@cdimauro

Quote:

But I can tell you that I consider complete idiots the engineers that have invented the above Slow-Mem: it's NOT usable as Chip-Mem NEITHER as Fast-Mem. It has ALL DISAVANTAGES and NO usage (despite as being a storage). That was the most idiotic thing on the Amiga land, since it severely crippled the games.

1989-era A500 Rev6A's ECS Agnus can address Slow RAM as Chip RAM at different address ranges e.g. copper pointer set to 0x090000.

You can have some kind of 1MB Chip RAM with 512KB+512KB configuration with an ECS Agnus chip.

For Amiga 500/2000 models from 1987 to 1988, Slow RAM design wasn't a good move.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@Gunnar

Quote:

Gunnar wrote: @Hammer I find it very hard to understand what your point is. With all your off-topic stuff and copy/paste from Wikipedia about PC hardware, I can not see what you actually try to say about Amiga.

It wasn't from Wikipedia.

My last post's point is about Commodore's management issues that affected the engineering results.

I provided other management examples with fewer fckups.

Last edited by Hammer on 05-Mar-2024 at 05:28 AM.
Last edited by Hammer on 05-Mar-2024 at 05:26 AM.
Last edited by Hammer on 05-Mar-2024 at 05:25 AM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@bhabbott

Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: @bhabbott https://en.wikipedia.org/wiki/UAE_(emulator) Not sure what what you are trying to say here. UAE was originally called the unusable Amiga emulator for good reason. It wasn't until 1997 that PCs became powerful enough to emulate an A500 in (close to) real time. Even after adding jit in 2000 it still struggled to emulate the AGA chipset properly.

What's the difference with a PC emulator running on Amiga? Have you ever used one?

With IBeM I was able to see how the Character Interface (CUI) was "drawing" the windows, menus, dialog boxes, etc... like a slideshow (Turbo Pascal 4.0 and so on). On my Amiga 2000.

Of course the situation improved, but this happens with ANY emulator on ANY platform.

@Hammer

Quote:

My last post's point is about Commodore's management issues that affected the engineering results. I provided other management examples with fewer fckups.

Let to sum up the positions:

Cesare: "Amiga not good because = Amiga engineers were un-creative and did a bad job"

Hammer: "Amiga not good because = Commodore Management was bad"

Gunnar: "Nothing is ever perfect. Hardware engineering is hard = Amiga was OK."

Are these our views?




Did I understand the argumentation right, that one core point in the AGA discussion is :
missing chunky support.

This is correct, AGA lacks chunky support.


For some games Chunky format is pretty beneficial.
I think one of the most succesfull games using Chunky was maybe DOOM.
DOOM came out late December 1993.
In 1994 DOOM was a big success.


The Amiga original chipset design work began in 1983.
In 1983 popular home-Computers had 16 KB Memory.
When Amiga development began the firsts Commodore C64 came out.



A like DOOM which needs several Mega-Byte Memory was
for sure difficult to foresee 11 years earlier in 1983.

Commodore made 2 iterations, improvements on the OCS chipset.

ECS = the A3000/600 chipset.
ECS added programming of the screen modes.
This means the chipset can switch between NTSC/PAL mode.

AGA = The A4000/1200 chipset
AGA added more colors (64 out of 4096) => (256 out of 16 Million)
AGA added wider sprites 16 = 32/64 pixel


I'm not sure exactly but I think the AGA development started in year 1989.
AGA came out in 1992.

My understanding is that the design goal of AGA was to
* play all existing Amiga games
* give features to make more nicer looking Amiga games


How did very succesfull Amiga games look like when work on AGA chipset started ?


* Turrican
* Shadow of the Beast


Is it not imaginable that the "goal" of AGA was
to run games like Turrican and to add features
that help to make more such "successful" games?


In my opinion AGA reaches this goal.
AGA can run pretty good such 2D games.


