@ppcamiga1

Commodore Amigas biggest mistake was firing Thomas Rattigan, who turned around Amiga when it was heading toward bankruptcy in the 80's. He respected the engineers and could run the business better than anyone at that point.

This was a major contribution to where they ended up.

Poor management takes awhile to undue it, given Commodore's financial situation and the attitude on production and engineers, it was dead for awhile and just didn't know it yet.

You need money not to be bankrupt, to have money you need great leadership.

Quote:

Matt3k wrote: Commodore Amigas biggest mistake was firing Thomas Rattigan, ...

From Wikipedia:
Quote:

In 1986 Rattigan took over as CEO from Marshall F. Smith. He continued the downsizing his predecessor had initiated, cut unprofitable product lines and initiated the development of two more models of Commodore's flagship Amiga 1000 computer.[1] Despite turning the company from a US$237 million three-quarter loss to a $22 million profit in one quarter, Rattigan was fired by major shareholder Irving Gould, who temporarily took over as CEO.

Sounds good, but he is also responsible for taking away the Amiga from the Los Gatos crew who had designed the Amiga to Commodore's core group of engineers in West Chester, this was good for the A500 but not for the A2000, that became the high end model with low end performance.

In 1987 he was fired because Gould thought the creation of the A500/A2000 needed to much time, instead C= got Mehdi Ali as a new manager and had to pay 9Million USD to Rattigan as compensation for breaking the contract.

We don't know what Rattigan would have done after the A500/A2000 but Mehdi Ali did no good.

Last edited by OneTimer1 on 23-Dec-2023 at 06:05 PM.

@ppcamiga1

Quote:

ppcamiga1 wrote: @Hammer this video is really interesting. few second more it show how much time was wasted on c2p. if aga has chunky pixels doom wil run at least two times faster. and be playable and commodore will not bankrupt in 1994

SNES has chunky pixels via Mode 7 and Mode 7 Direct Color. For SNES Doom, it needs 16-bit RISC SuperFX2 @ 20 Mhz

Commodore would have rejected the SuperFX2 stop-gap just as Commodore CEO Ali rejected Commodore UK's upgraded CD32 push.

_________________
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:

This is why so many PC clones came with a crappy trident VGA card.

1. That's a flawed argument when the PC market is very large.

Are you claiming there were only 44,000 ET4000 sales in 1992?
Are you claiming there were only 170,000 ET4000 sales in 1993?

https://dosdays.co.uk/topics/Manufacturers/tseng_labs.php
By 1991, according to IDC, Tseng Labs held a 25% market share in the total VGA market.


From https://www.intel.fr/content/dam/doc/report/history-1994-annual-report.pdf
Intel reported the following
1. In 1994's fourth quarter, Pentium unit sales accounted for 23 percent of Intel's desktop processor volume.
2. Millions of Pentiums were shipped.
3. During Q4 1993 and 1994, a typical PC purchase was a computer featuring the Intel 486 chip.
4. Net 1994 revenue reached $11.5 billion.
5. Net 1993 revenue reached $8.7 billion.
6. Growing demand and production for Intel 486 resulted in a sharp decline in sales for Intel 386 from 1992 to 1993.
7. Sales of the Intel 486 family comprised the majority of Intel's revenue during 1992, 1993, and 1994.
8. Intel reached its 6 to 7 million Pentiums shipped goal during 1994. This is only 23 percent unit volume.

How about 28 million 486s for a crushing blow?

By the end of 1994, Intel's Pentium PC install base crushed the entire Amiga install base of 4 to 5 million units!

By the numbers, Intel's unified X86 PC platform is a monster compared to the Amiga i.e. it mirrored the USA superpower military might against the smaller German military during WW2.

Doom 1's and Doom 2's attachment rate with PC's install base may suck, but still 4 million units sold.

Your argument is absurd.

2. Trident 8900CL is pretty good for Doom.

On the next advert, TLAB for $181, this is Tseng LABS.

https://i.ibb.co/0QfMNRb/APC-Nov-1993-prices-ET4000.png
Tseng LABS ET4000 for $145 AUD.

Quote:

Oh yeah a 386DX-40 for only $1320 - with that crappy 512k VGA card whose brand name we won't mention, and a miserable 1MB of RAM.

Add RAM. It's listed.

On the next advert, Extra 1 MB SIMM for $55.

Quote:

No sound card.

Add a sound card. It's listed. I recycled my ISA sound card on my Pentium 150 build in 1996.

Quote:

No hard drive.

Add 3.5 inch HDD. Superior MB per dollar.

Quote:

No operating system. This is only half a PC! Add all the stuff you need to make it usable and

There's a reason for MS's online activation. The Amiga had its X-Copy sub-culture.

Quote:

now what does it cost?

A1200 wasn't a player in Australia's Xmas Q4 1992. Check Australian Amiga Review magazine.

Try 1993 instead.

From https://i.ibb.co/0QfMNRb/APC-Nov-1993-prices-ET4000.png

Sound Blaster 2 = $135.

In 1993, it was the 486SX-33 era in the $1500 AUD price range.

$1400 AUD is about $1000 USD.

Last edited by Hammer on 11-Apr-2024 at 11:42 AM.
Last edited by Hammer on 11-Apr-2024 at 11:39 AM.
Last edited by Hammer on 11-Apr-2024 at 11:33 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)

@Matt3k
Huge +1

Where is amigaworlds like button when you really need it.

_________________
A posting a day keeps the sanity away...

Irving Gould was singly the worst thing that happened at Commodore. From being their money man in the early days, he pillaged the company to line his pockets, regardless if it was making money or not. He couldn't see it as a flourishing place of creativity and where thousands of people relied on it to put on the table, he grabbed the money regardless. Senior execs would come and go and have varying levels of success but in the end, Gould cashed out in the most crass way you could imagine and the company had no chance of investing in the future.

_________________
Amiga A1200, 3.1 ROMs, Blizzard 1230 MKIV 64MB & FPU, 4GB DoM SSD, Workbench 3.1

@cdimauro

Quote:

That's only the video card. What about complete configuration of a gaming PC AND the price of an Amiga 500?

Does A500 have a fast 256-color frame buffer without lossy color compression known as HAM mode?

Fact: Commodore had good revenue levels in 1990 and 1991.


It would be a tricky situation for the hardware generation switch. The Amiga platform resembles a game console platform with a keyboard since the Amiga graphics chipset is not PC graphics card modular.


Quote:

SNES was sold in 1990, not 1992. So, same year of the ECS.

SNES was released in Japan's 1990 market.

From 1990 to 1993, Commodore's revenue core market is Western Europe with the strongest Amiga market in the UK and Germany.

The year 1992 was SNES's European release and direct competition with Commodore.

SNES has packed pixels via Mode 7 and Mode 7 Direct Color modes, but it needs 16-bit RISC SuperFX2 @ 20Mhz for the Doom port.

IF A500++++++ had AGA with packed pixels, it needs a fast 68030 or at least 68020 @ 33Mhz!

SNES started to build its install base in 1990 and its larger install base was an advantage in for 3rd party developers during 1992. AGA was ground zero in 1992 with a tiny install base.

A500++++++/A1200/CD32 with AAA chipset and 68EC020 @ 14 Mhz CPU wouldn't change the outcome.

Quote:

Again, useless PADDING, since the CONTEXT was 1990!

From 1990 to 1993, Commodore's revenue core market is Western Europe with the strongest Amiga market in the UK and West Germany.

Quote:

The CONTEXT was Low-End GAMING market on 1990.

There's one thing for certain, Commodore CEO Ali will reject something like stop-gap SuperFX2 just as Ali rejected Commodore UK's upgraded CD32 initiative.


https://archive.computerhistory.org/resources/access/text/2013/04/102723262-05-01-acc.pdf
Page 119 of 981

For 1992
68000-12 = $5.5
68EC020-16 PQFP = $16.06, it's $15 in 1993 Q1.
68EC020-25 PQFP = $19.99, it's $18 in 1993 Q1.


Quote:

LOL! Then kick out the Blitter from all Amigas, because it's useless without a fast frame buffer.

Your words. Amiga's 16-bit Blitter @ 3.5 Mhz is so 1985.

My position on this matter is a dual 16-bit Blitter. The legacy Amiga software will only see the 1st 16-bit Blitter.

Quote:

No, it's in your a$$ only because you're an AMD die hard blind fanatical, so you're still burning for this.

You can't handle the truth straight from the Xbox team.

Did you forget the Amiga camp was "Intel Outside"? Did you forget MorphOS 3.12 Beta was demonstrated on the "AMD64, Ryzen 5 3600" PC? https://www.youtube.com/watch?v=oAUZivdbn-Q

Quote:

Neither the CPU, which was a regular P3 one (only with half the L2 cache).

You can't handle the truth straight from the Xbox team.

Removing Bill Gates also removes the close Intel link e.g. Xbox 360 with IBM PPE.

Quote:

It remains the most important component compared to the CPU. It's so much important that even an AMD die hard blind fanatical like you is using nVidia GPUs instead the ones of your beloved AMD...

Your attempt to paint me as an "AMD die-hard" is debunked by my NVIDIA GPU selection.

I have no problems switching to Qualcomm Oryon as long it matches the gaming PC's use case.

Quote:

LOL: TOTAL NON-SENSE! The FPU does NOT use the graphic card's fast-frame buffer for its calculations! There's NOTHING that could help the CPU's FPU for processing its 3D calculations.

What happens when you use K7 Athlon XP 2200+ (1800Mhz) with an original IBM VGA?

Hint: it's a roll-up slide show. LOL

A reminder... GPU I/O architecture basics


This is applicable for both AMD's VEGA/RDNA1,2,3 and NVIDIA's Maxwell/Pascal/Turing/Ampere/ADA.

Compute Shader IO path was handled by the CPU. Modern GPU's compute shaders took over these roles. Compute Shader usually uses TMUs for the I/O.

Pixel Shader IO path is handled by raster i.e. ROPS. Pixel Shader role was handled by CPU before it was handed to the GPU's pixel shaders. ROPS is the good old fill rate frame buffer IO units.

In RTX and RX RDNA 2/RNDA 3, the raytracing core's (BVH, boxes, geometry, search engine problem) processing is handled by their I/O path which is separated from ROPS I/O path!

