Was anyone else disappointed with the A1200?

[Thorham]

Originally Posted by Gorf

even then it is around 10GB ... the equivalent of 12.500 Workbench disks

Totally insane

Originally Posted by Gorf

For something you can't actually use for anything without additional software.

You can browse the internet

[TEG]

Regarding timeline and sales, one thing I don't understand about Commodore management, and it was at Tramiel time, is why they did not produced a C64 in a PC case style very early. I mean, when you look at the timeline, the C64 (1982) was outputted exactly 1 year after the IBM PC (1981), and 5 years after the apple II (1977). So the concept of a motherboard with optional daughter cards was here since long.

They had the experience of big cases with the PET series, it was not something new for them. So you have an architecture witch kick in the ass, why you don't try to use it against you competitor? Imagine a C64 in a PC case, having internals connectors, and with the bug which made the floppy drive so slowwww corrected. It would have looked good.

Now, was it too difficult to add slots to the C64 architecture? I don't know but having MOS Technology in their hands, they could have develop whatever glue chip was needed I guess. So your successful platform is present in several segments at a time when all is not yet decided. It's 1982, not 5 years later with the A500/A2000 and even less 10 with the A1200/A4000.

[hammer]

Originally Posted by TEG

Regarding timeline and sales, one thing I don't understand about Commodore management, and it was at Tramiel time, is why they did not produced a C64 in a PC case style very early. I mean, when you look at the timeline, the C64 (1982) was outputted exactly 1 year after the IBM PC (1981), and 5 years after the apple II (1977). So the concept of a motherboard with optional daughter cards was here since long.

They had the experience of big cases with the PET series, it was not something new for them. So you have an architecture witch kick in the ass, why you don't try to use it against you competitor? Imagine a C64 in a PC case, having internals connectors, and with the bug which made the floppy drive so slowwww corrected. It would have looked good.

Now, was it too difficult to add slots to the C64 architecture? I don't know but having MOS Technology in their hands, they could have develop whatever glue chip was needed I guess. So your successful platform is present in several segments at a time when all is not yet decided. It's 1982, not 5 years later with the A500/A2000 and even less 10 with the A1200/A4000.

Commodore tried with "big box" C64 https://en.wikipedia.org/wiki/Commodore_Educator_64

The problem is Commodore configured Educator 64 to be monochrome like the Apple II which is "zero sum" which wouldn't displace the incumbent. Educator 64 lacks the expansion slots of Apple II.

Commodore didn't update 65xx CPU family on par with Intel's 16-bit 8086, 16-bit 80286 (with MMU, 24-bit memory address) and 32-bit 80386. Acorn created ARM to address Commodore's 65xx CPU rubbish R&D road map. Commodore purchased the Amiga as a panic move.

Microsoft displaced incumbent text-based Lotus 123 and Word Perfect with Mac ports of MS Excel 2.2 and Word 2.0 for Windows 2.0.

[hammer]

Originally Posted by Thomas Richter

This is not really a defect.... Gayle *can* reset the PCMCIA port, but this is usually not done. The story is a bit longer.

The problem is that the PCMCIA port can/is also used as memory expansion. PCMCIA-cards can operate in two modes: As memory, or as I/O device. If you want to use it as memory device, it is important that the memory within it does not go away during a reset as this may crash the system if information such as interrupt vectors are placed in this RAM. Thus, one cannot simply reset the PCMCIA port during a hardware reset, and neither can SetPatch simply reset the port as it depends on what is in the port whether this makes sense or not.

Thus, this is not a bug, and certainly not an issue of the hardware.

It's a bug when it has "show stopper" behavior.

I already mentioned PCMCIA's "memory only" and "memory and I/O" modes in this topic.

I have a 3rd party AA-Gayle reset fix as mentioned in https://amitopia.com/gayle-reset-fix-amiga-1200/

Using PCMCIA as RAM for A1200's 32-bit 68EC020 is not recommended due to being 16-bit RAM, hence it's a pointless feature.

[hammer]

Originally Posted by Gorf

It was more like a "it's good enough" attitude. OSC/ESC could e.g. be easily upgraded to 5bit per colour to deliver 32K colors instead of 4K ... the bits in the 16bit wide register are simply unused.
Also an AA- (still on 16bit-Denise) could have been given the Double-CAS treatment to easily allow 8 bitplanes.

Amiga 3000 had the 32-bit Chip RAM with 44256k 120 ns access time.