Did they forsee in 1989 when AGA design work started the immense success of DOOM in 1994 ?
Probably not.

@Gunnar

Quote:

Let to sum up the positions: Cesare: "Amiga not good because = Amiga engineers were un-creative and did a bad job" Hammer: "Amiga not good because = Commodore Management was bad" Gunnar: "Nothing is ever perfect. Hardware engineering is hard = Amiga was OK." Are these our views?

Some narratives don't target the real cause.

My POV factors in Dave Haynie's (engineer at Commodore International) and David Pleasance's (Commodore UK MD) statements within Commodore.


Quote:

Did I understand the argumentation right, that one core point in the AGA discussion is : missing chunky support. This is correct, AGA lacks chunky support. For some games Chunky format is pretty beneficial. I think one of the most succesfull games using Chunky was maybe DOOM. DOOM came out late December 1993. In 1994 DOOM was a big success. The Amiga original chipset design work began in 1983. In 1983 popular home-Computers had 16 KB Memory. When Amiga development began the firsts Commodore C64 came out.

Jay Miner cited a military flight simulator during the original Amiga's design. The original Amiga was aiming higher i.e. workstation-like power for the micro-computer and game console markets.

Both 1985-era IBM's PGC/PGA and Amiga OCS's 4096 color palette targeted NEC PC-98's 4096 color palette (1981 era μPD7220).

For Nintendo 64, the SGI workstation's influence on a game console wasn't the 1st. The SGI team members who worked on Nintendo 64 have formed a company called ArtX.

The same idea with SUN GX workstation graphics influences NVIDIA's NV1 gaming PC card.

The same idea with SGI workstation graphics influences 3DFX's Voodoo gaming PC card.

Following 3DO M2, the 3DO MX project was led by an ex-SGI engineer. SGI was bleeding engineers.

IBM 8514 was workstation graphics in 1987 and made low cost by the SVGA cloners.

The idea is to obtain workstation graphics tech and make it low-cost via the highly integrated transistor chip methods.


Quote:

A like DOOM which needs several Mega-Byte Memory was for sure difficult to foresee 11 years earlier in 1983. Commodore made 2 iterations, improvements on the OCS chipset. ECS = the A3000/600 chipset. ECS added programming of the screen modes. This means the chipset can switch between NTSC/PAL mode.

My A500 Rev6A's ECS Agnus can switch between NTSC and PAL modes.

Programmable screen modes were part of ECS Denise.

Quote:

AGA = The A4000/1200 chipset AGA added more colors (64 out of 4096) => (256 out of 16 Million) AGA added wider sprites 16 = 32/64 pixel

AGA = The A4000/1200/CD32(+Akiko) chipset.

Quote:

I'm not sure exactly but I think the AGA development started in year 1989. AGA came out in 1992. My understanding is that the design goal of AGA was to * play all existing Amiga games * give features to make more nicer looking Amiga games How did very succesfull Amiga games look like when work on AGA chipset started ?

Amiga Ranger R&D was abandoned sometime in 1987. The original Los Gatos Amiga team attempted to counter Sharp X68000 and Apple's Macintosh II.

Los Gatos Amiga team was made redundant.

Commodore Germany's Amiga 2000 design won the corporate politics.

Amiga Ranger has 128 colors (7-bit planes) from a 4096 color palette and up to 1024×1024 pixel displays via VRAM. DRAM improves over time.

Amiga Ranger's 2 MB Chip RAM spec and the programmable display specs were preserved in Amiga ECS.

A3000UX with a TIGA (chunky pixels) card was released in 1990.

C65 has 256 colors (8-bit planes) from the 4096 color palette chipset R&D on an A500 level 7MB/s bus and this was completed on December 1990. C65 was canceled in 1991 by Commodore's chairman Irving Gould.

According to Dave Haynie, AA3000+'s AGA was operational in Q1 1991. AA3000+ lacks A600's PC world's IDE controller. After delays with ECS-focused "A1000 Jr" adventures, AA3000+ evolved into A4000. Blame ex-IBM PC Jr management import.