Certain concepts remained similar and they are hardware accelerated.


Quote:

And here comes back the famous Hammer's crystal ball... Another total non-sense, of course, since chips sold on 1990 were designed years before, and of course targeting 1990 and NOT 1993 or something like that.

I'm aware of this. Don't assume. For example, ET4000AX's 1989 release would have its R&D in the 1987 to 1988 time frame.

Quote:

You've absolutely no clue of how chips are designed and go to the market!

That's your assumption which makes you an ass.

Quote:

Again, you've no idea of what you talk about. Larrabee was a failure, yes, and it was replaced by the Xeon Phi: NOT Arc!

Wrong. Larrabee was a cGPU addon card product e.g. https://www.pcgamer.com/a-working-intel-larrabee-graphics-card-prototype-just-sold-for-dollar5000/

Larrabee (Aubrey Isle, End-of-Life) had 32 hardware texture map units.

Xeon Phi (Knights Corner, End-of-Life) also has 32 hardware texture map units. The successor is Intel H3C XG310 which is based on Xe-LP GPU architecture.

The successor to H3C XG310 is Data Center GPU series e.g.
Arctic Sound (Generation 12.5, DirectX12 Feature Level 12_1),
Flex series (Generation 12.7 similar to ARC, DIrectX12 Ultimate Feature Level 12_2),
Ponte Vecchio (Generation 12.5, DirectX12 Feature Level 12_1, includes raytracing cores)

The current end point successor to Larrabee is Intel DirectX12 FL12_1 and FL12_2 capable GpGPUs which includes ARC's Generation 12.7 architecture known Flex SKUs..

Xeon Phi and Larrabee are failures. Renaming Larrabee as Xeon Phi means little to me.

You're wrong.

Quote:

Only in the recent years Arc has replaced Xeon Phi.

I'm already aware of Xeon Phi and Sierra Forest (e-Core-based Xeon 6).

Our corporate RDP/VMware VSphere racks are the two 56-core Xeon servers from DELL,. These two servers are assigned to our division.

Last edited by Hammer on 11-Apr-2024 at 01:49 PM.
Last edited by Hammer on 11-Apr-2024 at 01:43 PM.
Last edited by Hammer on 11-Apr-2024 at 01:37 PM.
Last edited by Hammer on 11-Apr-2024 at 01:26 PM.
Last edited by Hammer on 11-Apr-2024 at 01:24 PM.
Last edited by Hammer on 11-Apr-2024 at 01:18 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)

@Matt3k

Quote:

Matt3k wrote: @ppcamiga1 Commodore Amigas biggest mistake was firing Thomas Rattigan, who turned around Amiga when it was heading toward bankruptcy in the 80's. He respected the engineers and could run the business better than anyone at that point. This was a major contribution to where they ended up. Poor management takes awhile to undue it, given Commodore's financial situation and the attitude on production and engineers, it was dead for awhile and just didn't know it yet. You need money not to be bankrupt, to have money you need great leadership.

It was around 1987 to 1989 for "read my lips, no new chips" directive during the A3000's development. The original Amiga team tried with the Amiga Ranger chipset upgrade around 1987 and they were dismantled.

Removing the original Amiga team comes back to bite Commodore in the ass with 3DO's 1993 release.

Last edited by Hammer on 11-Apr-2024 at 02:00 PM.
Last edited by Hammer on 11-Apr-2024 at 01:59 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)

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: That's only the video card. What about complete configuration of a gaming PC AND the price of an Amiga 500? Does A500 have a fast 256-color frame buffer without lossy color compression known as HAM mode?

No, and? The question has still no answer. Care to give one?
Quote:

Fact: Commodore had good revenue levels in 1990 and 1991.

Irrelevant: answer to the question.
Quote:

It would be a tricky situation for the hardware generation switch. The Amiga platform resembles a game console platform with a keyboard since the Amiga graphics chipset is not PC graphics card modular.

Still irrelevant.

The good thing here would be to provide a PC of the time that is a SBC + keyboard like an Amiga 500 and see how much it costed.
Quote:

Quote: SNES was sold in 1990, not 1992. So, same year of the ECS. SNES was released in Japan's 1990 market.

It was also imported from Japan, even if it wasn't available on other markets.
Quote:

From 1990 to 1993, Commodore's revenue core market is Western Europe with the strongest Amiga market in the UK and Germany.

The year 1992 was SNES's European release and direct competition with Commodore.[/quote]
The important thing in this part of the discussion was when it was released. And it was on 1990.

This is about what was possible (and available!) at that time.
Quote:

SNES has packed pixels via Mode 7 and Mode 7 Direct Color modes, but it needs 16-bit RISC SuperFX2 @ 20Mhz for the Doom port.

Irrelevant: the context is 1990, and Doom wasn't available.
Quote:

IF A500++++++ had AGA with packed pixels, it needs a fast 68030 or at least 68020 @ 33Mhz!

No, because that wasn't the context (read: it's still 1990).
Quote:

SNES started to build its install base in 1990 and its larger install base was an advantage in for 3rd party developers during 1992. AGA was ground zero in 1992 with a tiny install base. A500++++++/A1200/CD32 with AAA chipset and 68EC020 @ 14 Mhz CPU wouldn't change the outcome.

Same as above.
Quote:

Quote: Again, useless PADDING, since the CONTEXT was 1990! From 1990 to 1993, Commodore's revenue core market is Western Europe with the strongest Amiga market in the UK and West Germany.

Again, irrelevant.
Quote:

Quote: The CONTEXT was Low-End GAMING market on 1990. There's one thing for certain, Commodore CEO Ali will reject something like stop-gap SuperFX2 just as Ali rejected Commodore UK's upgraded CD32 initiative. https://archive.computerhistory.org/resources/access/text/2013/04/102723262-05-01-acc.pdf Page 119 of 981 For 1992 68000-12 = $5.5 68EC020-16 PQFP = $16.06, it's $15 in 1993 Q1. 68EC020-25 PQFP = $19.99, it's $18 in 1993 Q1.

Irrelevant: it was about 1990.
Quote:

Quote: LOL! Then kick out the Blitter from all Amigas, because it's useless without a fast frame buffer. Your words.

Actually it was because of YOUR words.
Quote:

Amiga's 16-bit Blitter @ 3.5 Mhz is so 1985.

Indeed, but now take a look at what YOU have written and then check WHY I've replied with the above sense.

Let me know if still don't understand it.
Quote:

My position on this matter is a dual 16-bit Blitter. The legacy Amiga software will only see the 1st 16-bit Blitter.

You continue to repeat this even on EAB like a parrot. Unfortunately this is because you've no clue at all of how an Amiga worked and, specifically, how game were developed using its hardware.

Just to be short: your proposal is a complete non sense.

However, since YOUR is the idea, then why don't you give MORE DETAILS about how it was supposed to work.
For example, let's say that you've a 48x48 pixels BOB to be drawn on the screen at an arbitrary position (X, Y): how game developers could have used your dual 16-bit Blitter'
Quote:

Quote: No, it's in your a$$ only because you're an AMD die hard blind fanatical, so you're still burning for this. You can't handle the truth straight from the Xbox team.

As someone already said you on EAB, it's childish. When do you think to grow and avoid those ridiculous ways to "argue"?
Quote:

Did you forget the Amiga camp was "Intel Outside"?

Yes, I forgot because I GREW, where other people are still nursing children.
Quote:

Did you forget MorphOS 3.12 Beta was demonstrated on the "AMD64, Ryzen 5 3600" PC? https://www.youtube.com/watch?v=oAUZivdbn-Q

No, and? Please, tell me: was it working ONLY on this PC? I mean: only on this AMD CPU?
Quote:

Quote: Neither the CPU, which was a regular P3 one (only with half the L2 cache). You can't handle the truth straight from the Xbox team.

The truth is the Xbox HAVE (I've highlighted for YOUR convenience) an Intel CPU.

If you don't know it you can even reach me and we open it, since I still have one model (which I still use with my kids, BTW).
Quote:

Removing Bill Gates also removes the close Intel link e.g. Xbox 360 with IBM PPE.

A complete non-sense. Care to PROVE it?
Quote:

Quote: It remains the most important component compared to the CPU. It's so much important that even an AMD die hard blind fanatical like you is using nVidia GPUs instead the ones of your beloved AMD... Your attempt to paint me as an "AMD die-hard"

Nothing to paint: you've proved it here SEVERAL TIMES.
Quote:

is debunked by my NVIDIA GPU selection.

You had no choice because AMD isn't a choice since very long time when talking about games, so you were and are still forced.

nVidia offers the best hardware for games.
Quote:

I have no problems switching to Qualcomm Oryon as long it matches the gaming PC's use case.

Which is? Mobile gaming? I don't think that your current video card suggests something like that...
Quote:

Quote: LOL: TOTAL NON-SENSE! The FPU does NOT use the graphic card's fast-frame buffer for its calculations! There's NOTHING that could help the CPU's FPU for processing its 3D calculations. What happens when you use K7 Athlon XP 2200+ (1800Mhz) with an original IBM VGA? Hint: it's a roll-up slide show. LOL A reminder... GPU I/O architecture basics [CRAPPY HUGE IMAGE DESTROYING THE SITES LAYOUT] This is applicable for both AMD's VEGA/RDNA1,2,3 and NVIDIA's Maxwell/Pascal/Turing/Ampere/ADA. Compute Shader IO path was handled by the CPU. Modern GPU's compute shaders took over these roles. Compute Shader usually uses TMUs for the I/O. Pixel Shader IO path is handled by raster i.e. ROPS. Pixel Shader role was handled by CPU before it was handed to the GPU's pixel shaders. ROPS is the good old fill rate frame buffer IO units. In RTX and RX RDNA 2/RNDA 3, the raytracing core's (BVH, boxes, geometry, search engine problem) processing is handled by their I/O path which is separated from ROPS I/O path! Certain concepts remained similar and they are hardware accelerated.

The usual Hammer's PADDING. Let me report the original context here:

"ME: How the Pentium-FPU has improved the VIDEO part of Quake will remain a mystery...

YOU: 3D render has a pipeline. Fast math power needs a fast-frame buffer."

Which is still:
LOL: TOTAL NON-SENSE! The FPU does NOT use the graphic card's fast-frame buffer for its calculations! There's NOTHING that could help the CPU's FPU for processing its 3D calculations.