280 ns read/write cycle translates into 3.57 Mhz. 32-bit Denise ECS+ would be required.

https://minuszerodegrees.net/memory/41256.htm
Hitachi HM51256P-12 (with 120 ns access time) has 210 ns read/write cycle.

GoldStar GM71C256A-80 (with 80 ns access time) has 145 ns read/write cycle translates into 6.89 Mhz.

GoldStar GM71C256A-70 (with 70 ns access time) has 130 ns read/write cycle translates into 7.69 Mhz.

VRAM in 1987 has 40 ns serial access. VRAM in 1990s has 20 ns serial access.

Something like GoldStar GM71C256A-70's would 130 ns read/write cycle allow for 16-bit 7 Mhz.

ET4000AX uses 70 ns to 80 ns access time DRAMs with memory interleave mastery. Tseng Labs claims VRAM like performance with memory interleave FP DRAMs.

[Thorham]

Originally Posted by hammer

Using PCMCIA as RAM for A1200's 32-bit 68EC020 is not recommended due to being 16-bit RAM, hence it's a pointless feature.

Indeed. Still good for a network, though.

[hammer]

Originally Posted by Thorham

Indeed. Still good for a network, though.

PCMCIA as a RAM expansion method has been displaced by SIMMs.

For single-slot PCMCIA equipped A600/A1200, swapping between PCMCIA RAM expansion and PCMCIA network card is not a good use case. SRAM PCMCIA RAM cards flopped in the PC market.

[Bruce Abbott]

Originally Posted by hammer

According to Lew, AAA had 1 year of serious R&D.

http://www.bambi-amiga.co.uk/amigahi...ggebrecht.html

Define 'serious'.

The AAA chipset project began in 1988. The spec for all 4 chips was completed in February 1989, with prototypes planned to be ready by July (Linda), August (Monica and Mary), and January 1990 (Andrea) - a very 'optimistic' time frame. That sounds pretty serious, but it was a much harder task than they expected. Despite having 8 engineers working on it, the schedule kept slipping. According to Ed Helper (one of the engineers working on Andrea), chips of that complexity usually took two to three years to develop.

AAA had a total of ~750,000 transistors spread over 4 chips, compared to only 80,000 transistors in AGA. Unlike AGA it wasn't just an enhanced ECS with much of the original intact (same blitter etc.), but a completely new design that still had to be 100% compatible. This was made doubly difficult by coders who (by accident or design) misused the chipset to get various effects.

Another problem is that a 1.25 micron process was needed, but CSG was already having trouble doing 2 micron (OCS was 5 micron). A $15 million investment was required for the new process, and it would take a while to get running smoothly. There was no urgency to get AAA finished sooner because they wouldn't be able to test it until they had prototype chips anyway.

Even if they had been 'serious' the whole time, AAA probably wouldn't have been ready before 1992, and it wouldn't have gone into low-end Amigas anyway due to its high cost. So while they might have made better progress on AAA by throwing more resources at it, that would have just compound the problem. If instead those 8 engineers had been working on AA from the start they might have been able to meet their tight schedule and have AA machines out in 1990!

Remember, Commodore had a "read my lips, no new chips" directive during A3000's R&D period.

That doesn't mean what what you are implying.

Legacy Amiga software wouldn't be aware of Lisa+'s extra burst modes, hence maintaining backward compatibility.

AAA scales to 64-bit VRAM. Tseng Labs ET4000 family wasn't crazy enough for AAA's 64-bit VRAM.

AAA was not compatible with AA. The reason for this is that AAA was developed separately (with ECS compatibility in mind) before AA existed.

PCMCIA got in the way.

No, it didn't.

The A300 project shouldn't have merged with the AA500 project.

It didn't. AA500 was abandoned in favor of the A500+ with ECS, to be followed by a cost-reduced version (A300) intended to be the last hurrah for the a500 line (like Sony's PS1 was for the PlayStation).

The A1200 was developed from the A600 because that was the fastest way to get the job done. The same engineer did both, so it was a doddle for him. Same goes for the case design. And by this time the factories were set up to do smd, so production was cheaper too.

C2P in AA baseline if chunky pixels weren't completed.

C2P hardware wasn't invented until 1993 (well OK that's not quite true - the Sidecar had it 1986, but only for 4 color CGA).