AGA was forked from AAA e.g. 16 million color palette.

Last edited by Hammer on 05-Mar-2024 at 08:32 AM.
Last edited by Hammer on 05-Mar-2024 at 08:31 AM.
Last edited by Hammer on 05-Mar-2024 at 08:23 AM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@Hammer


If you want to discuss, can you please try to put more structure in your posts?

I think if you "explain" what your point is in plain english then we can discuss about it much easier than if you "machine gun" out endless lists of wikipedia snippets.


Thank you.


What exactly is your point?

@Hammer

Maybe people have different views on what "GOOD" is?


Some people think the fastest most expensive system is GOOD.
Ferrari = GOOD

Some people think the most affordable system is GOOD.
FORD FIESTA = GOOD

Some might think the best system are a middle system



I think here our opinion on Amiga differ.



In my opinion the Amiga had the following key attributes

1) 68000 CPU = very coder friendly.
The coder frinedleness allowed kids to write nice games and demos.

2) Amiga DMA based chipset.
This chipset gave the Amiga its gaming power.
In my opinion the chipset was very balanced.
You can make nice games with Amiga chipset but its not 100% only for games.
You can also use Workbench and Application (Dpaint/Word/Webbrowser) very good with it.

3) Affordable
The Amiga 500/1200 were pretty inexpensive.
So that every family could afford one.

4) On purpose made to use TV frequency.
So that kids could use an existing TV - not need to buy a monitor.
To keep entry point cost low.
This "idea" made Amiga very popular with TV stations later.




I think the Amiga was very good balanced.

* The decided on purpose to use low clock CPUs in A500/1200 to keep the System "cheap" so that everyone can buy an Amiga.

* The decided on purpose to sell the system with no extra Fast memory, again to keep the entry systems at low cost.



I think we have to understand what the Goal(s) of the Amiga were.

Different people in Commodore and Engineering might have had different goals.

* Commodore management wanted to sell Home-computers that every Kid could get for XMas.

* Some engineer maybe wanted to make high end machines ...

There was obviously a conflict of goals.



The X68000 was technically a powerful system - but is was very pricy and therefore not such a success as the Amiga.

The NeoGeo was very good tuned for 2D games, but totally useless as Computer and was very expensive. NeoGeo was a huge market/sales failure compared to Amiga.


I think Commodore Management wanted an entry level system,
that kids can use to write some homework with,
and which is able to make nice games ... that kids like to play.


Yes the Ranger idea with VMEM had good performance.
But VMEM was very expensive.

Commodore did not want to sell system with unoptanium prices.
Commodore management rather preferred a lower cost entry system.

Commodore did not want to be Ferrari or Porsche -
they wanted to be sell the GOLF computer -
that everyone could afford but that gave good value for the money.


What do you think?

Last edited by Gunnar on 05-Mar-2024 at 09:24 AM.

@Hammer

When I worked for IBM, I was part of the Team which designed and developed the hardware for building the world fastest Super-Computer. https://de.wikipedia.org/wiki/IBM_Roadrunner
Being part of building the world fastest Computer was very exciting from engineering point a lot of fun.

But this IBM super-computer was build only 1 time.


The Commodore C64 on the other hand was sold over 10 million times.

Who had more success?

Which System was better?




I think: Both had very different goals and both did do a good job to reach their goal.


For Commodore the goal was to sell millions of home computers.
Did they do this?
YES

@Gunnar

Quote:

In my opinion the Amiga had the following key attributes 1) 68000 CPU = very coder friendly. The coder frinedleness allowed kids to write nice games and demos.

The PC's graphics architecture allowed certain bedroom programmers to create Catacomb 3D (1991), Wolfenstein 3D (1992), and Doom (1993).

