improving the audio, as you can also read from the interviews of the technical guys (and, especially, their leader).

Yes, they can do more, but Matt already gave good reasons why they weren't that much important / interesting.

You can find other reasons on my fresh article, which shows how those $ could have been much better spent on enhancing/evolving the Amiga platform in the right direction, yet keeping the costs down.

Regarding this, you always have to take into account that adding such DSP would have brought another $40-$50 ($20-$30 for the DSP, plus $20 for its required codec) to the BOM, which mean much more to be charged to the final users.

If you compare it to the costs of the entire chipset, you can figure out yourself how much non-sense would have been this stupid decision.

@vox

Quote:

vox wrote: @kolla @cdmauro Falcon was a failure but for reason Atari alalready abandoning home computing and gambling it all on Jaguar. On hw aspects its what Amiga could be on every level, even TOS was upgraded. Looking at what small niche Falcon community archieved as well as few companies that supported it like Silmaris its great machine On gaming level gfx wise is same or better then aga, on sound level its better. So those few titles surpass Amiga in Amiga Atari war :) On sound pro software it beats Amiga and was long time studio standard, last batches being rebranded as sound studio equipment

Sure, and? We have to thank our great engineers for having played with the dolls instead of providing us a decent enhancement of the platform.
Quote:

DSP was used as coprocessor for math, for voice synthesis etc. and made games like Doom or Duke possible on lower clock 030, extending m68k abilitie.

See Matt's comment and my last article on such topic.
Quote:

On ECS - major use was vga producitivity mode and super hi res for certain use case, but cbm failed to supply standard vga connector to exploit it

Major use of ECS was... GAMES! Hence -> the above are completely irrelevant enhancements.
Quote:

Plus more chip ram, surely and no more slow chip ram

2MB was enough at the time.

Slow RAM was just the more evident proof of the ineptitude of Commodore's engineers.
Quote:

Deinterlacer as in a3000 should be part of every Amiga, as interlaced modes were eye killers and usable only on paper.

Well, no: it was a completely dumb solution. In fact, there was no such Amber chips on AGA, which have solved the issue in a much, much better way (the obvious one!).
Quote:

But what if ... Managment did kill things that had market potential, like c65 or a3000 plus aga dsp,

Those had NO market! At all!!!
Quote:

to a500+ and a600, which is a bad choice

A500+ was okish, but A600 is a complete failure: the Amiga required cost reduction from one end, but a better and much more competitive chipset from the other end.
Quote:

A600 with numpad makes sense in bring ide pcmcia to amiga before aga, but a500 plus was waste in every aspect.

Indeed.

@cdimauro

The “what if� question is what software would have made us of it. I know from back at univ that there were some mac models with DSP, and that old Photoshop made use if it. The NeXTStation had motorola 56k, though exactly what it was used for I’d have to look up.

(Meanwhile I worked in a neuro lab where we had 8 or more DSPs on ISA cards, stuck into industrial AT cases with 286 boards (later moved to equivalent pentium systems and PCI cards), collecting signals from sensors mounted in rat brains…)

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

Quote:

kolla wrote: @cdimauro The “what if� question is what software would have made us of it. I know from back at univ that there were some mac models with DSP, and that old Photoshop made use if it. The NeXTStation had motorola 56k, though exactly what it was used for I’d have to look up. (Meanwhile I worked in a neuro lab where we had 8 or more DSPs on ISA cards, stuck into industrial AT cases with 286 boards (later moved to equivalent pentium systems and PCI cards), collecting signals from sensors mounted in rat brains…)

Of course, you can make use of them, but... on which contexts/markets? The ones that you've shown were all far away from Commodore's mainstream market.

Amigas had already several expansion cards which included DSPs and even a network of the famous Transputers, so they were used and there was a market. But WITHOUT have been integrated on the official machines.

@cdimauro

I thought the plan was A3000 type computer, ie “workstation� like, not the gaming market. But “what if� it had come out on lower ends as well, can you think of something slightly uhm …mainstream that dsp could have been useful for mid-late 90s?

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@matthey

Thanks for the exponation, but generally EC versions were rare, still quite expensive and Moto did not have sense for budget CPUs like e.g. 386SX, 486SX or 486 pimped Cyrix clones style were.

In time when no standard home computer had 030 (I dont count TT and A3000, A4000-030) Falcon was a monster.

It was planed to go with 040 but was again too expensive so 030 was insered to make more budget computer.