Chunky pixels aren't much of an advantage except for texture mapped 3D, which generally requires a fast CPU to do the 3D math anyway. The correct place to put C2P hardware is on the accelerator card where the CPU can use it at full speed. However once you get past 40MHz software C2P is just as fast because ChipRAM bandwidth is the bottleneck. C2P hardware on the A1200 would have been nice (as it was on the CD32) but not a game changer. Notice how many A1200 accelerator cards came with it after 1993. That's right, none - even though it could easily have been done in a small CPLD. Seems nobody though it was important enough...

DSP3210 for cheap math power increase in AA baseline.

Once again, it wasn't cheap. Just the main chip alone cost more that all the chips (apart from RAM) in the A1200.

Lisa modified for chunky pixels e.g. modified four bitplanes for 256 color chunky i.e.
1st bitplane has 8 bits (1st byte), 8 bits (5th byte),
2nd bitplane has 8 bits (2nd byte), 8 bits (6th byte),
3rd bitplane has 8 bits (3rd byte), 8 bits (7th byte),
4th bitplane has 8 bits (4th byte), 8 bits (8th byte) and etc.

It's based on VGA's four planes that spread the byte word across the four planes. Storage striping is still needed.

Forget bitplanes. Just rearrange the DMA to pull in consecutive longwords and send them to the CLUT one byte at a time. Instead of 4 DMA pointers each reading a different bitplane sequentially you just have one DMA pointer doing the same thing, advancing by 4 bytes (1 longword) each time. The data is stored in latches and then MUXed to the CLUT.

From 1990 to 1991, the entire Commodore business's sweet spot is selling about 900,000 to 1 million Amigas annually.

I agree. Foreign TIGA doesn't benefit Amiga's general-purpose graphics. C65 is another R&D wastage i.e. the R&D resources for C65 should have been the Amiga's.

I agree. 15 kHz resolution mode AA for CDTV. Double scan "business" resolution modes wouldn't be the focus i.e. not guaranteed to work...

I agree. 256 colors are for the lower-effort artwork without being Jim Sachs.

I'm glad we agree on some things. It only took 5000 posts!

"No A600 ECS" for killing the "A500" model sub-brand. "Amiga 500" is like a brand in itself e.g. BMW 3 series.

I agree with you here too. They should not have dropped the A500 until sales cratered. GVP would have been happy with that since they were selling heaps of HD-8s and A530's.

However the engineers were not happy with Commodore selling the A500 with KS1.3. They wanted everyone using KS2.0 so that developers would use the new OS features and Commodore wouldn't have to support the legacy OS. This is the proximate reason that they replaced the A500 with the A500+, then had to sell them in A500 boxes after they ran out of A500s before Christmas. Doh!

[hammer]

Originally Posted by Turrican_3

Yes the PC trajectory was almost vertical, but so was the C64 one at first. Who could predict its manufacturer would be gone in slightly more than a decade? As with many other things in life, hindsight is 20/20. :-P

But I believe the most important thing here is the fact back then, numbers were closer than they ever were compared to any other platform on the graph.

A - roughly - 75/10% split is not market pressure, is the very definition of market dominant and niche platform respectively.

I will stand corrected though: in 1987 it was probably already too late, I should have said 1983-86, when PC vs C64 was more or less 50/35 (although that is likely an apples to oranges comparison: C64 could do nothing against PC dominance, only a properly marketed and - later - improved Amiga might have)

CGS/MOS's 65xx CPU R&D stalled and WDC 65C816 was released in 1985 when competition like Intel released 32-bit 386.

Acorn designed ARM due to the "no new chips" mentality of Commodore. Commodore relied on external Amiga's technology injection for its 16-bit/32-bit transition.

The R&D rot existed after Commodore's MOS purchase and continued into the Amiga era. Commodore's main technology innovation for the A500 is cost reduction based on the existing A1000's technology innovation.

A1000's technology innovation factor wasn't repeated by Commodore since the original Los Gatos Amiga team was dismantled in 1987.

For the ET4000 series, Tseng Labs mastered memory interleave tech to squeeze out VRAM-like performance from low-cost 70-80 ns access time FP DRAM. ET4000AX was released in 1989.

Without mastering memory interleave tech, the VRAM path is a popular route. EDO was released in 1994, replacing FP DRAM.