In development Ultima Underworld: The Stygian Abyss (released in 1992)'s texture-mapped 3D graphics inspired John Carmark's Catacomb 3D (1991). John Carmack and John Romero had seen the Ultima Underworld's 1990 CES demo.

68000 is good for 32-bit OS road maps and linear memory address programming models.

Several major Unix big-endian vendors used Motorola 68K CPU family before switching towards big-endian RISC CPUs e.g. HP PA-RISC, Sun Sparc, and Hitachi SuperH.

Unix vendors who used 68000/68010/68012 CPUs have custom MMUs. 68010's companion 68451 MMU chip was slow.

SUN-2 used 68010 and SUN's custom PMMU.

The X86 competition has Xenix 286 and Xenix 386. Intel guaranteed MMUs on 286 and 386 CPUs. Without a guaranteed MMU install base on a particular platform, the PMMU OS transition has a chicken vs egg problem.

80386's 1985 release is too late for the Amiga's Lorraine project which started around 1983.

Linux was created on a 386-based PC.

386 large install base allowed Windows NT 3.x and Windows 3.1 Enhanced Mode (virtual memory, Win32S/WinG) software development decisions.

68K's early advantage with Unix wasn't continued with guaranteed PMMU inclusion in 68K CPUs. Motorola allowed their full 32-bit 68K family to be solid with PMMU-less SKUs while Intel's 386 standard has the mandatory PMMU.

MacOS X's origins NeXTSTEP was on a NeXT Computer with 68030 CPU and 688882 FPU and was released in 1988.

Quote:

2) Amiga DMA based chipset. This chipset gave the Amiga its gaming power. In my opinion the chipset was very balanced. You can make nice games with Amiga chipset but its not 100% only for games. You can also use Workbench and Application (Dpaint/Word/Webbrowser) very good with it.

Los Gatos Amiga team wanted an upgraded Amiga chipset.

Amiga Ranger targeted 7-bit (128 colors) planes which is a relatively minor update to Amiga's 5-bit color register and 6-bit planes OCS. Ranger's major update is its Denise and Agnus with fast 2MB Chip VRAM.

Sometime after Amiga Ranger's cancellation, A3000's R&D phase had the "read my lips, no new chips" management directive. Amiga OCS/ECS is stuck with a 1985-era memory controller which impacts AGA's Alice.


Quote:

I think we have to understand what the Goal(s) of the Amiga were. Different people in Commodore and Engineering might have had different goals. * Commodore management wanted to sell Home-computers that every Kid could get for XMas. * Some engineer maybe wanted to make high end machines ... There was obviously a conflict of goals.

R&D should continue without releasing the product. Engineers need to prove themselves with working and ready-to-ship upgraded chipsets and reference motherboard designs.

"Only the Paranoid Survives" - Andrew S. Grove, Intel Corp.

DRAM improves over time e.g. ET4000AX's 32-bit 60 ns to 80 ns DRAM or fast 16-bit VRAM.

A3000 has a native 68020/68030 25 Mhz 32-bit memory controller with Ramsey chip and Commodore didn't allow graphics IP made on it.

Meanwhile, C65's 8-bit planes (256 colors) with a 4096 color palette targeted CGS's 1989-1990 process node capabilities. It's a relatively minor update to Amiga's 5-bit color register and 6-bit plane chipset.


Quote:

The X68000 was technically a powerful system - but is was very pricy and therefore not such a success as the Amiga. The NeoGeo was very good tuned for 2D games, but totally useless as Computer and was very expensive. NeoGeo was a huge market/sales failure compared to Amiga.

SNK Corporation's NeoGeo was released in 1990.
Neo Geo AES and the Neo Geo CD have sold 980,000 units.
One million Neo Geo MVS units have been shipped worldwide as of April 1997.
Neo Geo's unit sales are superior to Commodore's AGA (under a million units).

SNK Corporation went bust in 2001 and it has IP takeover battles.