You don't know of what you talk about!!!
Quote:

Quote: And here comes back the famous Hammer's crystal ball... Another total non-sense, of course, since chips sold on 1990 were designed years before, and of course targeting 1990 and NOT 1993 or something like that. I'm aware of this. Don't assume. For example, ET4000AX's 1989 release would have its R&D in the 1987 to 1988 time frame.

Your words does NOT show that you were aware of it. Care to READ AGAIN the original context AND understand it?
Quote:

Quote: You've absolutely no clue of how chips are designed and go to the market! That's your assumption which makes you an ass.

Oh, now you're going personal: what a news!

That's the only thing that you can do to "argue" on the discussion, since you've shown (AGAIN!) to have NOT CLUE AT ALL of what was about it.
Quote:

Quote: Again, you've no idea of what you talk about. Larrabee was a failure, yes, and it was replaced by the Xeon Phi: NOT Arc! Wrong. Larrabee was a cGPU addon card product e.g. https://www.pcgamer.com/a-working-intel-larrabee-graphics-card-prototype-just-sold-for-dollar5000/ Larrabee (Aubrey Isle, End-of-Life) had 32 hardware texture map units. Xeon Phi (Knights Corner, End-of-Life) also has 32 hardware texture map units. The successor is Intel H3C XG310 which is based on Xe-LP GPU architecture. The successor to H3C XG310 is Data Center GPU series e.g. Arctic Sound (Generation 12.5, DirectX12 Feature Level 12_1), Flex series (Generation 12.7 similar to ARC, DIrectX12 Ultimate Feature Level 12_2), Ponte Vecchio (Generation 12.5, DirectX12 Feature Level 12_1, includes raytracing cores) The current end point successor to Larrabee is Intel DirectX12 FL12_1 and FL12_2 capable GpGPUs which includes ARC's Generation 12.7 architecture known Flex SKUs.. Xeon Phi and Larrabee are failures. Renaming Larrabee as Xeon Phi means little to me. You're wrong.

Again, you don't know AT ALL of what you're taking about!

I was QA engineer at Intel working or... rolling drum... the Xeon Phi family just after that Larrabee was declared EoL and "converted" to Kights Ferry first and then to the more successful (we had several customers for it) Knights Corner (I started working on this!).

Xeon Phi was a PCI-Express card EXACTLY LIKE Larrabee.

So, it was a COPROCESSOR. Even because it was NOT able to execute a regular PC OS, since it LACKED SEVERAL INSTRUCTIONS. Do you know this "detail", Google-search man? I don't think so, because you're a complete ignorant about this topic.
Quote:

Quote: Only in the recent years Arc has replaced Xeon Phi. I'm already aware of Xeon Phi and Sierra Forest (e-Core-based Xeon 6).

Clearly not: see above!
Quote:

Our corporate RDP/VMware VSphere racks are the two 56-core Xeon servers from DELL,. These two servers are assigned to our division.

IRRELEVANT. As usual...

@bhabbott

From EAB: https://eab.abime.net/showpost.php?p=1679143&postcount=3574

Quote:

bhabbott wrote: This article has some interesting historical information that some people may not know about, but the rest of it is unrealistic armchair engineering.

Which you completely failed to prove here. You tried, but you miserably failed. But this doesn't stop you to continue writing your non-sense on EAB.
Quote:

No, ECS could not have had 'enormously more useful' 256 colors. In fact they would be pretty useless, because:- 1. 8 bitplanes would suck up all the bandwidth.

Wrong. You don't know how the Amiga works. E.g.: you "forgot" the horizontal and vertical blank periods.
Quote:

Combine that with double the blitter load and it would be horribly slow.

With your completely wrong math, yes. However 8 bitplanes instead of 6 is NOT double the load for the Blitter in the real world...
Quote:

The same problem would also occur with dual playfields.

Yes, and see above.
Quote:

This is why AGA has 2x and 4x fetch modes.

Which does NOT mean that you've so many slots available for the Blitter. I've already provided you complete math / proof about that some time ago: have you forget it?
Quote:

2. The main reason for having 256 colors is to match the graphics in PC games. But when they consist of 32 colors with 6 brightnesses achieved by bit shifting it becomes very difficult to match palettes. The darker shades lose color resolution, so in practice you end up with far less than 256 useful colors.

This hasn't stopped Archimedes to have it implemented with even 16 shades.

Which, BTW, is much worse than an Amiga having 8 shares because of the 32 colours base palette (Archimedes had only 16 of them).
Quote:

OCS already had 64 colors with EHB, yet very few games used it because even 6 bitplanes taxed the system too much. if people thought AGA was disappointing they would be even less impressed by ECS with 256 colors done this way.

First of all, let GAME developers decide what and how to use the Amiga hardware. You've NOT worked at Amiga games, so you have no experience about it.
For example, a game of 1989.
Amiga version: https://www.mobygames.com/game/534/indiana-jones-and-the-last-crusade-the-graphic-adventure/screenshots/amiga/156067/
PC VGA version: https://www.mobygames.com/game/534/indiana-jones-and-the-last-crusade-the-graphic-adventure/screenshots/dos/10783/
What do you prefer, the first or the second one? Do you know that a version very similar to the second one could have been possible with my proposal?

Second, ECS means at least 1MB of Chip Mem. Which means that games like Fightin' Spirit could have been implemented using 256 colours (instead of 64 with EHB) with less frames for the characters or all frames with around 1.2 MB Chip Mem, while keeping the existing frame rate. Yes, you've read it correctly: the game could have run with 256 colours.
Quote:

He says adding stereo panning to Paula would be 'trivial'. But as you can see from the die photo, there is no spare silicon to add this feature to the existing design. A completely new chip would have the designed, and as Paula is a mixed signal (digital and analog) IC this would not be easy.

You've already got replies about, but again: NOT! It could have been possible and there was still some space.

In fact, you don't even need new DACs for this panning it.
Quote:

The suggestions for making ChipRAM access faster don't apply to the 68000,

Guess what: it was NOT proposed for the 68000!
Quote:

and would be useless on faster CPUs without greatly increasing ChipRAM bandwidth.

Again, another sentence by an expert game developer? Do you understand how a game was developed and worked on an Amiga?
Quote:

This would be a lot more than just a little tweak.

Oh, yes. Checking the CPU bus request every 2 cycles instead of every 4 requires dramatic changes, eh!
Quote:

The caching ideas would also involve a lot of extra circuitry and tricky logic,

You don't know of what you've talked about. Clearly NO, as I've already fully explained in the article. Care to read AND understand it, finally?
Quote:

with significant compatibility issues.

ALL my proposals are FULLY BACKWARD COMPATIBLE!

You haven't read or, much worse, understood the article!
Quote:

This article derides the "Read my lips – no new chips" command, but if it was intended to avoid stuff like this then it was justified.

Clearly no, since the ECS brought SEVERAL NEW FEATURES. AKA more transistors used. Even for completely USELESS staff, like the UHRES bitplane and sprite.
Quote:

Trying to jam all these features into ECS could have set back its release by years.

Well, it depends on the engineers which should have implemented them.

For example, even the Commodore 16 had the possibility to set 8 level of shares for its fixed 16 colours palette...
Quote:

Commodore had to get the A3000 out while it still had a chance to be relevant - thus 'no new chips' - but in fact it had heaps of new chips in it already.

Which needed TIME and RESOURCES to be developed. Which could have used much better.

Understood now?

@cdimauro

Mike Sinz's statements on A3000's "no new chips"
http://ftp.bambi-amiga.co.uk/amigahistory/mikesinz.html

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")


Engineers did NOT see ECS as "new".

_________________
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)

@cdimauro

Quote:

No, and? The question has still no answer. Care to give one?

Commodore being dead speaks for itself.

Quote:

Irrelevant: answer to the question.

It's relevant. A mistake with the main model transition will result in Commodore's bankruptcy.

A600's release, A600's sales flop, and A500's cancelation have caused Commodore 1992 revenues to drop and post unsustainable P&L losses.

Quote:

Irrelevant: it was about 1990.

Again, Commodore's revenues are in a pretty good state in 1990 and 1991.

The AGA transition should been executed Q3-to-Q4 Xmas 1991.

Releasing the A500 with AGA would cause PR issues for the new A500 Q4 1991 customers.

There should be enough AGA Amiga stockpiles for Q3-to-Q4 Xmas 1991 sales.

From http://ftp.bambi-amiga.co.uk/amigahistory/leweggebrecht.html

Question: Tell me about AAA - it's been worked on since 1989?

Lew Eggebrecht: "Yes, we worked on it from an architectural point of view for a long time but it's only been serious for about a year. It was obvious that AAA was not going to meet our cost targets for the mid to low end systems. We wanted to continue that development and we also had to have an enhancement quickly so, AA was the solution to that problem. It would have been nice to have AAA at the same time as AA but we just couldn't get there."


AA (AGA) and AAA have been in development since 1989. The 1987 and 1988 time is gone.

For AGA's 1990 release, the 256-color chipset R&D should been started in early 1988 and mostly completed in 1989.

C65's 256-color capable chipset was mostly completed in Dec 1990.

AA3000+ (AGA) was booting state in Feb 1991.

There was IDE controller mandate demanded by Commodore Germany (cite Commodore the Inside Story - The Untold Tale of a Computer Giant, David Pleasance).

AA3000+ doesn't have Gayle (Gary replacement, IDE, PCMCIA) and Budgie (cost-reduced Buster and Ramsey).

Mid-range ECS-based A1000Jr still has Super Buster, Fat Gary, Ramsey, and 68EC020-25.

Quote:

You continue to repeat this even on EAB like a parrot. Unfortunately this is because you've no clue at all of how an Amiga worked and, specifically, how game were developed using its hardware.

I'm aware that Alice has 16-bit data pins while Lisa has 32-bit data pins.

Quote:

However, since YOUR is the idea, then why don't you give MORE DETAILS about how it was supposed to work. For example, let's say that you've a 48x48 pixels BOB to be drawn on the screen at an arbitrary position (X, Y): how game developers could have used your dual 16-bit Blitter'

For new software, the dual 16-bit Blitter should act like a 32-bit Blitter in gang mode.

Quote:

As someone already said you on EAB, it's childish. When do you think to grow and avoid those ridiculous ways to "argue"?