Commodore didn't have a completed graphics chipset for VRAM's price reduction. 3DO team completed a graphics chipset for 20 ns serial access time VRAM (1 MB) and low cost 80 ns access time FP DRAM (2MB), released in Q4 1993. It would be costly for 2 MB VRAM as UMA Chip RAM.

Sony's PS1 project had early access to NEC's SGRAM and the game console was completed in early 1994. SGRAM (SDR for graphics) replaced VRAM.

S3 Trio64 was released in 1995 with 64-bit wide EDO DRAM. S3 Vision 864/868 used VRAM. S3's approach was cheap EDO DRAM with a wide 64-bit bus.

NVIDIA's NV1 used 64-bit wide EDO DRAM and was released in 1995. NVIDIA was targeting game console partnerships during NV1 and NV2. NVIDIA's approach was cheap EDO DRAM with a wide 64-bit bus.

3DFX's Voodoo 1 used 128-bit EDO DRAM and was released in 1996. 3DFX's approach was cheap EDO DRAM with a very wide 128-bit bus.

In modern times, NVIDIA has a close partnership with Micron for GDDR6X. A close partnership with a high-performance DRAM manufacturer is important for graphics chipset designers.

Commodore had "double rate" leading and failing edge CPU technology. AMD hired C65's CPU designer for K7 Athlon since this CPU has a DDR front-side bus and AMD-760 chipset has DDR SDRAM support.

[Bruce Abbott]

Originally Posted by Thorham

Indeed. Still good for a network, though.

Indeed, especially after somebody wrote a driver for 'standard' PC cards.

The only problem with PCMCIA is that there's only one port. Yesterday I saw a nice Sony VAIO PCMCIA CD-ROM drive being sold on eBay for a good price. I was tempted, but it would mean unplugging the network card to use it. It's easier for me to just mount CD-ROMs over the network to my WinXP PC.

Yesterday I received two more cheap PCMCIA cards from TradeMe (NZ auction site). I want to experiment with them, but not on the A1200 (don't want to risk blowing it up). So I am thinking of making a PCMCIA port adapter for the A500. I may combine this with an ISA bus slot so I can play with those cards too (PCMCIA and ISA bus are actually quite similar).

Originally Posted by hammer

Using PCMCIA as RAM for A1200's 32-bit 68EC020 is not recommended due to being 16-bit RAM, hence it's a pointless feature.

Not pointless, it just doesn't provide the speedup that 32-bit RAM in the trapdoor slot does. Note that like FastRAM on the A500 etc. it's still faster than ChipRAM for some stuff. One reason for using it was that these cards were already on the market for use with the A600, and were cheaper than A1200 RAM boards (when they became available). If you just wanted more RAM and were happy with the current CPU speed then why not?

The other reason was that it put you in good Company with the Atari Falcon and Mac LC, both of which had 16-bit RAM (source for the goose...).

What a wonderful design the Mac LC was, having everything the A1200 should have had - pizza box case with separate keyboard, built in PSU, 16Mhz fully 32-bit 68020 CPU running at full speed, 256 kB of VRAM supporting a display resolution of 512×384 pixels non-interlaced at 8-bit color (or 16-bit color and 640×480 pixels with the 512k upgrade), built in 3.5" SCSI hard drive option, 1.44 MB floppy drive, SIMM sockets for up to 8MB of extra RAM.

What more could you want? Well perhaps not being crippled by a 16-bit bus, more than just a single graphics resolution, some hardware acceleration like blitter and sprites etc., broadcast standard PAL/NTSC video output with genlocking, and 4 channel sound with independent playback frequencies that can use the whole 2MB instead of only precious VRAM. A proper multitasking OS would be nice too.

Oh yeah, and some kind of external expansion port would be great. Having to use up the Processor Direct slot for your (proprietary, expensive) network card is a bummer. Why couldn't they have put an industry standard PCMCIA slot in it? Oh silly me, it's Apple. They don't follow industry standards.

[Bruce Abbott]

Originally Posted by Thomas Richter

This is not really a defect.... Gayle *can* reset the PCMCIA port, but this is usually not done. The story is a bit longer...

Thus, this is not a bug, and certainly not an issue of the hardware.

It is an issue in the A1200, but only because the method of resetting the port is different from the A600 (and undocumented). The root problem is that because PC cards are usually designed for PCs they may not reset properly on power up and/or take too long to initialize. AA Gayle has a PCMCIA reset output, but it doesn't work on the A1200 because the OS doesn't know how to do it. That's where the program CardReset comes in.