Sega's 1988 released Genesis/Mega Drive covered the low-cost gaming segment until SNES's 1990 Japanese release.

NEC PC-98 has an 18 million install base and it was crushed by Microsoft Windows 95's standard PC 386 architecture.

NEC/Hudson Soft has PC Engine game consoles with 5.62 million (HuCard) and 2.02 million (CD-ROM+HuCard) install base. Nintendo's SNES and Sony PlayStation crushed NEC's game console adventure.

NEC is still alive. Konami Digital Entertainment merged with Hudson Soft.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@Hammer



Can you please try to put more logical structure in your posts?

Please try to make your point in plain English
then we can discuss about it much easier.

Please stop "machine gun" out endless bullet points made from random Wikipedia snippets.

Thank you!
What exactly is your point?

@Hammer


Hammer:
Quote:

Los Gatos Amiga team wanted an upgraded Amiga chipset.

Which is exactly what I said.


Gunnar:
Quote:

I think we have to understand what the goal(s) of the Amiga were. Different people in Commodore and Engineering might had different goals. * Commodore management wanted to sell Home-computers that every Kid could get for XMas. * Some engineer maybe wanted to make high end machines ... There was obviously a conflict of goals.

@Hammer

Quote:

Amiga OCS/ECS is stuck with a 1985-era memory controller which impacts AGA's Alice.

I'm not sure what you try to say?

Fact is that AGA did get a big memory controller upgrade.
AGA Alive has 4 times the memory bandwidth compared to OCE/ECS Agnes.

So what you say is not really true.

Last edited by Gunnar on 06-Mar-2024 at 07:23 AM.

@Hammer

Do you agree or disagree with my post?

Gunnar:
Quote:

The NeoGeo was very good tuned for 2D games, but totally useless as Computer and was very expensive. NeoGeo was a huge market/sales failure compared to Amiga.

Hamme:
Quote:

One million Neo Geo MVS units have been shipped worldwide as of April 1997.

What does the number show?

The number show that NeoGeo was produced and sold by SNK the same number of years as Commodore sold Amiga.

In the same amount of years SNK only reached about 15% of the sales of Commodore Amiga.
Was this not exactly my point?


Commodore goal was to produce entry level home computers that every family could afford to reach very good sales numbers.

Is this goal wrong?


I've started my Amiga experience with the Amiga 1000.
The Amiga 1000 was an excellent machine. But it was expensive.
It was much too expensive for every family to afford.
Commodore pushed to make it a lot cheaper.
The Amiga 500 was a lot cheaper.
Only with the much more affordable Amiga 500 - the Amiga became a real market success.


Engineers often like to do Rocket Science stuff.
They enjoy to work on high performance and expensive stuff.

But Commodores market segment was not "Ferrari".
Commodotes market segment was low end.
So Commodore Management tried to "tune" the Amiga for the low end segment.

As the Amiga was made very affordable - it became a sales success and many games were made for it.


Lets assume Amiga would not have been bought by Commodore but by IBM.
Then Amiga might had been got all the Fancy stuff like VMEM.
But Amigas would have been sold for maybe 10 times the price Commodore did ask for.
And Amiga would have been a niche product with very few users.

@Hammer

Quote:

Los Gatos Amiga team wanted an upgraded Amiga chipset. Amiga Ranger targeted 7-bit (128 colors) planes which is a relatively minor update to Amiga's 5-bit color register and 6-bit planes OCS. Ranger's major update is its Denise and Agnus with fast 2MB Chip VRAM. Sometime after Amiga Ranger's cancellation, A3000's R&D phase had the "read my lips, no new chips" management directive. Amiga OCS/ECS is stuck with a 1985-era memory controller which impacts AGA's Alic

I can understand you think this.
Only I have the feeling your opinion as based on some "misunderstandings".


Let me try to offer you my view...

I think many ideas of the development team were accepted by Commodore
and actually came into the New Amiga Chips.