They can't handle the truth from the Xbox team.

Xbox creator apologizes to AMD over last-minute switch to Intel CPUs 20 years ago.
https://www.techspot.com/news/91749-xbox-creator-apologizes-amd-over-last-minute-switch.html

Quote:

Yes, I forgot because I GREW, where other people are still nursing children.

For the original Xbox, Bill Gates has Intel bias at the expense of AMD.

Quote:

A complete non-sense. Care to PROVE it?

January 13, 2000: Bill Gates steps down from his role as Microsoft CEO.

Quote:

you've proved it here SEVERAL TIMES.

Debunked by my NVIDIA RTX usage.

Hint: Intel LGA 1700 is a dead-end platform and lacks dedicated NVMe version 5.0 lanes and Ice Lake+ AVX-512 support. Intel LGA 1700 has no road map for post-RaptorLake AVX-512 support.

During the Intel Pentium III and Pentium 4 era, I had AMD K7 and K8.

During AMD's Bulldozer era, I had Intel Core 2 Duo E6800, Core i7-740QM, i7-2600, i7-3770K, i7-4790K.

During AMD Zen 1, I had i9-9900K CoffeeLake-R and i7-7820X SkyLake X.

I received a free dual socket Xeon X5680 server and Quadro P2000 from work and I use them for my NAS role Windows Server 2019 EE use case. I have another recently obtained free decommissioned two-socket Broadwell-EP Xeon Server on standby from my work. I plan for another NAS role with Windows Server 2022. Older Xeons still works well enough for my NAS Windows Server-based use case. They come with ECC RDIIMM memory installed.

My i7-7820X SkyLake X's X299 chipset (ASUS ROG Strix X299-E Gaming) doesn't support ECC memory.

ASUS/ASRock AM5 X670 motherboards support ECC UDIMM DDR5.

You're wrong.

Quote:

You had no choice because AMD isn't a choice since very long time when talking about games, so you were and are still forced.

Wrong, professional apps with raytracing. For raster gaming, Radeon RX 7900 XTX is the price and performance alternative to RTX 4080/4080 Super.

Quote:

nVidia offers the best hardware for games.

AMD's raster gaming use case is price-performance competitive.

My use case includes other non-gaming e.g. OpenAI related which I use two RTX 3080 Ti for the job.


Quote:

The usual Hammer's PADDING. Let me report the original context here: "ME: How the Pentium-FPU has improved the VIDEO part of Quake will remain a mystery... YOU: 3D render has a pipeline. Fast math power needs a fast-frame buffer." Which is still: LOL: TOTAL NON-SENSE! The FPU does NOT use the graphic card's fast-frame buffer for its calculations! There's NOTHING that could help the CPU's FPU for processing its 3D calculations.

https://www.youtube.com/watch?v=octArwHpaiY
IBM's original VGA with AMD K7 Athlon XP 2200+ (1800Mhz) running Quake.
Results of timedemo were:
360x480 - 3.2FPS
320x240 - 7FPS
320x200 - 8.6FPS

Tested on Athlon XP 2200+, Soltek SL75-KAV (Via KT133A), 512MB SDR CL3.

Again, 3D render has a graphics pipeline. Fast math power needs a fast-frame buffer.
3D render's performance MUST account from the start to the end!

PS; AGA's result is superior when attached to Emu68's 1.8 GHz ARM Cortex A72 CPU.

A graphics pipeline's stage can be done on the CPU e.g. software BVH raytracing (box and triangle test) from World of Tanks https://youtu.be/dXbjmF--QVc (Intel promoted software BVH raytracing on CPU's SIMD extensions).





When compared to DX9 GPUs, DirecX10 GPU takes over additional CPU-like workloads in the graphics pipeline.

Successive GPU generation has been taking over the CPU's render workload in an evolving manner.

Graphics pipeline doesn't mean it's 100 percent video card only.

The frame buffer issue:
Hint: Xbox One's 32 MB ESRAM stores 1080p and 900p render target frame buffers while the rest of DDR3 memory has texture storage.

Xbox One's 1080p split render target frame buffer with 2X MSAA example has 11 MB on 32 MB ESRAM and 4.8 MB on 256-bit DDR3-2133. 32 MB ESRAM soaks up the bulk of GPU's high memory bandwidth needs.

Render target frame buffers usually have small storage with high bandwidth needs.

PS4's Infamous Second Son with almost 4.5 GB usage has the following:
290 MB render targets (render frame buffers)
2500 MB for loaded data (e.g. art assets).
Flexbile memory = 500 MB
Atlases + Buffers = 370 MB
Movie = 270 MB
Command Buffers = 115 MB
Fixed = 100 MB
Heap = 100 MB

Recent GPUs have delta color compression (DCC). PS4 GPU doesn't have delta color compression.

Modern PC GPU's large video memory mostly stores textures to minimize stalls while render targets usually have small video memory storage requirements.

Like Xbox One's 32 MB ESRAM, recent PC GPUs from AMD's RDNA 2 and NVIDIA's ADA have double-digit MB on-chip caches e.g.

RTX 4090 has a 72 MB L2 cache before delta color compression (DCC) is applied.
RTX 4080 has a 64 MB L2 cache before delta color compression (DCC) is applied.
RX 7900 XTX has 96 MB L3 cache before delta color compression (DCC) is applied.

NVIDIA Pascal GPUs have 2:1, 4:1, and 8:1 compression ratio DCC modes.

For modern GPUs, external video memory is not the fastest video memory storage.

Last edited by Hammer on 18-Apr-2024 at 07:54 AM.
Last edited by Hammer on 17-Apr-2024 at 02:54 PM.
Last edited by Hammer on 17-Apr-2024 at 02:52 PM.
Last edited by Hammer on 17-Apr-2024 at 02:50 PM.
Last edited by Hammer on 17-Apr-2024 at 02:12 PM.
Last edited by Hammer on 17-Apr-2024 at 02:02 PM.
Last edited by Hammer on 17-Apr-2024 at 01:59 PM.
Last edited by Hammer on 17-Apr-2024 at 01:56 PM.
Last edited by Hammer on 17-Apr-2024 at 01:48 PM.
Last edited by Hammer on 17-Apr-2024 at 01:43 PM.
Last edited by Hammer on 17-Apr-2024 at 01:39 PM.
Last edited by Hammer on 17-Apr-2024 at 01:38 PM.
Last edited by Hammer on 17-Apr-2024 at 01:30 PM.
Last edited by Hammer on 17-Apr-2024 at 12:58 PM.
Last edited by Hammer on 17-Apr-2024 at 12:52 PM.
Last edited by Hammer on 17-Apr-2024 at 12:43 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)

@Hammer

Quote:

Hammer wrote: @cdimauro Mike Sinz's statements on A3000's "no new chips" http://ftp.bambi-amiga.co.uk/amigahistory/mikesinz.html 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") Engineers did NOT see ECS as "new".

And what's the point? I'm an engineer as well and I say that those are new chips. With the big difference that I can PROVE it.

In fact, ECS has introduced NEW features which required ADDITIONAL TRANSISTORS for being implemented. Which NECESSARILY brought to NEW chips to be designed.

Do you spot the difference?

And what about the Amber chip? Could you tell me how this BIG chip matches the "read my leaps, no new chips"? Why you systematically AVOID giving a feedback about this?
Quote:

Hammer wrote: @cdimauro Quote: No, and? The question has still no answer. Care to give one? Commodore being dead speaks for itself.

Again, you do NOT answer to the question:

That's only the video card. What about complete configuration of a gaming PC AND the price of an Amiga 500?

When do you plan answer this? I've to recall you that this question comes because of something that YOU have stated.
Quote:

Quote: Irrelevant: answer to the question. It's relevant. A mistake with the main model transition will result in Commodore's bankruptcy. A600's release, A600's sales flop, and A500's cancelation have caused Commodore 1992 revenues to drop and post unsustainable P&L losses.

Which, again, has nothing to do with the above question that I've reported for YOUR convenience.
Quote:

Quote: Irrelevant: it was about 1990. Again, Commodore's revenues are in a pretty good state in 1990 and 1991.

And the context was 1990, as I've already reported several time. Do you understand why I've written this article (and second one)? Do you understand that the context was this:

"Low-End GAMING market on 1990"

?
Quote:

The AGA transition should been executed Q3-to-Q4 Xmas 1991. Releasing the A500 with AGA would cause PR issues for the new A500 Q4 1991 customers. There should be enough AGA Amiga stockpiles for Q3-to-Q4 Xmas 1991 sales.

See above: irrelevant, since it's completely out of context.
Quote:

From http://ftp.bambi-amiga.co.uk/amigahistory/leweggebrecht.html Question: Tell me about AAA - it's been worked on since 1989? Lew Eggebrecht: "Yes, we worked on it from an architectural point of view for a long time but it's only been serious for about a year. It was obvious that AAA was not going to meet our cost targets for the mid to low end systems. We wanted to continue that development and we also had to have an enhancement quickly so, AA was the solution to that problem. It would have been nice to have AAA at the same time as AA but we just couldn't get there."

Nice that you've reported it. Now you should CAREFULLY read it AND understand what he telling.

You always have put all fault/negligence towards Commodore's management. Which has clearly it's big part on the company's demise.

However declarations like those that you've reported FURTHERLY (because I've already reported other things on the same topic) show that Commodore's engineers have also their big contribution to that. IF you understand what they did (and, even more, what they have NOT done) and being intellectually honest about it.
Quote:

AA (AGA) and AAA have been in development since 1989. The 1987 and 1988 time is gone. For AGA's 1990 release, the 256-color chipset R&D should been started in early 1988 and mostly completed in 1989. C65's 256-color capable chipset was mostly completed in Dec 1990. AA3000+ (AGA) was booting state in Feb 1991. There was IDE controller mandate demanded by Commodore Germany (cite Commodore the Inside Story - The Untold Tale of a Computer Giant, David Pleasance). AA3000+ doesn't have Gayle (Gary replacement, IDE, PCMCIA) and Budgie (cost-reduced Buster and Ramsey). Mid-range ECS-based A1000Jr still has Super Buster, Fat Gary, Ramsey, and 68EC020-25.