[hammer]

Originally Posted by Bruce Abbott

Define 'serious'.

The AAA chipset project began in 1988.

Prove it.

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

The AAA (Advanced Amiga Architecture) chipset was a secret project initiated by Commodore around 1989 to create the next generation of Amiga technology.

Disprove the bambi-amiga claim with a valid counter source.

http://www.bambi-amiga.co.uk/amigahi...ggebrecht.html

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

Lew: "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 an d 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."

"Architectural point of view" is just paper.

Originally Posted by Bruce Abbott

The spec for all 4 chips was completed in February 1989, with prototypes planned to be ready by July (Linda), August (Monica and Mary), and January 1990 (Andrea) - a very 'optimistic' time frame.

There's a difference between "completing the spec" (paper) vs actual implementation.

After a year of "serious" development, AAA reached the silicon stage with bugs.

Originally Posted by Bruce Abbott

That sounds pretty serious, but it was a much harder task than they expected. Despite having 8 engineers working on it, the schedule kept slipping. According to Ed Helper (one of the engineers working on Andrea), chips of that complexity usually took two to three years to develop.

Provide the source as per academic referencing standards.

AAA wouldn't solve the 3D problem i.e. AAA is an ET6000 mistake.

Originally Posted by Bruce Abbott

AAA had a total of ~750,000 transistors spread over 4 chips, compared to only 80,000 transistors in AGA. Unlike AGA it wasn't just an enhanced ECS with much of the original intact (same blitter etc.), but a completely new design that still had to be 100% compatible. This was made doubly difficult by coders who (by accident or design) misused the chipset to get various effects.

I'm aware of AAA's transistor count and AAA built-in register level compatibility with AA.

Due to the missing resource tracking in the graphics API feature, the change in timing across various DMA'ed processing units will cause software incompatibility. Note why I mentioned "PS4 mode" for PS4 Pro and PS5 i.e. timing incompatibility is a bitch with hit-the-metal programming with two smart chips i.e. CPU and GpGPU. Vampire's "turtle mode" concept is not new.

Originally Posted by Bruce Abbott

Another problem is that a 1.25 micron process was needed, but CSG was already having trouble doing 2 micron (OCS was 5 micron). A $15 million investment was required for the new process, and it would take a while to get running smoothly. There was no urgency to get AAA finished sooner because they wouldn't be able to test it until they had prototype chips anyway.

AAA wouldn't solve the 3D problem i.e. the same problem for 1996 era 2D only ET6000.

Tseng Labs exited the PC graphics business due to late 3D R&D and a shortage of funds to develop 3D capable ET6300 and sold their graphics business to ATI.

ET6000 has excellent 2D, but it's missing 3D, hence Tseng Labs exit.

On 16th December 1997, Tseng Labs announced the sale of its graphic design assets to ATI Technologies, Inc. for approximately $3 million USD, and agreed ATI would lease their Newtown, Pennsylvania facility for at least 3 years.

On 21st April 1998, Tseng announced it was actively seeking merger and acquisition candidates. They were bought by ATI Technologies shortly after.

My point, AAA will not solve the 3D transition issue.

Originally Posted by Bruce Abbott

Even if they had been 'serious' the whole time, AAA probably wouldn't have been ready before 1992, and it wouldn't have gone into low-end Amigas anyway due to its high cost. So while they might have made better progress on AAA by throwing more resources at it, that would have just compound the problem. If instead those 8 engineers had been working on AA from the start they might have been able to meet their tight schedule and have AA machines out in 1990!

AAA will not solve the 3D transition issue.

AA with packed pixels will NOT solve the 3D transition issue.

Packed pixel-capable SNES still needs SuperFX or various math 3D DSP add-ons.

The PC competition has an existing workstation 3D experience!

Originally Posted by Bruce Abbott

That doesn't mean what what you are implying.

Prove it.

Originally Posted by Bruce Abbott

AAA was not compatible with AA. The reason for this is that AAA was developed separately (with ECS compatibility in mind) before AA existed.

AAA is only register as compatible with AA.

Note why I mentioned "PS4 mode" for PS4 Pro and PS5 i.e. timing incompatibility is a bitch with hit-the-metal programming with two smart chips i.e. CPU and GpGPU. Vampire's "turtle mode" concept is not new.

Originally Posted by Bruce Abbott

No, it didn't.

Fact: AA-Gayle is dependent on A600's Gayle R&D.