Do you know that the ECS chipset has the registers to use the Ranger-VMEM?
ECS was actually planned with more new features than most think.
But the relative high price of the VMEM - killed this as product.


Some people look down on the Amiga 600.
Commodore planned for the A600 to use much smaller PCB and smaller form factor and the new SMD soldering - to be much less expensive.
Commodore wanted to sell the A600 for much less than the A500.
But the production cost of the early SMD was higher than planned - so it not worked as planned. Otherwise Commodore would maybe have flooded the world with A600 for very very low price.


Development work on AAA chips with TrueColor / Chunky / 16bit audio and all features you can dream of - was done. They were just not market ready.

The AGA chips as simplified solution were developed and came out.
And AGA does offer improvements over ECS/OCS.

The thinking that AGA did not improve the memory controller is wrong.
The truth is that ALICE (AGA) memory bandwidth is 4 times more than OCS/ECS.


Everyone can have different wishes and opinions.

The notion that Commodore Management was all idiots and did everything wrong is maybe a little to simple.


You have to see that Commodore tried to be the "Burger King" of restaurants.
They aimed to sell many cheap Burgers to millions of people.

I can fully understand that many engineers would have loved to be Chef in a 5 Star restaurant,
and only do fancy lopsters - instead doing burgers.
But Burger-King aka Commodore paid their salaries.

Last edited by Gunnar on 06-Mar-2024 at 08:11 AM.
Last edited by Gunnar on 06-Mar-2024 at 08:09 AM.

@Hammer

Quote:

80386's 1985 release is too late for the Amiga's Lorraine project which started around 1983.

No one would have wanted the 386.

Maybe you not know this but the intel 80386 is not software compatible to the Motorola 68K!


Please understand that the 68000 was selected on purpose for 2 very important reasons.


a) very easy to use in Hardware design.

b) Very nice and easy to program.
Easy to develop software!


The Amiga design does NOT need a fast CPU.
The Amiga design pulls its power from the DMA based AMIGA Chipset
and its super-power is the very coder friendliness of the 68K-CPU.



The Amiga was build around a excellent CPU (excellent to program) which was not expensive..

The 68020 was available as higher performance option for the Amiga.
For those with to much money 68020- CPU cards were available even at the A1000 time.

Last edited by Gunnar on 06-Mar-2024 at 10:28 AM.

@Gunnar

Of course you are right the chipset was good... for it was designed. At its time even far ahead any other platforms

but there were two limitations... it was good for 2D games but needed special developed games that make use of the hardware. Then you could do things with it that needed much more resources on other platforms- And it of course not supported 3D games, Doom as a famous example.

The main problem was most of the commercial software was not exclusive developed for amiga. In early days many games were developed for Atari ST and then 1:1 ported to Amiga without really using the better hardware. Later main platform was PC and then ported to Amiga. Again not really using the hardware that compensated the lower resources (processor and ram).

You cannot blame the chipset for it but that was the reality. At least from my memory.

@OlafS25

I agree with you in all points

@Gunnar

Quote:

Which is exactly what I said.

You stated


2) Amiga DMA based chipset.
This chipset gave the Amiga its gaming power.
In my opinion the chipset was very balanced.
You can make nice games with Amiga chipset but its not 100% only for games.
You can also use Workbench and Application (Dpaint/Word/Webbrowser) very good with it.

and

I think we have to understand what the goal(s) of the Amiga were.
Different people in Commodore and Engineering might had different goals.

* Commodore management wanted to sell Home-computers that every Kid could get for XMas.
* Some engineer maybe wanted to make high end machines ...

There was obviously a conflict of goals.


It's obvious there's conflict.

The core problem is Commodore management had a "read my lips, no new chips" status quo directive.

Commodore management repeated the same problem as C64 evolved into C128 i.e. recycling aging C64 game hardware with "business" high resolution with low colors.

When the Amiga faced an aging C64 game hardware situation, Commodore management couldn't buy another company for easy advanced tech injection.