When 040 became viable Atari already dropped the ball.

https://temlib.org/AtariForumWiki/index.php/Falcon_040

I understand DSP isn`t easy to grasp, but that was time when people did m68k ASM.
You can find numerous expamples how much was squeezed out of Falcon thanks to DSP, more then e.g. a3000. like Q2 engine, Halflife style engine etc.

Its not miracle, but added extended 16 bit sound plus some abilities. I have no doubt A3000 DSP AGA would make good or more widespread software used as compared to just same Paula on A1200 and A4000.

Now one is trying to make miracle of DSP, but as one solution existing as prototype, of the time, its better then what they came with to masses.

Also A3000 AGA had SCSI and CPU slot as in A3000, no longer in A4000 I believe. Too bad flicker fixer wasnt also carried on.
https://bigbookofamigahardware.com/bboah/product.aspx?id=23

@cdmauro

C65 had some market potential as still there were a lot of C64 fans in East Europe and 3rd world countries that haven`t moved on to Atari. A3000 AGA surely did have potential as it was prototype ready way earlier then A4000 or A1200.

ECS games just used extra chip RAM and RAM, otherwise they were OCS.
I tell you, major advances of ECS were more chip RAM and more productivity modes,
coupled with advancements of OS 2.0. And even that wasnt fully used as we had VGA compatibile modes without VGA connector (saying that as Amiga monitors were quite expensive) which would be good for productivity.


@All

Only real missed oportunity is giving Amiga chip acessable know hardware to evolve, and putting all hands on coding for it.

Imaginary:

a) Sorting Amiga lincesing mumbo jumbo
b) Giving Pi5 license to make and pack special Pi5 AIO keyboard model with AMiga name and Amiga sw bundle, where additional $20 per sold unit would go to Amiga IP owners and $10 to some controlled software development fund
c) Making AmigaOS 5 for it e,g.
- Taking OS 3.2.2 and AROS ARM as start codebase incl AROS multicore and x64 support, moving OS to threaded multicore x64 as possible, from current state.
- Taking best and free modern elements of MorphOS, Enhancer V46 OS, Amiga 600GS and PiStorm software ARM optimized and AmiKit XE pi for 1$ per sold model licenses
- Standardizing these so all future software development standards are known
d) Hardwarewise Amiga Pi 5
would also be able to run all Pi Software, AmigaOS would be just one preinstalled multiboot option
e) Built in 68k compatibility layer as in e.g. MacOS 8 early X from PPC to m68k and UAE ARM optimized, or special FPGA with AGA or SAGA chipset license to have backward compatibility to classic game library
d) Licensing bundling art pack, some apps and games as standard with OS from fund, or as free downloads to encourage both development and funding, or charging additional 1 to 10$ month donations from AOS 5 users for certain sw development - building some software model
e) Making ARM AmigaOS coding docs public and making easy cross compile tools to encourage ports from linux, android and other ARM OS`s to AmigaOS 5

With enough devotion, I believe such transition would be possible in shorter then making new OS 4 version or Tabor using known hardware and best OS elements :D

Last edited by vox on 14-Sep-2024 at 07:46 PM.
Last edited by vox on 14-Sep-2024 at 07:07 PM.
Last edited by vox on 14-Sep-2024 at 07:05 PM.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@kolla

It could do software modem eliminating need for external modem, which wasn`t cheap back then :D

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@vox

f) profit!!111

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@vox

I recall mpeg encoded audio (and video) files being rather popular at some point, some would argue they still are.

Last edited by kolla on 15-Sep-2024 at 12:34 AM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

vox Quote:

Thanks for the explanation, but generally EC versions were rare, still quite expensive and Moto did not have sense for budget CPUs like e.g. 386SX, 486SX or 486 pimped Cyrix clones style were.

Motorola sold mostly EC and LC 68k CPUs after the 68000. Apple was one of the few major buyers of MMU+FPU 68k CPUs and some low end 68040 Macs lacked FPUs. I don't like Motorola's CPU marketing. Personally, I think they would have been better off making the MMU standard on 68030, 68040 and 68060 as a MMU was relatively cheap to add (maybe 50,000 transistors for 68030 MMU). There was already the 68020 without MMU if a 68030 without MMU was wanted. They could have made up the small loss of not charging for a MMU by charging more for higher clock rated parts. The FPU was likely several times more expensive than a MMU and required more caches. My logic is to simplify the lineup offering half caches and no FPU or FPU with full caches. Motorola wasted too much time making EC and LC variants that had minimal production cost differences and wasted developer time bringing to market. Some of the 68060 EC and LC variants were full 68060s before they had time make EC and LC masks which was a high and seemingly only priority. For a similar developer effort, a 68060+ with double the caches likely could have been produced.