Budgie has bugs e.g. timing bugs with the expansion card. My A1200 1d4 has timing bugs due to a production error.

Originally Posted by Bruce Abbott

It didn't. AA500 was abandoned in favor of the A500+ with ECS, to be followed by a cost-reduced version (A300) intended to be the last hurrah for the a500 line (like Sony's PS1 was for the PlayStation).

"PS One" is portable with an optional LCD screen attachment. Bundled LCD with "PS One" in Y2002.

A300 has no effort to be portable i.e. where's the optional LCD screen attachment?

Competitors like SNES have been building their install base since 1990 which is very important for mainstream 3rd party studios.

AGA only has the full 1993 for a solid sales year. OCS Amigas acted like "Atari ST" against AGA Amigas.

PC competition has been building a VGA install base since 1987 and ramped up in 1990 due to twin Windows 3.0 and Wing Commander releases.

Originally Posted by Bruce Abbott

The A1200 was developed from the A600 because that was the fastest way to get the job done.

Without integrated IDE and PCMCIA mandates, A3000's Fat Gary works as is.

Originally Posted by Bruce Abbott

The same engineer did both, so it was a doddle for him. Same goes for the case design. And by this time the factories were set up to do smd, so production was cheaper too.

Fact: AA-Gayle is dependent on A600's Gayle R&D.

Without integrated IDE and PCMCIA mandates, A3000's Fat Gary works as is.

AA3000+ project was frozen for "more than 6 months" and it was the Pandora AA prototype. This is Commodore management's fault.

Originally Posted by Bruce Abbott

C2P hardware wasn't invented until 1993 (well OK that's not quite true - the Sidecar had it 1986, but only for 4 color CGA).

Akiko's C2P hardware was within a 1-day job.

Originally Posted by Bruce Abbott

Chunky pixels aren't much of an advantage except for texture mapped 3D, which generally requires a fast CPU to do the 3D math anyway. The correct place to put C2P hardware is on the accelerator card where the CPU can use it at full speed. However once you get past 40MHz software C2P is just as fast because ChipRAM bandwidth is the bottleneck. C2P hardware on the A1200 would have been nice (as it was on the CD32) but not a game changer. Notice how many A1200 accelerator cards came with it after 1993. That's right, none - even though it could easily have been done in a small CPLD. Seems nobody though it was important enough...

Motorola wasn't selling 40 Mhz 68030 for $20 i.e. Motorola glued their focus on Intel's 386DX prices instead of AMD's Am386-40 clone or other RISC competitors.

Akiko C2P hardware was the official Commodore position on the packed pixel issue.

https://bigbookofamigahardware.com/b...t.aspx?id=1604

The “chunky to planar” logic was thought out in a lunchtime conversation between Beth Richard (system chip design), Chris Coley (board design), and Ken Dyke (software) over Subway sandwiches on a picnic table in a nearby park one day, because Ken was telling us how much of a pain it was to shuffle bits in software to port games from other platforms to the Amiga planar system. We took the idea to Hedley Davis, who was the system chip team manager and lead engineer on Akiko and he said we could go ahead with it. I showed him the “napkin sketch” of how I thought the logic would work and was planning on getting to it the next day as it was already late afternoon by that point. I came in the next morning and Hedley had completed it already, just from the sketch!

Akiko's C2P hardware was within a 1-day job.

Beyond 1992, Motorola's 68040 prices weren't competitive against Intel 486s.

Once again, it wasn't cheap. Just the main chip alone cost more that all the chips (apart from RAM) in the A1200.

1. Commodore is willing to spend on a higher-cost FMV module with $50 CL-450.

2. Lew administration has DSP3210 for ALL Amigas.

3. A1200 has a "healthy profit margin" for Commodore UK and Commodore International.


CD32's FMV module has the following:

1. 24-bit DAC (STM's STV8438CV) for 16.7 million colors display.

2. MPEG-1 decoder (C-Cube CL450, 352 x 240 pixels @ 30hz, 352 x 288 pixels at 25 Hz, pixel interpolation and frame duplication to produce output formats of 704 x 240 pixels at 60 Hz or 704 x 288 pixels at 50 Hz ),

https://websrv.cecs.uci.edu/~papers/...LES/060803.PDF

CL450 has about 398K transistors with up to 40 MHz. CL450 includes a programmable on-chip "purpose-built" RISC processor with some assist hardware. In quantities of 100K or more per year, the price is less than $50 in 1992.

3. LSI l64111qc (Digital Audio Decoder, 16-bit DAC),

4. 512 KB local RAM, NEC 423260 DRAM 4Mbit (512 KB) with 80 ns.

5. Lattice ispLSI 1024-60LJ CPLD.

Commodore is willing to spend on this non-core business by following the failed CDI.

Commodore says NO for RISC co-processor @ 40 Mhz with Amiga's general purpose.

Commodore says NO for 512 KB Fast RAM with Amiga's general purpose.

Commodore says NO for 24-bit color 704 x 288p with Amiga's general purpose.

Commodore says NO for 16-bit stereo audio Amiga's with general purpose.

Forget bitplanes. Just rearrange the DMA to pull in consecutive longwords and send them to the CLUT one byte at a time. Instead of 4 DMA pointers each reading a different bitplane sequentially you just have one DMA pointer doing the same thing, advancing by 4 bytes (1 longword) each time. The data is stored in latches and then MUXed to the CLUT.

Data striping is needed due to the slowish memory subsystem. VGA can't escape this factor.

I agree with you here too. They should not have dropped the A500 until sales cratered. GVP would have been happy with that since they were selling heaps of HD-8s and A530's.

Keeping the A500(with AdIDE style) avoids the A600 debacle, but it wouldn't solve the 256-color texture-mapped 3D problem.

The difference is loss wouldn't be a $300 million mortal financial injury i.e. Commodore International could survive into 1995 instead of stalling on April 29, 1994.

However the engineers were not happy with Commodore selling the A500 with KS1.3. They wanted everyone using KS2.0 so that developers would use the new OS features and Commodore wouldn't have to support the legacy OS.
This is the proximate reason that they replaced the A500 with the A500+, then had to sell them in A500 boxes after they ran out of A500s before Christmas. Doh!

Commodore provided ReloKick1.3 on the magazine cover disk.

ReloKick1.3 should been bundled with A500P.

[Gorf]

Originally Posted by TEG

Regarding timeline and sales, one thing I don't understand about Commodore management, and it was at Tramiel time, is why they did not produced a C64 in a PC case style very early.
....
So the concept of a motherboard with optional daughter cards was here since long.
...
Now, was it too difficult to add slots to the C64 architecture?

No, that's quite possible and ExpansionPort doublers (or more) from 3rd parties do exist, e.g.:
https://www.go4retro.com/products/x-pander-3/

It is even possible to use two cards at the same time, as long as they are designed for such usage.

Sadly not even Bill Herd thought of this for the C128D...

[Thomas Richter]

Originally Posted by Bruce Abbott

It is an issue in the A1200, but only because the method of resetting the port is different from the A600 (and undocumented). The root problem is that because PC cards are usually designed for PCs they may not reset properly on power up and/or take too long to initialize. AA Gayle has a PCMCIA reset output, but it doesn't work on the A1200 because the OS doesn't know how to do it. That's where the program CardReset comes in.

I believe I responded to that already, but: No, that's not true. It was designed *not* to issue a card reset when booting because that could hang your system if any operating system component was sitting in the RAM on the card. Of, course, by today's standards, using a PCMCIA card to expand the memory of your machine sounds silly, but back then, it was considered an option.

[Thomas Richter]

Originally Posted by hammer

AAA wouldn't solve the 3D problem i.e. the same problem for 1996 era 2D only ET6000.Tseng Labs exited the PC graphics business due to late 3D R&D and a shortage of funds to develop 3D capable ET6300 and sold their graphics business to ATI.ET6000 has excellent 2D, but it's missing 3D, hence Tseng Labs exit.

The problem started already earlier on. Tseng's ET4000 design had a very nice fast ISA bridge, but ISA was dying and replaced by Vesa-Local and PCI, and the Bus speed mattered less with more capable bus systems such as PCI. Tseng also missed to integrate the RAMDAC and the clock generator, which means that their design was not only more expensive (two additional chips needed), it was also limited in resolution as the VGA core of the Tseng has to fetch up to four times wider screens for true-color.

Its registers are 11 bit wide, allowing 2048 bytes(!) to be fetched by scanline. Divide by 3 to get the maximal width for 24-bit truecolor, and consider that you need thrice the pixel speed to get the same amount of pixels on the screen.

The ET6000 was already too late in the market to make a substantial difference - