@Hammer

Quote:

Hammer wrote: @agami --- Wall of data ---

Yes. Evidence that they didn't keep up.
Not that they couldn't.

_________________
All the way, with 68k

@agami

Quote:

agami wrote: Yes. Evidence that they didn't keep up. Not that they couldn't.

"They could" argument is a "What IF".

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

@Gunnar

Quote:

Gunnar wrote: @cdimauro Quote: Again? It took SEVEN years for the AGA, and it was crap! No, Commodore did not work on AGA for 7 years. The Amiga team worked on other ideas, AAA for example. As the other project were delayed, not ready, cancelled, they cooked AGA "Spaghetti" for us in 10 minutes. They not worked 7 years on AGA. They had limited time only. If you have a limited time, then you need to focus. You can only do "so many" features in this time. And every feature that you try to do has a risk to fail - and the whole project might then fail. AGA has significant improvements over OCS. And they limited the features to the amount that they could do, and could debug. This was a good decision. Its better to have a working and released chipset with 5 new improvements, than an non working non released with 10 new features.

AA project was started during September 1989.

From Commodore - The Final Years
Quote:

AA First Prototypes Back in September 1989, George Robbins proposed an intermediate level Amiga chipset, called Pandora, to bridge the gap until AAA appeared. The chipset, consisting of a graphics chip called Lisa and an improved Agnus called Alice, was supposed to be ready in early 1990. Rubin had given the go ahead and six chip engineers had been working full time under the project leader, Bob Raible.

Cdimauro's narrative is false.

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

@ppcamiga1

Quote:

ppcamiga1 wrote: @Karlos 060 was released after Commodore bankruptcy. So it is still valid Commodore bankrupt because AGA has not chunky pixel.

Motorola released 68060 on April 19, 1994. https://www.techmonitor.ai/technology/motorola_plans_to_sample_the_68060_next_quarter

Commodore International announced voluntary bankruptcy and liquidation on April 29, 1994.

Actual, 68060 was released before Commodore International's April 29, 1994 voluntary bankruptcy announcement.

Last edited by Hammer on 05-Sep-2024 at 08:34 AM.

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

@Hammer

motorola_plans_to_sample_the_68060_next_quarter

so 68060 was not released before c= bankruptcy
moto just anonsed that samples will be send in next quarter

@ppcamiga1
Both events in April are announcements.

Commodore's official statement follows:
Quote:

"Commodore International Limited announded today that its Board of Directors has authorized the transfer of assets to trustees for the benefit of its creditor and has placed its major subsidiary, Commodore Electronics Limited, into voluntary liquidation. This is the intial phase of an orderly liquidation of both companies, which are incorporated in the Bahamas, by the Bahamas Supreme Court." "This action does not affect the wholly-owned subsidiaries which include Commodore Business Machines (USA), Commodore Buisiness machines LTD (Canada), Commodore/Amiga (UK), Commodore Germany, etc. Operations will continue normally."

Commodore International was defunct around May 6, 1994.

https://bs.vlex.com/vid/re-commodore-international-ltd-792694421
Osadebay, J. (Acting): On the 24th June, 1994 I made an Order for the Winding up of the “‘Commodore International Limited, under the provisions of the Companies Act, 1992.

Commodore Electronics Limited (C.E.L) is a subsidiary of C.I.L and is also a Company incorporated and existing under the laws of The Bahamas and has its Registered Office in Nassau, New Providence, Bahamas. C.I.L. is the parent of a multinational group of companies engaged in the manufacturing and selling computers in many parts of the world. By a Resolution of the Company passed on April, 29th, 1994, C.E.L. went into a Voluntary Winding — up (i.e. Voluntary Liquidation). On 2nd of May, 1994, C.E.L. filed a petition with the Supreme Court of the Bahamas requesting that the winding — up be continued under the supervision of the Court. This petition was heard on the 31st May, 1994, and C.E.L's liquidation was ordered to be continued under the supervision of the Supreme Court of The Bahamas. C.E.L therefore is presently in liquidation under the supervision of the Court.