Btw: 3DO has about 2 million installbase which beats Amiga AGA's installbase.

Last edited by Hammer on 06-Mar-2024 at 09:38 PM.
Last edited by Hammer on 06-Mar-2024 at 09:33 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@Gunnar

Quote:

Gunnar wrote: @Hammer Quote: Amiga OCS/ECS is stuck with a 1985-era memory controller which impacts AGA's Alice. I'm not sure what you try to say? Fact is that AGA did get a big memory controller upgrade. AGA Alive has 4 times the memory bandwidth compared to OCE/ECS Agnes. So what you say is not really true.

Wrong. Refer to Amiga 3000's 32-bit Chip RAM with 7MB/s CPU access which is similar to AGA's 7MB/s CPU access on its 32-bit Chip RAM.

The Alice is not efficient with the given 32-bit memory bandwidth.

Alice's Blitter is still 16-bit like Agnus Blitter. In certain fetch modes, Lisa has up to 28 MB/s. Lisa's sprite's color registers have low color limitations while being 4X wider. There are gotchas with higher fetch modes.

Alice was modified for 2-cycle Fast Page RAM.

32-bit 14 Mhz RAM's raw bandwidth is 56 MB/s and 68EC020 CPU (any fast 68K) only has about 7MB/s from it.

Since Alice's Blitter was stuck in 1985, there's a need for a fast CPU for Lisa's 256 colors (8-bit planes) modes.

With a fast CPU and Fast RAM, AGA's Lisa is satisfactory for 320x256 with 256 colors (8-bit planes) e.g. 50 hz is attenable.

Like $699 Amiga 500 in 1987, there's $699 hardware that was released in late 1993 and that's 3DO. 3DO reached about 2 million install base which more than doubles AGA install base's below 1 million. It's too bad that the 3DO is just a games console.

3DO's custom chip has quadrilateral 3D complexity similar to Sega Saturn's quadrilateral 3D (4-sided sprites with distortion engine) system.

3DO had a custom math co-processor for geometry (fixed point) with a weak 12 Mhz ARM60 command CPU (68030 level @ 25Mhz). The two custom chips and 16-bit DSP are clocked at 25 Mhz. 3DO is from Los Gatos Amiga's key engineers.

Doom on the 3DO made a poor showing and has negatively impacted its PR performance.

Meanwhile, Dave Haynie pushed for AGA to be paired with AT&T's low-cost DSP3210 @ 50 Mhz (13.3 MIPS and 25 MFLOPS FP32) on 32-bit Fast RAM. Unlike Apple's Quadra 605, 68LC040 @ 25 Mhz wasn't factored in. On paper, AT&T DSP3210's MFLOPS FP32 nearly rivals 68060 @ 50Mhz MFLOPS.

The Commodore Hombre team selected the correct triangle-based 3D system, targeted SGI's OpenGL, 100 Mhz PA-RISC 7150-like CPU and they ran out of time. Desktop Amiga Hombre has Hitachi's PA-RISC implementation and may have A1200 on a chip.

For 1992 to 1994, the Amiga 1200 should been at least 386DX-33 (68020 @ 28Mhz or 68030 @ 33Mhz) with a near ET4000 (chunky pixel) level solution. 40 to 50 Mhz 68030 would be needed for C2P overheads.

Last edited by Hammer on 07-Mar-2024 at 04:18 AM.
Last edited by Hammer on 06-Mar-2024 at 10:39 PM.
Last edited by Hammer on 06-Mar-2024 at 10:31 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@Gunnar

Quote:

No one would have wanted the 386.

That's IBM's speak during OS/2's development. Cloners have beaten IBM on the 386 PC release.

i386 instruction set and PC VGA (via SVGA clones) won the desktop war against the 68K instruction set and Amiga OCS/ECS/AGA.