OK, and? All that has nothing to do with the CONTEXT and, specifically, with my two articles.
Quote:

Quote: You continue to repeat this even on EAB like a parrot. Unfortunately this is because you've no clue at all of how an Amiga worked and, specifically, how game were developed using its hardware. I'm aware that Alice has 16-bit data pins while Lisa has 32-bit data pins.

That's something that even a stone can repeat, because you're stating the obvious.

See above now.
Quote:

Quote: However, since YOUR is the idea, then why don't you give MORE DETAILS about how it was supposed to work. For example, let's say that you've a 48x48 pixels BOB to be drawn on the screen at an arbitrary position (X, Y): how game developers could have used your dual 16-bit Blitter' For new software, the dual 16-bit Blitter should act like a 32-bit Blitter in gang mode.

"gang mode"?!? What is it? Care to finally explain it? Wasn't it about YOU that talked this "dual 16-bit Blitter"? It wasn't me, right? Then explain how it should have worked out.

Specifically, how it works with the most common Blitter operation that developers have extensively used: cookie cut. Here it is:
"let's say that you've a 48x48 pixels BOB to be drawn on the screen at an arbitrary position (X, Y): how game developers could have used your dual 16-bit Blitter?"

BTW, do you understand that a DUAL Blitter means at least DOUBLING the transistors required for implementing it on (the new) Agnus?
Quote:

Quote: As someone already said you on EAB, it's childish. When do you think to grow and avoid those ridiculous ways to "argue"? They can't handle the truth from the Xbox team.

I've no problem with the truth but I've problem with people which write non-sense and invent things.
Quote:

Xbox creator apologizes to AMD over last-minute switch to Intel CPUs 20 years ago. https://www.techspot.com/news/91749-xbox-creator-apologizes-amd-over-last-minute-switch.html

OK, and what's the point here? They can still cry about that decision, and?
Quote:

Quote: Yes, I forgot because I GREW, where other people are still nursing children. For the original Xbox, Bill Gates has Intel bias at the expense of AMD.

First of all you've NEVER PROVEN IT. It's YOUR statement and YOU have to prove it.

The only relevant thing is that Microsoft chose Intel for the CPU of its first console. DOT.

Which was a LEGIT choice. DOT #2.

What you're stating here clearly shows your bias in favour of AMD. But that's YOUR problem.
Quote:

Quote: A complete non-sense. Care to PROVE it? January 13, 2000: Bill Gates steps down from his role as Microsoft CEO.

Another non-sense and you still refuse to PROVE YOUR (YOUR!) statements.
Quote:

Quote: you've proved it here SEVERAL TIMES. Debunked by my NVIDIA RTX usage.

What's not clear to you that you, as gamer, had NO choice other than nVidia?
Quote:

Hint: Intel LGA 1700 is a dead-end platform and lacks dedicated NVMe version 5.0 lanes and Ice Lake+ AVX-512 support. Intel LGA 1700 has no road map for post-RaptorLake AVX-512 support.

Hammer's PADDING.
Quote:

During the Intel Pentium III and Pentium 4 era, I had AMD K7 and K8.

What a news...
Quote:

During AMD's Bulldozer era, I had Intel Core 2 Duo E6800, Core i7-740QM, i7-2600, i7-3770K, i7-4790K. During AMD Zen 1, I had i9-9900K CoffeeLake-R and i7-7820X SkyLake X.

Because you had no choices: AMD had crappy products at the time, and no one with a salt of grain could have bought them.

But when you had choice, you've always preferred AMD. See above, for example.
Quote:

I received a free dual socket Xeon X5680 server and Quadro P2000 from work and I use them for my NAS role Windows Server 2019 EE use case. I have another recently obtained free decommissioned two-socket Broadwell-EP Xeon Server on standby from my work. I plan for another NAS role with Windows Server 2022. Older Xeons still works well enough for my NAS Windows Server-based use case. They come with ECC RDIIMM memory installed.

Sure. And only because they came for free...
Quote:

My i7-7820X SkyLake X's X299 chipset (ASUS ROG Strix X299-E Gaming) doesn't support ECC memory. ASUS/ASRock AM5 X670 motherboards support ECC UDIMM DDR5.

Hammer's PADDING.
Quote:

You're wrong.

Care to tell me from where your "Hammer" username is coming?
Quote:

Quote: You had no choice because AMD isn't a choice since very long time when talking about games, so you were and are still forced. Wrong, professional apps with raytracing. For raster gaming, Radeon RX 7900 XTX is the price and performance alternative to RTX 4080/4080 Super.

Gamers like to use the most modern technologies: raster gaming is a thing of the past.

Gamers which had money on their pocket (so, YOU) bought, and continue to buy, nVidia graphic cards because they are interested on playing with the latest technologies.

BTW, DLSS can be used with raster graphics as well, and nVidia has a clear edge here as well.
Quote:

Quote: nVidia offers the best hardware for games. AMD's raster gaming use case is price-performance competitive.

See above: no. Because DLSS helps here as well.
Quote:

My use case includes other non-gaming e.g. OpenAI related which I use two RTX 3080 Ti for the job.

Do you play games or not?
Quote:

Quote: The usual Hammer's PADDING. Let me report the original context here: "ME: How the Pentium-FPU has improved the VIDEO part of Quake will remain a mystery... YOU: 3D render has a pipeline. Fast math power needs a fast-frame buffer." Which is still: LOL: TOTAL NON-SENSE! The FPU does NOT use the graphic card's fast-frame buffer for its calculations! There's NOTHING that could help the CPU's FPU for processing its 3D calculations. https://www.youtube.com/watch?v=octArwHpaiY IBM's original VGA with AMD K7 Athlon XP 2200+ (1800Mhz) running Quake. Results of timedemo were: 360x480 - 3.2FPS 320x240 - 7FPS 320x200 - 8.6FPS Tested on Athlon XP 2200+, Soltek SL75-KAV (Via KT133A), 512MB SDR CL3. Again, 3D render has a graphics pipeline. Fast math power needs a fast-frame buffer. 3D render's performance MUST account from the start to the end! PS; AGA's result is superior when attached to Emu68's 1.8 GHz ARM Cortex A72 CPU. A graphics pipeline's stage can be done on the CPU e.g. software BVH raytracing (box and triangle test) from World of Tanks https://youtu.be/dXbjmF--QVc (Intel promoted software BVH raytracing on CPU's SIMD extensions). When compared to DX9 GPUs, DirecX10 GPU takes over additional CPU-like workloads in the graphics pipeline. Successive GPU generation has been taking over the CPU's render workload in an evolving manner. Graphics pipeline doesn't mean it's 100 percent video card only. The frame buffer issue: Hint: Xbox One's 32 MB ESRAM stores 1080p and 900p render target frame buffers while the rest of DDR3 memory has texture storage. Xbox One's 1080p split render target frame buffer with 2X MSAA example has 11 MB on 32 MB ESRAM and 4.8 MB on 256-bit DDR3-2133. 32 MB ESRAM soaks up the bulk of GPU's high memory bandwidth needs. Render target frame buffers usually have small storage with high bandwidth needs. PS4's Infamous Second Son with almost 4.5 GB usage has the following: 290 MB render targets (render frame buffers) 2500 MB for loaded data (e.g. art assets). Flexbile memory = 500 MB Atlases + Buffers = 370 MB Movie = 270 MB Command Buffers = 115 MB Fixed = 100 MB Heap = 100 MB Recent GPUs have delta color compression (DCC). PS4 GPU doesn't have delta color compression. Modern PC GPU's large video memory mostly stores textures to minimize stalls while render targets usually have small video memory storage requirements. Like Xbox One's 32 MB ESRAM, recent PC GPUs from AMD's RDNA 2 and NVIDIA's ADA have double-digit MB on-chip caches e.g. RTX 4090 has a 72 MB L2 cache before delta color compression (DCC) is applied. RTX 4080 has a 64 MB L2 cache before delta color compression (DCC) is applied. RX 7900 XTX has 96 MB L3 cache before delta color compression (DCC) is applied. NVIDIA Pascal GPUs have 2:1, 4:1, and 8:1 compression ratio DCC modes. For modern GPUs, external video memory is not the fastest video memory storage.

Another massive PADDING which is totally irrelevant for the specific context, which I report for YOUR convenience:

"How the Pentium-FPU has improved the VIDEO part of Quake will remain a mystery"

Now, context is: QUAKE. Understood?

Quake was first released as PURELY SOFTWARE ENGINE. It means that CPU did ALL the work.

Now, I repeat it again: how the FPU has improved Quake's VIDEO part?

VIDEO = the OUTPUT (read: on the VRAM) on the graphic cards.

Care to finally give an answer instead of reporting complete non-sense like the super fast embedded memory of the XBox 360?

CONTEXT: QUAKE. CONTEXT: SOFTWARE ENGINE. CONTEXT: VIDEO PART.

Understood?

@cdimauro

Quote:

And what's the point? I'm an engineer as well and I say that those are new chips. With the big difference that I can PROVE it.

You're not a Commodore engineer.

Quote:

In fact, ECS has introduced NEW features which required ADDITIONAL TRANSISTORS for being implemented. Which NECESSARILY brought to NEW chips to be designed. Do you spot the difference? And what about the Amber chip? Could you tell me how this BIG chip matches the "read my leaps, no new chips"? Why you systematically AVOID giving a feedback about this?

A2320 card also used the Amber chip.

Amber with frame buffer is a workaround for the "no new chips" directive and it delivered stable 640x400p 16 colors on VGA monitors. It effectively matches VGA's 640x480 16-color baseline.

You're making the argument outside of Commodore based on retail product releases!

1. ECS Agnus existed with the 1989 released A500 Rev 6A. ECS Agnus A is missing the AB and B model's 2MB Chip RAM address.

2. 2MB Chip RAM configuration existed with 1989 A500 Rev 6A's PCB design.

Commodore would have an internal use-only A500 Rev 6A prototype with a 2MB Chip RAM test configuration.

Quote:

Again, you do NOT answer to the question: That's only the video card. What about complete configuration of a gaming PC AND the price of an Amiga 500?

Amiga 500 had its "power without the price" advantage from 1987 (starting at $699 USD in 1987) to 1990. This is not in dispute.

For the low-cost gaming use case, a game console like SNES is a threat to the majority of low-end Amiga OCS/ECS.

1992-1994 Gaming PC delivered "new 32-bit 2.5D/3D gaming experience" above SNES's late 16-bit strong 2D gaming experience.

IBM laid the foundation for the rise of the "gaming PC" in 1987 when IBM established VGA's 256-color use case and it was cloned, improved, and cost-reduced.


Commodore had a choice, deliver a strong 2D gaming experience or deliver a "new 32-bit 2.5D/3D gaming experience".


The hardware sales are based on the use cases, delivered experience, and price.

A1200 entered the US market at about $599 in Q4 1992.

With US inflation, A500's $699 in 1987 is $804.22 in 1990 or $863.29 in 1992.


SNES entered the US market in 1991 with a $199 price tag. Commodore wouldn't be matching SNES's $199 USD due to a different business model.


For the gaming experience, what does $599 give to the consumer? Another late 16-bit experience?


Quote:

And the context was 1990, as I've already reported several time. Do you understand why I've written this article (and second one)? Do you understand that the context was this: "Low-End GAMING market on 1990"

For the Low-End GAMING market in 1990", 256 color display capable SNES is a threat.

http://www.bambi-amiga.co.uk/amigahistory/leweggebrecht.html
Lew Eggebrecht, Vice President of Engineering at Commodore.

Lew Eggebrecht: We do get squeezed with clone PCs at the top and Sega underneath, and also boxes like 3D0,


Lew Eggebrecht: "We can't make that decision right now - it's something we'll have to look at but in that time frame, even in the low end, every machine is likely to have a DSP. It's a cost thing - although the AT&T chip itself is only $20 to $30 or so. AT&T has a number of lower cost options, as well, that are designed more specifically to go on the motherboard

------

DSP3210 has a $20 to $30 price range and it's in the game plan, but Commodore runs out of time and money.

Lew Eggebrech is open for Nintendo-style officially supported DSP for ALL Amigas. It's too bad Eggebrech was not in the hot seat earlier in the 1988 timeframe.

RISC core DSP3210 at 50Mhz (13.5 MIPS INT32, 27.5 MFLOPS FP32) is superior to 16 bit RISC SuperFX2 @ 20Mhz (20 MIPS INT16).

Lew Eggebrech is better than Bill Sydnes.

Last edited by Hammer on 19-Apr-2024 at 08:42 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)

@cdimauro

Quote:

You always have put all fault/negligence towards Commodore's management. Which has clearly it's big part on the company's demise. However declarations like those that you've reported FURTHERLY (because I've already reported other things on the same topic) show that Commodore's engineers have also their big contribution to that.

Prove Commodore engineers canceled the A500 and tanked Commodore's revenues in 1992.

Quote:

IF you understand what they did (and, even more, what they have NOT done) and being intellectually honest about it.

Commodore management dismantled the proven original Los Gatos Amiga team.

Key original Los Gatos Amiga engineers led the 1993 released 3DO project.

Read https://en.wikipedia.org/wiki/Amiga_Ranger_Chipset

Commodore killed the purpose of buying Amiga Corporation in the 1st place!

You're not intellectually honest about it.
Quote:

Care to tell me from where your "Hammer" username is coming?

Star Wars. A black-colored Super Star Destroyer named the Night Hammer (renamed Knight Hammer, Executor-class Star Dreadnought) from Star Wars Darksaber (novel) was released in 1995. A nice pointy black starship.

There are a few starships named Hammer in the Star Wars Imperial Navy.

I was a fan of Star Wars Extended Universe until Katherine Kennedy killed it.

Quote:

"gang mode"?!? What is it? Care to finally explain it? Wasn't it about YOU that talked this "dual 16-bit Blitter"? It wasn't me, right? Then explain how it should have worked out. Specifically, how it works with the most common Blitter operation that developers have extensively used: cookie cut. Here it is: "let's say that you've a 48x48 pixels BOB to be drawn on the screen at an arbitrary position (X, Y): how game developers could have used your dual 16-bit Blitter?"

Double the parameters. The programmer has to think about two items at the same time. Think SIMD. Xbox One APU has multiple move engines.

7 Mhz clock speed increase with the 16-bit Blitter method can cause incompatibility with timing-sensitive games. A setting is needed for 3.5 Mhz or 7 Mhz modes.

AAA's 64-bit config has a dual 32-bit chipset approach i.e. two 32-bit Blitters. http://www.bambi-amiga.co.uk/amigahistory/amigaaaa.html

Quote:

BTW, do you understand that a DUAL Blitter means at least DOUBLING the transistors required for implementing it on (the new) Agnus?

Amiga Hombre has a 1 million transistors budget.


Quote:

Because you had no choices: AMD had crappy products at the time, and no one with a salt of grain could have bought them.

Pro-Bulldozer management in AMD was removed from the company.

Quote:

But when you had choice, you've always preferred AMD. See above, for example.

1. LGA 1700 is a dead end with a disabled AVX-512 road map.

https://www.techpowerup.com/review/intel-core-i9-14900k/18.html
Ryzen 7 7800X3D is still the king for PC gaming over Intel Core i9-14900K.

AMD's AM5 road map is superior when compared to Intel's LGA 1700.

2. Intel couldn't deliver AM4's longevity.

3. Intel hasn't delivered unified ECC and non-ECC UDIIMM with their mainstream chipsets.

https://www.asus.com/au/motherboards-components/motherboards/prime/prime-b650-plus/techspec/
ASUS's lower-cost AM5 B650 with ECC UDIMM support.

https://www.asrock.com/mb/AMD/B650%20Pro%20RS/index.asp#Specification
ASrock's lower-cost AM5 B650 with ECC UDIMM support.

Intel has a separate LGA 1700 W680 chipset for ECC UDIMM support.

Intel will "nickel and dime" ECC UDIIMM feature.

4. Intel wants customers to purchase a higher-cost W790 Xeon HEDT workstation platform for 16 P-Core Sapphire Rapids, discrete NVMe 5.0 lanes, IceLake+ AVX-512, and ECC memory. W790 Xeon workstation HEDT platform competes against AMD's Ryzen Threadripper 7000 workstation HEDT platform. I rather spend my money on NVIDIA's RTX 4090 to speed up raytracing. I don't have a fast FP64/FP80 use case that needs the HEDT platform.

There's is choice, spend more on a higher-cost W790 Xeon HEDT workstation platform with 16 P-Core Sapphire Rapids. My forum sig doesn't show my other gaming/Blender 3d /OpenAI PC box with RTX 4090. RTX 4090 has ECC GDDR6X memory option in the control panel.

Intel has a product stack problem with the high-end Core i9 LGA 1700.
Core i9 LGA 1700 Z790 is not matching RTX 4090's ECC support.

Last edited by Hammer on 20-Apr-2024 at 01:22 AM.
Last edited by Hammer on 20-Apr-2024 at 01:20 AM.
Last edited by Hammer on 20-Apr-2024 at 01:13 AM.
Last edited by Hammer on 19-Apr-2024 at 10:02 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)

Quote:

Hammer wrote: Commodore management dismantled the proven original Los Gatos Amiga team. Key original Los Gatos Amiga engineers led the 1993 released 3DO project. Read https://en.wikipedia.org/wiki/Amiga_Ranger_Chipset Commodore killed the purpose of buying Amiga Corporation in the 1st place!

The purpose of buying the Amiga was to have a 16 bit architecture, because that's the way the market was going. It was always intended that the original Amiga team would be 'disbanded' after the project was done. However they were kept on to provide diversity in developing the next models.

During this time the Los Gatos Amiga team's aspirations changed. They were working on a cost-reduced A1000, but wanted to produce a more advanced (and more expensive) machine that would compete in the workstation market. They saw the A500 as competition against their A1000CR, and the A2000 as competition against their high-end 'Ranger' system. In both cases however Commodore's other teams' designs won because they were more practical. The Los Gatos team had become redundant.

IMO that's how it should have turned out. The A1000 was a nice concept, but the A500 was the machine which was needed to popularize the Amiga, and the A2000 was the right 'workhorse' for professional and industrial applications.

Quote:

Hammer wrote: [quote]Amber with frame buffer is a workaround for the "no new chips" directive and it delivered stable 640x400p 16 colors on VGA monitors. It effectively matches VGA's 640x480 16-color baseline.

No. The A3000 had a new (as in 'new chips') native VGA resolution and Amber. The purpose of Amber was was to make the standard Amiga resolutions compatible with VGA. The A2000 already had this since 1998 via Microway's 'AGA 2000' flicker fixer, which was a popular addon for that machine. Therefore it made sense for Commodore to add it into the A3000.

What didn't make sense was putting all those new chips into a machine that would be far too expensive for the average home computer user. Commodore's engineers had become infected with the same desire for high-end toys that made the Los Gatos team go off the rails.

Quote:

Amiga 500 had its "power without the price" advantage from 1987 (starting at $699 USD in 1987) to 1990. This is not in dispute. For the low-cost gaming use case, a game console like SNES is a threat to the majority of low-end Amiga OCS/ECS. 1992-1994 Gaming PC delivered "new 32-bit 2.5D/3D gaming experience" above SNES's late 16-bit strong 2D gaming experience. IBM laid the foundation for the rise of the "gaming PC" in 1987 when IBM established VGA's 256-color use case and it was cloned, improved, and cost-reduced.

256 color VGA was definitely a threat because now the PC had more bitmap colors than the Amiga. However most of the threat was because it was a huge step up for PCs, and therefore anything less was perceived as being no better than EGA with its miserable 16 colors and 64 color palette. But the Amiga already had 64 colors in EHB, and up to 4096 colors with copper lists and HAM.

Quote:

Commodore had a choice, deliver a strong 2D gaming experience or deliver a "new 32-bit 2.5D/3D gaming experience".

Commodore also had to consider its existing userbase. With less than 10% share of the personal computer market it was already struggling to maintain visibility. So it needed to consolidate its 'strong 2D gaming experience' until the 'new 32-bit 2.5D/3D gaming experience' took over.

For 2.5/3D gaming the Amiga needed a much faster CPU, but that would dramatically raise the price. They tried that on the A3000 and it was a disaster. The answer was to just make the 2D chipset a bit stronger, give it 256 colors, and give the machine a low cost 32 bit CPU similar to contemporary 386SX's but with open-ended expandability so it could take future more powerful CPUs as they became available. Which is what they did with the A1200.

Some fans argue that the AGA chipset didn't do enough. I agree with them in one aspect only - it should have had a 256 color chunky mode. Not that it really needed it, but as a marketing ploy. It would have ticked that box which was perceived as being so important, and made porting PC games that little bit easier as well as slightly improving performance.

But like other stuff that armchair engineers think would have been a doddle to add, chunky mode was just to hard to implement in the short time frame they had. Perhaps if the engineers were all geniuses with over 200 IQ they might have managed it - but they weren't. And technically it wasn't needed anyway.

An A1200 with popular 50MHz 030 accelerator card matched the performance of a 386DX-40 (the fastest 386 you could buy). By the time Doom was ported to the Amiga the price of these cards had come way down to the point where many Amiga fans could afford them. That wasn't the case in late 1993 when Doom came out, but then a 386DX system was also very expensive compared to the A1200.

It would not be reasonable to expect Commodore to provide 386DX-40 performance on a machine costing much less. That applied to the SNES too. It couldn't run Doom. Even after adding a Super FX 2 chip in the cartridge they still had to dumb the game down. But an accelerated A1200 runs the original Doom perfectly, providing an identical gaming experience to the PC.

Quote:

SNES entered the US market in 1991 with a $199 price tag. Commodore wouldn't be matching SNES's $199 USD due to a different business model.

Indeed. The very idea of a fully fledged computer matching the price of a game console is ludicrous.

Quote:

For the gaming experience, what does $599 give to the consumer? Another late 16-bit experience?

32 bit actually. :)

It gives the consumer a machine capable of running a greater variety of game genres that appeal to a wider audience. And it gives them the ability do other computing-related tasks on it too - write a letter, paint a picture, compose some music etc., just like people with 'real' computers do. Maybe even get a modem and join the social media of the day (bulletin boards, newsgroups, irc...). Finally it give them access to all those interesting demos and a huge library of (ahem) 'free' software - the real reason most people bought an Amiga!


Quote:

For the Low-End GAMING market in 1990", 256 color display capable SNES is a threat.

Not really. The SNES didn't hit the US until mid 1991, and the UK until 1992. Amiga 500 games were still getting better even as late as 1994. Some were so good that it was hard to tell they weren't AGA!

But the NES was already entrenched in the US before then, despite not being that great. US customers weren't not buying the A500 because it wasn't as good as a console, but because it cost a lot more and wasn't a console. In the US 'computer' meant PC, and 'games machine' meant 'not a computer'. The Amiga didn't fit most people's preconceptions - and those that it did fit already had a C64!

Quote:

DSP3210 has a $20 to $30 price range and it's in the game plan, but Commodore runs out of time and money. Lew Eggebrech is open for Nintendo-style officially supported DSP for ALL Amigas. It's too bad Eggebrech was not in the hot seat earlier in the 1988 timeframe.

Perhaps, but DSP was still not the answer IMO. The Atari Falcon had DSP, what good did that do it?

Quote:

RISC core DSP3210 at 50Mhz (13.5 MIPS INT32, 27.5 MFLOPS FP32) is superior to 16 bit RISC SuperFX2 @ 20Mhz (20 MIPS INT16). Lew Eggebrech is better than Bill Sydnes.

Sure, but this represents a new architecture which is divorced from the Amiga. Maybe a more financial Commodore could have floated this along with some degree of backwards compatibility, but many fans would not be interested. If you have to make that leap then you might as well just buy a PC!

@bhabbott

Quote:

No. The A3000 had a new (as in 'new chips') native VGA resolution and Amber. The purpose of Amber was was to make the standard Amiga resolutions compatible with VGA. The A2000 already had this since 1998 via Microway's 'AGA 2000' flicker fixer, which was a popular addon for that machine. Therefore it made sense for Commodore to add it into the A3000.

Nope. "No new chips" context is with Amiga's graphics chipset capabilities.

http://www.bambi-amiga.co.uk/amigahistory/mikesinz.html
Mike Sinz: 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")


A1200's Budgie is a cost-reduced Buster and Ramsey.

Gayle replaced Gary/Fat Gray while adding IDE and PCMCIA functions.

Quote:

What didn't make sense was putting all those new chips into a machine that would be far too expensive for the average home computer user. Commodore's engineers had become infected with the same desire for high-end toys that made the Los Gatos team go off the rails.

Your argument is made without context. Refer to Mike Sinz's statement.


When US inflation is factored in, the AA1000+ $800 target in 1991 is close to A500's $699 in 1987.

$699 in 1987 is $837.83 in 1991 or $921.25 in 1993.

When 486SX-25 and S486X33 PC clones entered $1000 price range in 1993, it was game over for the 1987's $699 product.

3DO's $699 in 1993 is $549.81 in 1987.

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

Add 33% for an Australian price estimation.

"Your mileage may vary with weaker currencies."

Quote:

An A1200 with popular 50MHz 030 accelerator card matched the performance of a 386DX-40 (the fastest 386 you could buy). By the time Doom was ported to the Amiga the price of these cards had come way down to the point where many Amiga fans could afford them. That wasn't the case in late 1993 when Doom came out, but then a 386DX system was also very expensive compared to the A1200.

386DX-40 class compute power with 50 Mhz 68030 is too late for the Amiga platform when PC has moved into Pentium class and PS1 game experiences.

"Economies of scale" factors are real.


Quote:

It would not be reasonable to expect Commodore to provide 386DX-40 performance on a machine costing much less.

For 1992's wholesale price, 68EC020-25 is considerably cheaper than Am386DX-40.


https://archive.computerhistory.org/resources/access/text/2013/04/102723262-05-01-acc.pdf
Page 119 of 981

For 1992
68000-12 = $5.5
68EC020-16 PQFP = $16.06, it's $15 in 1993 Q1.
68EC020-25 PQFP = $19.99, it's $18 in 1993 Q1.

68EC030-25 PQFP = $35.94

68030-16 CQFP = $70
68030-25 CQFP = $108.75

68040-25 = $418.52
68EC040-25 = $112.50
---
Competition

AM386-40 = $102.50
386DX-25 PQFP = $103.00

486SX-20 PQFP = $157.75
486DX-33 = $376.75
486DX2-50 = $502.75


Commodore UK MD argued for an upgraded CD32 with minimal cost increase.

Commodore CEO has milked the cow too hard.


Quote:

That applied to the SNES too. It couldn't run Doom. Even after adding a Super FX 2 chip in the cartridge they still had to dumb the game down.

SNES has packed pixels with Mode 7 and Mode 7 Direct Color. SNES has one less problem.

Quote:

But an accelerated A1200 runs the original Doom perfectly, providing an identical gaming experience to the PC.

AGA's small install base problem says Hi.

Gaming PC has superior timed exclusives with new 32-bit 2.5/3D gaming experiences that are beyond SNES's 16-bit strong 2D gaming experience.

Fast 32-bit gaming PC didn't directly compete against SNES's 16-bit 2D gaming experience since the PC offered the "new 32-bit 2.5/3D gaming experience".

Quote:

Indeed. The very idea of a fully fledged computer matching the price of a game console is ludicrous.

SNES had the bulk of A500/A1200 gaming use cases.

It shows that the Amiga's non-gaming market wasn't able to sustain Commodore.

Commodore Germany's IDE demand for A600 was a sales flop and it tanked Commodore's revenues with it.

https://youtu.be/5kaBIMuW74Q?t=115

Quote:

32 bit actually. :) It gives the consumer a machine capable of running a greater variety of game genres that appeal to a wider audience. And it gives them the ability do other computing-related tasks on it too - write a letter, paint a picture, compose some music etc., just like people with 'real' computers do. Maybe even get a modem and join the social media of the day (bulletin boards, newsgroups, irc...). Finally it give them access to all those interesting demos and a huge library of (ahem) 'free' software - the real reason most people bought an Amiga!

Amiga's non-gaming market wasn't able to sustain the Commodore.

"We don't care about games" - Bernd Meyers. Amithlon... enough said.

Doom/Doom 2 has 4 million sales with an estimated 15 million copies floating i.e. show the Jolly Roger.

Microsoft allowed Windows 3.x and 95 to be shown with Jolly Roger.

Quote:

Not really. The SNES didn't hit the US until mid 1991, and the UK until 1992. Amiga 500 games were still getting better even as late as 1994. Some were so good that it was hard to tell they weren't AGA!

What you don't get is SNES was building its install base earlier than AGA's Q4 1992. This is important for mainstream 3rd party game developers.

Without the IDE directive (demand by Commodore Germany), the A500 and AA3000+ could have been released with AGA in Q4 1991. AA3000+ has the DSP.

IDE-equipped AGA Amigas could be released later.

A500 AGA with 68EC020 would need Fat Gary (Gayle in A1200) and a cut-down Super Baster(just a side single connector, Budgie in A1200) and Ramsey (Budgie in A1200). A500 AGA would have no PCMCIA and IDE to blow up the price which needed A1200's Budgie's cost reduction method.


Quote:

But the NES was already entrenched in the US before then, despite not being that great. US customers weren't not buying the A500 because it wasn't as good as a console, but because it cost a lot more and wasn't a console. In the US 'computer' meant PC, and 'games machine' meant 'not a computer'. The Amiga didn't fit most people's preconceptions - and those that it did fit already had a C64!

SNES had the bulk of A500/A1200 gaming use cases.


Quote:

Perhaps, but DSP was still not the answer IMO. The Atari Falcon had DSP, what good did that do it?

1. Falcon's DSP is a fixed point 24-bit integer with about 16 MIPS INT24.

"13000-14000 Falcons sold with an additional 4164 units in stock is the last official number"

DSP3210 @50 Mhz has 12.5 MIPS INT32 and 25 MFLOPS FP32. DSP3210's 32-bit datatype is common with the host CPU.

When compared to 32-bit integers and 32-bit floating-point datatypes, a fixed point 24-bit integer for 3D is not wise for cross-platform game engine development.

The baseline Atari Falcon has 1 MB RAM and 68030-16 is not cheap when compared 68EC020-16 or 68EC020-25.

68030's MMU is a premium feature for Motorola's product segmentation.

For 1992,
68EC020-16 PQFP = $16.06, it's $15 in 1993 Q1.
68EC020-25 PQFP = $19.99, it's $18 in 1993 Q1.

68EC030-25 PQFP = $35.94

68030-16 CQFP = $70
68030-25 CQFP = $108.75

The component selection for Atari Falcon is not wise. Atari's MIDI music niche wasn't able to sustain the 68K TOS platform and results are worse than Amiga's Video Toaster niche.

2. Atari Falcon's 16-bit front side bus for 68030 @ 16 Mhz was mocked.

I remembered when PC gamers mocked this next-generation "16-bit" Atari garbage.

The Amiga press mocked Falcon's "16-bit bus" with the CPU. Atari tried to return with a "64-bit" Jaguar (between 125,000-250,000 units).

3DO has a better record and it's just the competition from Saturn (9.26 million unit sales) and PS1 are stronger.

Last edited by Hammer on 22-Apr-2024 at 01:30 AM.
Last edited by Hammer on 22-Apr-2024 at 01:24 AM.
Last edited by Hammer on 22-Apr-2024 at 01:20 AM.
Last edited by Hammer on 21-Apr-2024 at 02:06 PM.
Last edited by Hammer on 21-Apr-2024 at 01:58 PM.
Last edited by Hammer on 21-Apr-2024 at 01:56 PM.
Last edited by Hammer on 21-Apr-2024 at 01:35 PM.
Last edited by Hammer on 21-Apr-2024 at 12:42 PM.
Last edited by Hammer on 21-Apr-2024 at 12:38 PM.
Last edited by Hammer on 21-Apr-2024 at 12:37 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)

@bhabbott

Quote:

bhabbott wrote: The purpose of buying the Amiga was to have a 16 bit architecture, because that's the way the market was going. It was always intended that the original Amiga team would be 'disbanded' after the project was done. However they were kept on to provide diversity in developing the next models.

Commodore had an in-house 16-bit C900 and to be manufactured by Commodore West Germany.

https://en.wikipedia.org/wiki/Commodore_900
The ECS's low color and "business" high-resolution mindset are shown in this project.


Quote:

During this time the Los Gatos Amiga team's aspirations changed. They were working on a cost-reduced A1000, but wanted to produce a more advanced (and more expensive) machine that would compete in the workstation market. They saw the A500 as competition against their A1000CR, and the A2000 as competition against their high-end 'Ranger' system. In both cases however Commodore's other teams' designs won because they were more practical. The Los Gatos team had become redundant. IMO that's how it should have turned out. The A1000 was a nice concept, but the A500 was the machine which was needed to popularize the Amiga, and the A2000 was the right 'workhorse' for professional and industrial applications.

From 1987 to 1991, A2000's addons didn't benefit Amiga's general-purpose graphics.

A1500/A2000's numbers remained in the several thousand. e.g. 124,500 units in Germany.

A1500/A2000 didn't develop Amiga's RTG market's economies of scale. The Video Toaster niche wasn't able to sustain Commodore.

Last edited by Hammer on 21-Apr-2024 at 01:46 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)

@Hammer

Quote:

Hammer wrote: @bhabbott Quote: No. The A3000 had a new (as in 'new chips') native VGA resolution and Amber. The purpose of Amber was was to make the standard Amiga resolutions compatible with VGA. The A2000 already had this since 1998 via Microway's 'AGA 2000' flicker fixer, which was a popular addon for that machine. Therefore it made sense for Commodore to add it into the A3000. Nope. "No new chips" context is with Amiga's graphics chipset capabilities.

ECS was new for the A3000. Amber was part of the graphics chipset in the A3000. If the A3000 was commanded to have 'no new [graphics] chips' then it would have been OCS.

The truth is, the 'no new chips' directive referred to getting the design finalized without further delays, which there would have been if they had added more features. So what do you want - ECS, or nothing? AAA was in development for years and they couldn't get it working properly. CSG engineers doubted that even AGA would be possible using the process they had, let alone the extra stuff people here are proposing.

Quote:

http://www.bambi-amiga.co.uk/amigahistory/mikesinz.html Mike Sinz: 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")

Mike Sinz can call the A3000 design 'radical' if he wants, but in reality it wasn't that different. Zorro-III was basically just Zorro-II with a multiplexed bus to get 32 bit. Super DMAC was just a DMA controller. RAMsey was just a DRAM controller. 25MHz 030 had already been used on the A2630 card (which also used ZIP RAM). So in what way was it a 'radical departure'?

Quote:

Quote: What didn't make sense was putting all those new chips into a machine that would be far too expensive for the average home computer user. Commodore's engineers had become infected with the same desire for high-end toys that made the Los Gatos team go off the rails. Your argument is made without context. Refer to Mike Sinz's statement.

The context is clear. They stuffed the A3000 full of expensive new chips, including the 2MB ECS Agnus and Amber. They had visions of it being used as a Unix workstation, but this was not the machine the market needed. They already had the A2000 (which the Video Toaster actually fitted into!), and the A2630 already had a 32 bit expansion socket onboard which they could have added a SCSI controller and graphics card to (just like GVP did with their G-Force 030). Or they could have gone 'radical' by using the bridgeboard slots as a 32 bit extension connector like VL bus did. Instead they gave us a crippled Zorro-III with less then 1/10th of the theoretical bandwidth, and a flaky DMA controller.

Quote:

When US inflation is factored in, the AA1000+ $800 target in 1991 is close to A500's $699 in 1987.

But they couldn't meet that target. It would come in at over $1000, and by then the A500 was much cheaper.

Quote:

When 486SX-25 and S486X33 PC clones entered $1000 price range in 1993, it was game over for the 1987's $699 product.

In 1993 the A1200 was selling for US$350 and the CD32 was US$250.

In 1991 I opened my shop selling Amigas and PCs. I stocked up on various PC models including a 486SX-25 with 8MB RAM. This was initially priced at NZ$5000 (~US$2500). It took 2 years to sell this machine because nobody needed that much power in 1991. In 1993 the most popular machines were 386SXs ('low-end') and 386DXs ('mid-range'). Only rich people bought a 486. This didn't just apply in New Zealand. Few people bought the 'cheap' 486 PCs advertised in magazines because they came with iffy components and poor support.

Quote:

3DO's $699 in 1993 is $549.81 in 1987.

1987 prices are irrelevant in 1993, 6 years later!

Quote:

386DX-40 class compute power with 50 Mhz 68030 is too late for the Amiga platform when PC has moved into Pentium class and PS1 game experiences.

Commodore was long gone by the time Pentium PCs became affordable. Those who had an A1200 bought them several years before, and would now be looking at upgrading their favorite machine to do more with it. This was much cheaper than buying a PC.

Quote:

Quote: It would not be reasonable to expect Commodore to provide 386DX-40 performance on a machine costing much less. For 1992's wholesale price, 68EC020-25 is considerably cheaper than Am386DX-40.

It's not just the CPU you have to consider. A 25MHz 020 is a waste when the rest of the system only runs at 14MHz. Some have suggested overclocking to 28MHz for synchronous operation, but this is a no-no. if for some reason they were flaky, Motorola would say "Your fault, not ours. No warranty!".

The A1200 had to be £399 because that was a 'magic' price. Make it a bit more expensive and there will be a lot more buyer resistance. At that price Commodore was making 25% gross margin (same as I put on my computers).

It was only a bit more expensive than the A500, but considerably more powerful (and much neater too!). Considering that many Amiga fans stuck with their A500 anyway because it was good enough for them, having a slightly more powerful CPU wouldn't have made much difference. The A1200 was meant as a replacement for the A500, not another class of computer. And as you have pointed out, an Amiga selling for close to the price of a PC wouldn't fly - even if it had similar power.

Quote:

Commodore UK MD argued for an upgraded CD32 with minimal cost increase.

Commodore ended up selling the CD32 at cost price, and it was still at the upper end for a console. The 'minimal' cost increase you talk of wasn't. Besides, one of the attractions of the CD32 was that games for it would also run on the A1200 (with a CD-ROM drive). That wouldn't be the case if the CD32 was more powerful.

Quote:

Commodore CEO has milked the cow too hard.

On the contrary, they weren't milking it enough. They stopped production of the best-selling A500 too early because they thought the A1200 would be ready by then. They did everything they could to keep the price down so customers would get a good deal - rather than being 'milked' like the PC industry was doing, forcing people to upgrade every 2 years in order to have a powerful enough machine to run the latest games etc.

Quote:

Quote: That applied to the SNES too. It couldn't run Doom. Even after adding a Super FX 2 chip in the cartridge they still had to dumb the game down. SNES has packed pixels with Mode 7 and Mode 7 Direct Color. SNES has one less problem.

'One less problem' wasn't nearly enough to make up for its deficiencies.

Quote:

AGA's small install base problem says Hi.

Do I care? 30 years later you can still get an A1200 for a reasonable price, whereas other models like the A1000, A3000, CDTV, A4000 and CD32 are going for stupid prices. Pop in a PiStorm for a few hundred bucks and you have insane performance that you couldn't buy for a million dollars in 1993.

Quote:

SNES had the bulk of A500/A1200 gaming use cases.

I dispute that. But the A500 was generally bought in preference to a game console because it was a lot more. Parents bought one because they thought it would educate the kids - which in many cases it did. Hobbyists bought them because they wanted to do more than play mindless arcade games. Those who were only interested in games bought it for the huge library of 'free' software available (total cost of ownership much cheaper than an SNES!).

Quote:

It shows that the Amiga's non-gaming market wasn't able to sustain Commodore.

That certainly wasn't the case here. People weren't just buying it to play games!

Quote:

Commodore Germany's IDE demand for A600 was a sales flop and it tanked Commodore's revenues with it.

No. Stopping production of the A500 and not introducing the A1200 were the causes. Disappointment at getting the A600 instead of the promised A1200 was what tanked interest among fans. But existing Amiga owners weren't the target market. Once it hit the shops it actually sold well, which shows that the idea of a smaller low cost Amiga was good. It just shouldn't have been the only choice (and wouldn't have been if the A1200 had arrived on time).


Quote:

"We don't care about games" - Bernd Meyers. Amithlon... enough said.

Games were a big part of the Amiga, but not all of it. Many Amiga owners spent more time using it for other stuff. Same thing applied to other home computers, including the PC. Difference on the Amiga was you got great games and an advanced multitasking GUI OS that was a joy to use for much less than a PC - and a lot less frustrating too!