-----------------------------

Subsidiaries Commodore UK and Commodore BV (Netherlands) survived Commodore International's bankruptcy.

Commodore BV (Netherlands) dissolved in early 1995.

Commodore UK went into liquidation on August 30, 1995.

Escom AG declared bankruptcy on 15 July 1996.

Last edited by Hammer on 06-Sep-2024 at 12:32 AM.
Last edited by Hammer on 06-Sep-2024 at 12:29 AM.
Last edited by Hammer on 06-Sep-2024 at 12:19 AM.

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

@Hammer

Quote:

Hammer wrote: @Gunnar Quote: Gunnar wrote: @cdimauro No, Commodore did not work on AGA for 7 years. The Amiga team worked on other ideas, AAA for example. As the other project were delayed, not ready, cancelled, they cooked AGA "Spaghetti" for us in 10 minutes. They not worked 7 years on AGA. They had limited time only. If you have a limited time, then you need to focus. You can only do "so many" features in this time. And every feature that you try to do has a risk to fail - and the whole project might then fail. AGA has significant improvements over OCS. And they limited the features to the amount that they could do, and could debug. This was a good decision. Its better to have a working and released chipset with 5 new improvements, than an non working non released with 10 new features. AA project was started during September 1989. From Commodore - The Final Years Quote: AA First Prototypes Back in September 1989, George Robbins proposed an intermediate level Amiga chipset, called Pandora, to bridge the gap until AAA appeared. The chipset, consisting of a graphics chip called Lisa and an improved Agnus called Alice, was supposed to be ready in early 1990. Rubin had given the go ahead and six chip engineers had been working full time under the project leader, Bob Raible. Cdimauro's narrative is false.

Unfortunately, the problem here is another: it's Nature which was an evil step mother with you.

In fact, here is the context: https://amigaworld.net/modules/newbb/viewtopic.php?topic_id=44362&forum=2&start=220&viewmode=flat&order=0#855859

I think AGA was really an improvement over OCS.

It's for sure. However it was very modest. After 7 (SEVEN) years...

which is where all this part of discussion started. Read AGAIN: that's the context.

However, you aren't able to follow the discussions, as you've greatly proved, and try to answer like a bot which don't know the context: by catching words and blindly replying on them...

Yes, bot is the perfect words that describes you. Bots don't need to be smart, in fact: they just mechanically write things...

@Hammer

Quote:

Hammer wrote: @agami Quote: agami wrote: Yes. Evidence that they didn't keep up. Not that they couldn't. "They could" argument is a "What IF".

If I were indeed making a "they could" argument, then I would have to provide evidence to support such an argument. But I wasn't making that claim.

What I was pointing out is that all the evidence you've presented only speaks to the fact they they "didn't keep up", and your inference from that evidence that "they couldn't" is a post hoc fallacy.

If you wish to claim that Commodore et al "couldn't keep up", then you need to present evidence speaking to the capabilities of individuals and other corporate and commercial conditions which could be barriers.

A series of recorded instances only ever showing a person walking, is not evidence of their inability to run.

_________________
All the way, with 68k

7 years is how long ago it was announced by Gunnar that SAGA will be open source!

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

Quote:

kolla wrote: 7 years is how long ago it was announced by Gunnar that SAGA will be open source!

But how accurate that statement was?

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

@NutsAboutAmiga

On Amiga level it was vast improvement especially on hi res where jump from 16 to 256 cols is evident .
Also ham8 makes better conversion of true colour images then ham6, while keeping reduced file size.
In Amiga world its almost revolutionary as first ecs minus weak 020ec to complement it, no fast ram as standard and no sound improved.

Outside of it it wasnt that special as ocs ait its time. PC and consoles could do more and were moving beyond 256 colours and to 16 bit sound and cd rom.

If ecs was skipped and aga released earlier, while something 16 bit sound and beyond released in 1992 it could keep Amiga on top, but as we know, CBM was unable to do that and kept economical approach vs innovation, which killed them and Amiga ...