X86 PC world's clone business model has logistics distribution superiority. X86's software and hardware clones sold the standard PC platform.

Quote:

Maybe you not know this but the intel 80386 is not software compatible to the Motorola 68K!

There was no 3rd party legacy software for the Amiga Lorraine project during 1983.

In terms of legacy software, moving from MOS 65xx CPUs into Motorola 68K or Acorn ARM is ground zero.

The Unix workstation world was using 68K before Amiga Corp or Jack Tramiel's Atari.
Apple Lisa (with 68000) was released in January 1983.

Commodore's 65xx CPU R&D road map wasn't competitive against Intel's x86 R&D map. Blame Jack Tramiel.

Jack Tramiel's Commodore's 16-bit in-house solution was C900 with Z8001 CPU and CSG/MOS 8563 graphic chipset and to be manufactured by Commodore Germany.


Quote:

Please understand that the 68000 was selected on purpose for 2 very important reasons. a) very easy to use in Hardware design. b) Very nice and easy to program. Easy to develop software!

68000's linear memory address programming model has an advantage.

Again, the Unix workstation world was using 68K before Amiga Lorraine.

Quote:

The Amiga design does NOT need a fast CPU. The Amiga design pulls its power from the DMA based AMIGA Chipset and its super-power is the very coder friendliness of the 68K-CPU.

At a certain point, Amiga accelerator hardware turned into a decelerator.

IF the Amiga doesn't need a fast CPU, degrade your AC68080.

Due to 68000 @ 7.1 Mhz not delivering 7.1 MIPS, Amiga's custom chip shifts the bottleneck towards OCS 16-bit memory bandwidth. You're looking at 3.5MB/s to 7MB/s hog mode which is about 3.5 MIPS to 7 MIPS equivalent. The custom chips can't exceed the allocated memory bandwidth. Amiga's custom chips weren't designed for 7.1 MIPS fixed point 3D, let alone 25 MIPS or 28 MIPS 3D math co-processors.

6LC8040 @ 25 Mhz with nearly 20 MIPS with local 32-bit Fast RAM can do a better composing job. Lisa can display the completed frame buffer after it's copied from Fast RAM.

AT&T DSP3210 @ 50Mhz can reach 25 MFLOPS FP32 on A3000's 25 Mhz 32 bit Fast RAM.

Quote:

The 68020 was available as higher performance option for the Amiga.

AmigaOS's Kickstart 1.3/Workbench 1.3.x was aware of 68020/68030 for official CPU-accelerated A2000 bundles.

Back to this topic, A1200's $599 USD price target in 1992 was about $480 USD in 1987.

A slightly higher-priced Atari Falcon had 16 Mhz 68030 and 56K DSP (24-bit integer, 16 MIPS) instead. Jack Tramiel did a typical Commodore bonehead price reduction trick, 16-bit bus coupled with 68030 and 56K DSP.

56K DSP's 16 MIPS (INT16) matches 16 bit @ 16 Mhz memory bus.

AGA's Alice is mostly 16-bit hence there's something common with Jack Tramiel's mindset.

Both Commodore twins' "next-generation" hardware sold less than 3DO's 2 million install base.

Due to 3DO's lack of desktop computer SKU, 3DO directly faced against Sony's PlayStation.


Quote:

For those with to much money 68020- CPU cards were available even at the A1000 time.

1985 wasn't the issue for this topic.

This topic has John Carmark's statements in mind.

Last edited by Hammer on 07-Mar-2024 at 06:04 AM.
Last edited by Hammer on 07-Mar-2024 at 05:58 AM.
Last edited by Hammer on 07-Mar-2024 at 05:49 AM.
Last edited by Hammer on 06-Mar-2024 at 11:36 PM.
Last edited by Hammer on 06-Mar-2024 at 11:33 PM.
Last edited by Hammer on 06-Mar-2024 at 11:30 PM.
Last edited by Hammer on 06-Mar-2024 at 11:29 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)