vox Quote:

In time when no standard home computer had 030 (I dont count TT and A3000, A4000-030) Falcon was a monster. It was planed to go with 040 but was again too expensive so 030 was inserted to make more budget computer. When 040 became viable Atari already dropped the ball. https://temlib.org/AtariForumWiki/index.php/Falcon_040 I understand DSP isn`t easy to grasp, but that was time when people did m68k ASM. You can find numerous examples how much was squeezed out of Falcon thanks to DSP, more then e.g. a3000. like Q2 engine, Halflife style engine etc. Its not miracle, but added extended 16 bit sound plus some abilities. I have no doubt A3000 DSP AGA would make good or more widespread software used as compared to just same Paula on A1200 and A4000. Now one is trying to make miracle of DSP, but as one solution existing as prototype, of the time, its better then what they came with to masses.

Much of the beauty of the 68k Amiga is the ease of programming. The 68k is one of the easiest to program architectures ever. Many developers used 68k assembly because it was so good and all that was needed. Many 68k consoles also banged the hardware using 68k assembly. I expect most 68k console games were written in 68k assembly. I doubt this was true of later generation MIPS, ARM or PPC consoles. DSPs are usually more difficult to program in assembly than RISC CPUs and compilers often generate poor code for them. DSPs are small low power cores with lots of compute performance which they provide by stripping out stall avoidance niceties and data caches. It is practically the opposite philosophy of 68k CPUs which avoid stalls and maximize cache efficiency. Adding a general purpose DSP ruins the ease of use of the system as programmers will be expected to use it to remain competitive. A general purpose DSP was not necessary and the popularity of using DSPs quickly faded with better performance general purpose CPUs. The 68060 could outperform many DSPs for DSP workloads yet it has all the advantages of a general purpose CPU and the famous ease of programming of the 68k.

vox Quote:

Also A3000 AGA had SCSI and CPU slot as in A3000, no longer in A4000 I believe. Too bad flicker fixer wasnt also carried on. https://bigbookofamigahardware.com/bboah/product.aspx?id=23

The Amiga 4000 dropped SCSI and the flicker fixer. The Amiga 4000T brought back SCSI. The Amiga 3000 flicker fixer was nice like a DSP would have been but they were expensive kludges for chipset handicaps.

@kolla

Quote:

kolla wrote: @cdimauro I thought the plan was A3000 type computer, ie “workstation� like, not the gaming market.

And that could be fine for such workstations, but I need to point out that you're embracing a component coming from outside, so you're linking yourself to it for the future.
Quote:

But “what if� it had come out on lower ends as well, can you think of something slightly uhm …mainstream that dsp could have been useful for mid-late 90s?

I've already expressed my opinion here and, more precisely, on my last article series. Matt has also reported other useful information.

In short: it's super expensive and alien to the Amiga ecosystem, requiring A LOT of work to squeeze the most from it, and to be used on specific scenarios which might not be so much mainstream.

Think about a simple thing: it's not able to access Chip RAM, but only system/Fast RAM, due to problems on bus mastering the memory. It means that you're forced to add Fast RAM to your system, increasing the costs. And, still, you cannot use it for things like packed/chunky to planar conversion, because it's not well suited for those kind of operations.

I invite to you think about how the Amiga was used to draw your conclusions:
- games;
- OS and applications.

Which kind of contributes it can give on both? That's an homework for you, kolla, and the DSP fans.

After that you can think about what could have better helped in those two macro areas.

I've already exposed my vision on the series: adding the mentioned features / evolutions to the chipset, which costed very little. And Matt has giving the second part of answer: using a better processor from Motorola.
Those two things would have greatly and positively impacted our beloved platform.

@vox

Quote:

vox wrote: I understand DSP isn`t easy to grasp, but that was time when people did m68k ASM. You can find numerous expamples how much was squeezed out of Falcon thanks to DSP, more then e.g. a3000. like Q2 engine, Halflife style engine etc. Its not miracle, but added extended 16 bit sound plus some abilities. I have no doubt A3000 DSP AGA would make good or more widespread software used as compared to just same Paula on A1200 and A4000.

I've great doubts, instead, because DSPs were very difficult to program, and to squeeze the most from them you were forced to go to assembly language as well, which was a bloodbath for developers.

I think that the main problem here is that people don't know how those DSPs worked (how it was their architectures, and how to correctly and usefully program it), and just take a look at raw numbers, which don't find a match with the reality.
Quote:

Now one is trying to make miracle of DSP, but as one solution existing as prototype, of the time, its better then what they came with to masses.

Well, NO: the right solution was to expand the existing chipset, which was much more useful for the overall platform and for all users.
Quote:

Also A3000 AGA had SCSI and CPU slot as in A3000, no longer in A4000 I believe.

SCSI was so much expensive, for... what benefits? For professionals? That's ok: give it to professionals.

However, IDE was much cheaper, so it was a much better for Commodore's primary market.
Quote:

Too bad flicker fixer wasnt also carried on.

It was so good: a dumb decision from the usual "great" engineers. And very expensive, BTW.

So, again: for professionals it might be a good stop gap (waiting a much better solution), but NOT for the mainstream market.
Quote:

@cdmauro C65 had some market potential as still there were a lot of C64 fans in East Europe and 3rd world countries that haven`t moved on to Atari. A3000 AGA surely did have potential as it was prototype ready way earlier then A4000 or A1200.

At the beginning of 90s it was another very stupid decision from Commodore's management, which had very little impact on the market.

The time of 8 bit systems was already well gone, and especially the limited C64 hardware. Even enhancing it in the proposed way, it was a complete disaster.

This project should have been done replacing the C128, and it would have made sense. But definitely not so many years after.

Plus, certainly not with those very stupid decisions from the engineers about how to enhance the C64 hardware.

Here it was The perfect storm: a very "good" combination of stupid decisions from both management and engineers...
Quote:

ECS games just used extra chip RAM and RAM, otherwise they were OCS.

HOW they used it? Just as a cache, or as a replacement for the Slow RAM.

Read: it was NOT very useful and definitely not a game changer for... games.
Quote:

I tell you, major advances of ECS were more chip RAM

Not used in the correct way, as I've said.
Quote:

and more productivity modes,

See above for that: good ONLY for professionals. NOT for the mainstream market.
Quote:

coupled with advancements of OS 2.0.

That's orthogonal to the platform.
Quote:

And even that wasnt fully used as we had VGA compatibile modes without VGA connector (saying that as Amiga monitors were quite expensive) which would be good for productivity.

Again, only good for professionals.
Quote:

@All Only real missed oportunity is giving Amiga chip acessable know hardware to evolve, and putting all hands on coding for it.

Indeed, and I've already mostly covered it (the last article is missing from my series).
Quote:

Imaginary: a) Sorting Amiga lincesing mumbo jumbo b) Giving Pi5 license to make and pack special Pi5 AIO keyboard model with AMiga name and Amiga sw bundle, where additional $20 per sold unit would go to Amiga IP owners and $10 to some controlled software development fund c) Making AmigaOS 5 for it e,g. - Taking OS 3.2.2 and AROS ARM as start codebase incl AROS multicore and x64 support, moving OS to threaded multicore x64 as possible, from current state. - Taking best and free modern elements of MorphOS, Enhancer V46 OS, Amiga 600GS and PiStorm software ARM optimized and AmiKit XE pi for 1$ per sold model licenses - Standardizing these so all future software development standards are known d) Hardwarewise Amiga Pi 5 would also be able to run all Pi Software, AmigaOS would be just one preinstalled multiboot option e) Built in 68k compatibility layer as in e.g. MacOS 8 early X from PPC to m68k and UAE ARM optimized, or special FPGA with AGA or SAGA chipset license to have backward compatibility to classic game library d) Licensing bundling art pack, some apps and games as standard with OS from fund, or as free downloads to encourage both development and funding, or charging additional 1 to 10$ month donations from AOS 5 users for certain sw development - building some software model e) Making ARM AmigaOS coding docs public and making easy cross compile tools to encourage ports from linux, android and other ARM OS`s to AmigaOS 5 With enough devotion, I believe such transition would be possible in shorter then making new OS 4 version or Tabor using known hardware and best OS elements :D

GOSH. I prefer to don't spend other words on that...

@vox

Quote:

vox wrote: @kolla It could do software modem eliminating need for external modem, which wasn`t cheap back then :D

That was the plan for the DSP as well.

However, how many people used a modem at the time?

You wanted to put a $40-$50 cost for a DSP (to the BOM. Which means much more to the final users!) that ALL customers had to pay, even when not using it.

Does it make sense? Clearly NO!


@kolla

Quote:

kolla wrote: @vox f) profit!!111

?!?
Quote:

kolla wrote: @vox I recall mpeg encoded audio (and video) files being rather popular at some point, some would argue they still are.

WHEN? WHO used them?

Again, the main problem here is understanding the market target. Specifically, being able to understand Commodore's mainstream market.