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

@vox

Doesn't this belong to the other thread? :p

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

@kolla

On vanp

And we know they will keep both saga and licensing 080 to them, as they tend to milk the cow. Only pistorm kills it as more affordable fast accel

On other unkept promises was gold3 limited for v2 users, all sound/hdmi redirected for v2 ...

After all, split with majsta made apollo team lose right to vampire name, hence various creature names for v4 cards.

Implementation is interesting, aga reverse engineered and improved, as 080 seems to be quite corrected and faster 060. Too bad only such hard ass owneelr has it, with little hope to spread and mass the improvement.

Like post cbm cbm2 or almost commodore USA

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

@pixie

Quote:

pixie wrote: @kolla Quote: kolla wrote: 7 years is how long ago it was announced by Gunnar that SAGA will be open source! But how accurate that statement was?

It is. He has written it on the Apollo forum very long time ago, but yet there's no public repo where it has pushed the SAGA sources.

kolla Quote:

7 years is how long ago it was announced by Gunnar that SAGA will be open source!

Which SAGA?

1. Original SAGA written by Thomas Hirsch in AHDL?
2. SAGA Gunnar converted to VHDL?
3. SAGA Gunnar obtained from Thomas Hirsch before he joined the Apollo Team?
4. SAGA Gunnar has supposedly rewritten completely without any credit given to Thomas Hirsch?

https://amigaworld.net/modules/newbb/viewtopic.php?topic_id=45221&forum=25&start=120&134#870436

Only Amiga makes it impossible. Why does Amiga get all the mental cases? Did Amiga cases use lead paint?

@Gunnar

Quote:

Gunnar wrote: @Hammer Quote: Commodore created two 256-color display capable chipsets i.e. C65 (completed in December 1990) and AGA (completed in March 1991). Commodore made spaghetti and marinara pasta. Where these the same developers? Big companies have many teams working on different solutions in different areas.

Double tasks on the same engineers have occurred.

From Commodore - The Final Years,
Quote:

Linda, Monica, and Mary were supposed to be done in the summer of 1989, but by September they were not even close. It became clear the AAA chipset was not running on schedule. “It was just very complex,� explains Ed Hepler. “Complex chips like that take two or three years to develop and compared to other companies, we had a very limited staff working on things.� Engineer Bob Raible soon noted the project was “slipping at a nice predictable rate of one week per week.� AAA was to join the ECS, Amiga 3000 and C65 as yet another project where the deadline was constantly slipping. Something was happening in Commodore. Either all of the management, previously capable, decided all at once to allow their projects to fall hopelessly behind, or there was a root cause. Most engineers felt they knew the root and the solution. Commodore needed an HR department, and it needed to hire more engineers for the vital project. “The AAA chipset had four chips, and there was really only one person working on each one as far as the main architect,� explains Hepler. “I don't have a problem with that, but we could have used some extra support helping to simulate and so forth.� Although Victor Andrade was supposed to have been working on AAA all year, he had instead been working on the 4510 chipset until the middle of October 1989, at which time he began on AAA. By then it was clear they could not meet the early 1990 schedule for working silicon. With only two engineers per chip (and not dedicated ones at that) it would be impossible to meet the schedule. “There is usually one designer, one circuit designer, and one layout person per chip and that's sort of how things went,� says Hepler. “If you look at the way companies build chips these days, there are far more people working on a chip than back then.�

------------

Continuing from Commodore - The Final Years,
Quote:

Unix Resurrection This obsession was driven by Gerard Bucas along with engineers Bob Welland and George Robbins, who saw the A500 as the ultimate in low-cost Unix platforms. According to coworker Bil Herd, (skip) Ominously, Commodore's engineering group jumped from 49 engineers in 1987 to 86 engineers in 1988. It was the C900 project all over again, one that had proven costly and helped bring Commodore close to its destruction. “That was definitely a failure of management, one that I didn't fully appreciate at the time, but in retrospect I can see it was just fatal,� says Nesbitt.

Commodore's master of none and the lack of focus R&D.

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

Quote:

@cdimauro A 32-bit Blitter was needed and was feasible. Even a 64-bit one, considered that the fetch was extended to 64-bit However if Commodore smart engineers didn't wanted to improve the Blitter this way, they had a very simple way to improve its speed: increase its clock frequency. Providing an equivalent (and different!) fmode register for selecting the frequency to keep the backward-compatibility.

AAA and AA's timeline is shown in "Commodore the Final Year" book. The blame is on management i.e. blame Henri Rubin for Amiga's lost 1986 to 1988 years.

Henri Rubin is instrumental in removing Thomas Rattigan.

In mid-1986, Henri Rubin ordered the original Amiga team to design the #metoo monochrome high-resolution Denise and canceled the Amiga Ranger project. Henri Rubin's #metoo R&D directive was flawed when IBM released VGA and 8514 in 1987. R&D time was wasted on monochrome high-resolution Denise.

The Commodore LSI team blames the monochrome high-resolution Denise debacle on the original Amiga team when the real blame is on Henri Rubin. This corporate politics lead to the original Amiga team's removal.

Jeff Porter argued for 8 bitplanes with 16 million colors in 1987 which is effectively AA R&D project that started in September 1989.

Henri Rubin's removal is due to Ali's closer power nexus relationship with Irving Gould.


Quote:

@cdimauro I think that from 7 to 28Mhz in 7 (SEVEN) years should have been possible. If you take into account how much the chips frequencies increased over that long period.

Without a cache or memory interleave, it depends on the selected memory performance.

Game console like NeoGeo has a high-performance 84 KB VRAM which is enough for 320x224x8 resolution games. Also applicable for small VRAM-equipped Sega Mega Drive and SNES. These game consoles are designed for a single purpose.

Amiga's general-purpose desktop computer design would require larger VRAM e.g. Super A500 has expensive 1 MB VRAM as general-purpose Chip RAM. A3000's VRAM configuration has 2MB VRAM as Chip RAM.

IBM 8514's 512 KB VRAM is not general purpose like Amiga's Chip RAM.

The solution is a hybrid UMA design as shown by 3DO's MADAM (Agnus/Alice role) which has DRAM and VRAM access. Hybrid UMA is used in Xbox Series X i.e. 10 GB 320-bit bus and 6 GB 192-bit bus.

For 1989 released ET4000AX, Tseng Labs implemented memory interleave to lower latencies into "VRAM-like" (claimed by Tseng Labs) with normal FP DRAM.

If VRAM is expensive, fast and smaller VRAM should used as a video cache for the slower and larger DRAM.

Current PC GPUs have a multi-MB on-chip cache against the slower and larger GDDR6/GDDR6X video memory.

Amiga OCS's 3.5 Mhz is based on 260 ns read/write cycle DRAMs. AA Lisa is designed for 140 read/write cycle DRAMs.

Typical DRAM "access times" marketing like "80 ns" don't show read/write cycle times.

IBM 8514's VRAM has 40 ns sequential (serial) access time and pixels read/write workloads are usually sequential. 3DO's VRAM has 20 ns sequential (serial) access time.

40 ns sequential access translates into 25 Mhz effective.

20 ns sequential access translates into 50 Mhz effective. 3DO's MADAM (Agnus/Alice role) and CLIO (Lisa/Denise role) custom chips have about 24.54 Mhz each.

3DO's downside is the CPU selection i.e. ARM60 @ 12.5 Mhz and crap MUL instruction implmentation, hence CPU PIO-driven matrix math co-processor in the MADAM chip. ARM60 @ 12.5 Mhz is effectively 68030 @ 40 Mhz class.

CPU PIO-driven matrix math co-processor is the original Amiga engineers' rush job due to compressed R&D time.

Jeff Porter's MIPS-X CPU @ 40Mhz from CL-450 SoC selection has strong MUL instruction implementation.

For game consoles and workstations from 1990 to 1996, MIPS CPU beats ARM CPUs. Sony's PlayStation for the win.

My point, a RISC-based 28 Mhz graphics co-processor without cache would need memory technology faster than A1200/A4000's 140 ns read/write cycle FP DRAM.

Last edited by Hammer on 09-Sep-2024 at 03:57 AM.
Last edited by Hammer on 09-Sep-2024 at 02:56 AM.
Last edited by Hammer on 09-Sep-2024 at 02:40 AM.
Last edited by Hammer on 09-Sep-2024 at 02:34 AM.
Last edited by Hammer on 09-Sep-2024 at 02:28 AM.

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

@agami

Quote:

agami wrote: If I were indeed making a "they could" argument, then I would have to provide evidence to support such an argument. But I wasn't making that claim. What I was pointing out is that all the evidence you've presented only speaks to the fact they they "didn't keep up", and your inference from that evidence that "they couldn't" is a post hoc fallacy. If you wish to claim that Commodore et al "couldn't keep up", then you need to present evidence speaking to the capabilities of individuals and other corporate and commercial conditions which could be barriers. A series of recorded instances only ever showing a person walking, is not evidence of their inability to run.

That's crap. "Commodore - The Final Years" book should be a mandatory textbook for this topic.

Last edited by Hammer on 09-Sep-2024 at 03:51 AM.

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

@Hammer

Quote:

Hammer wrote: Quote: @cdimauro A 32-bit Blitter was needed and was feasible. Even a 64-bit one, considered that the fetch was extended to 64-bit However if Commodore smart engineers didn't wanted to improve the Blitter this way, they had a very simple way to improve its speed: increase its clock frequency. Providing an equivalent (and different!) fmode register for selecting the frequency to keep the backward-compatibility. AAA and AA's timeline is shown in "Commodore the Final Year" book. The blame is on management i.e. blame Henri Rubin for Amiga's lost 1986 to 1988 years. Henri Rubin is instrumental in removing Thomas Rattigan. In mid-1986, Henri Rubin ordered the original Amiga team to design the #metoo monochrome high-resolution Denise and canceled the Amiga Ranger project. Henri Rubin's #metoo R&D directive was flawed when IBM released VGA and 8514 in 1987. R&D time was wasted on monochrome high-resolution Denise. The Commodore LSI team blames the monochrome high-resolution Denise debacle on the original Amiga team when the real blame is on Henri Rubin. This corporate politics lead to the original Amiga team's removal. Jeff Porter argued for 8 bitplanes with 16 million colors in 1987 which is effectively AA R&D project that started in September 1989. Henri Rubin's removal is due to Ali's closer power nexus relationship with Irving Gould.

Already discussed about it. And already answered.

Hint: DISCUSSIONS about how to enhance the Amiga with 256 colours from 16 million colours started on September 1987 (after Apple's Mac II and IBM's VGA), and this requires time. AFTER the technical discussions (which are needed), THEN the implementation is needed, which requires time as well.
Quote:

Quote: @cdimauro I think that from 7 to 28Mhz in 7 (SEVEN) years should have been possible. If you take into account how much the chips frequencies increased over that long period. Without a cache or memory interleave, it depends on the selected memory performance.

The topic in this case was the BLITTER. Which needs a very small cache ONLY for the planar graphics, to keep the mask without reloading it each time for each bitplane to be processed.

Besides that, no cache at all is needed for Blitter operational activities.
Quote:

Game console like NeoGeo has a high-performance 84 KB VRAM which is enough for 320x224x8 resolution games. Also applicable for small VRAM-equipped Sega Mega Drive and SNES. These game consoles are designed for a single purpose. Amiga's general-purpose desktop computer design would require larger VRAM e.g. Super A500 has expensive 1 MB VRAM as general-purpose Chip RAM. A3000's VRAM configuration has 2MB VRAM as Chip RAM. IBM 8514's 512 KB VRAM is not general purpose like Amiga's Chip RAM. The solution is a hybrid UMA design as shown by 3DO's MADAM (Agnus/Alice role) which has DRAM and VRAM access. Hybrid UMA is used in Xbox Series X i.e. 10 GB 320-bit bus and 6 GB 192-bit bus. For 1989 released ET4000AX, Tseng Labs implemented memory interleave to lower latencies into "VRAM-like" (claimed by Tseng Labs) with normal FP DRAM. If VRAM is expensive, fast and smaller VRAM should used as a video cache for the slower and larger DRAM.

VRAM = professional and very expensive graphic cards -> Not needed for the Amiga.

DRAM was enough 'til '92. After that, VRAM could have been considered, because their cost greatly reduced -> an option for the Amiga.
Quote:

Current PC GPUs have a multi-MB on-chip cache against the slower and larger GDDR6/GDDR6X video memory.

Irrelevant, off-topic: time ranges 'til '94 (Commodore's bankrupt).
Quote:

Amiga OCS's 3.5 Mhz is based on 260 ns read/write cycle DRAMs. AA Lisa is designed for 140 read/write cycle DRAMs.

And the chipset made use of the Fast Page mode (AKA: read the data from the same opened RAM page) -> double the bandwidth.
Quote:

Typical DRAM "access times" marketing like "80 ns" don't show read/write cycle times. IBM 8514's VRAM has 40 ns sequential (serial) access time and pixels read/write workloads are usually sequential. 3DO's VRAM has 20 ns sequential (serial) access time. 40 ns sequential access translates into 25 Mhz effective. 20 ns sequential access translates into 50 Mhz effective. 3DO's MADAM (Agnus/Alice role) and CLIO (Lisa/Denise role) custom chips have about 24.54 Mhz each.

Only the graphic chip (Denise, Lisa, etc.) needs to work at higher (usually double) clock frequencies compared to the remaining chips, to generate the proper video signals.

The remaining could have worked at less frequencies (half), because their activities are purely dominated by the RAM bandwidth.
Quote:

3DO's downside is the CPU selection i.e. ARM60 @ 12.5 Mhz and crap MUL instruction implmentation, hence CPU PIO-driven matrix math co-processor in the MADAM chip. ARM60 @ 12.5 Mhz is effectively 68030 @ 40 Mhz class.

Not needed: Amiga had a custom chipset to which delegate many important activities.
Quote:

CPU PIO-driven matrix math co-processor is the original Amiga engineers' rush job due to compressed R&D time.

The DSP? It was NOT needed, as already proved. More to come on the next article.
Quote:

Jeff Porter's MIPS-X CPU @ 40Mhz from CL-450 SoC selection has strong MUL instruction implementation.

Same here.
Quote:

For game consoles and workstations from 1990 to 1996, MIPS CPU beats ARM CPUs. Sony's PlayStation for the win.

1995..1996 -> outside of the context.
Quote:

My point, a RISC-based 28 Mhz graphics co-processor without cache would need memory technology faster than A1200/A4000's 140 ns read/write cycle FP DRAM.

Your point is very well known, and the answer is always the same: NOT needed.

The Amiga had already its custom chipset to deal with the graphics and all other things. Chipset which needs to be upgraded, of course, and you don't necessarily need to go for Commodore's LSI group for keeping the costs down (see CD32's cost for the chipset: a few dollars).

@cdimauro

Quote:

@cdimauro Already discussed about it. And already answered. Hint: DISCUSSIONS about how to enhance the Amiga with 256 colours from 16 million colours started on September 1987 (after Apple's Mac II and IBM's VGA), and this requires time. AFTER the technical discussions (which are needed), THEN the implementation is needed, which requires time as well.

Wrong. Hint: the R&D implementation's "go ahead" from Henri Rubin is required e.g. Henri Rubin's AA R&D implementation approval is during Sep 1989.

Jeff Porter can argue for "8 bitplanes with 16 million colors" in 1987 which amounts to nothing burger until Henri Rubin's R&D implementation approval in Sep 1989.

Quote:

@cdimauro VRAM = professional and very expensive graphic cards -> Not needed for the Amiga. DRAM was enough 'til '92. After that, VRAM could have been considered, because their cost greatly reduced -> an option for the Amiga.

Wrong narrative.

FP DRAM was enough when the implementation included multiple memory controllers with memory interleave e.g. ET4000AX to W32 series.

Commodore didn't master memory interleave, hence the VRAM direction.

1988-era Super A500's expensive issue was mixed up with $77 68020-16 along with $100 1 MB VRAM. Commodore tolerated the $50 cost for A1200's FP DRAM (cite: Commodore_Post_Bankruptcy pdf).

Game consoles selected small VRAM storage for aggressive cost control while delivering VRAM performance at certain game resolution targets.

VRAM sizes
Nintendo SNES = 64 KB
Sega Mega Drive = 64 KB

2D game consoles such as SNES, Mega Drive, and Neo Geo implement discrete video memory bus separated from system memory bus.

Stock Amiga 500's UMA already lost against the mentioned game consoles on memory bandwidth.

Quote:

@cdimauro Irrelevant, off-topic: time ranges 'til '94 (Commodore's bankrupt). when Commodore can tolerate $52 for A1200's 2MB FP DRAM.

It's relevant for hybrid UMA ideas for 1993 released 3DO (designed in 1992) and modern-day game consoles.

Like 3DO's MADAM, Amiga Hombre's Natalie has a hybrid UMA with FP DRAM and VRAM access.

Quote:

@cdimauro And the chipset made use of the Fast Page mode (AKA: read the data from the same opened RAM page) -> double the bandwidth.

Again, AA Lisa has access to around 140 ns read/write rated FP DRAM which is 7.1 Mhz effective. Lisa's 64-bit fetch is wrapped in two sequential 32-bit transfers.

For Alice, A1200's FP DRAM improves the 1985-era 16-bit Blitter.

Your counter-argument is meaningless since I already know about Fast Page DRAM mode.

Quote:

@cdimauro The DSP? It was NOT needed, as already proved. More to come on the next article.

Why the anti-DSP? Did you know the Amiga Blitter has a rudimentary ALU?

3DO's 3D DSP is an in-house matrix math co-processor design that is fed like Akiko's C2P i.e. ARM CPU writes to matrix math co-processor's registers and watches for competition status register. The lack of DMA for 3DO's matrix math co-processor is due to compressed R&D time.

DSP is a custom math chip for a target workload like any other custom chip.

3DO has no problems delivering arcade-quality strong 2D Super Street Fighter 2 Turbo.

When 3DO's matrix math co-processor @ 25 Mhz and texture mapper @ 25 Mhz hardware are used, 3DO can run Tomb Raider.

https://www.youtube.com/watch?v=vscj6y6oNac
Tomb Raider comparison between 3DO Vs PlayStation. The Tomb Raider port on 3DO is pushing to its limits. PlayStation still has faster results.

PlayStation's hardware was designed by mostly 3rd parties i.e. LSI (R3000A CPU, MIPS-based GTE matrix math) and Toshiba (GPU which includes texture mapper, raster, and display). Ken Kutaragi provided the leadership to combine hardware components and game developer relationships.

Amiga Hombre's Natalie Blitter IP was modified for texture mapping fill and raster operations processing. Custom PA-RISC with 3D extensions provides matrix math and game logic compute power.

For DSP3210, math libraries can abstract the math functions.

Quote:

@cdimauro 1995..1996 -> outside of the context.

Wrong. PSX was fully operational in December 1993 which shows Namco's Ridge Racer. Sony was waiting for the launch games in 1994 and 1995 releases.

Your retail release argument is NOT consistent with your arm-chair tech leadership. You're a hypocrite.

PSX's hardware IP and design was before December 1993's Namco's Ridge Racer demonstration for 3rd party game developers.

Last edited by Hammer on 10-Sep-2024 at 04:04 AM.
Last edited by Hammer on 10-Sep-2024 at 03:56 AM.
Last edited by Hammer on 10-Sep-2024 at 03:43 AM.

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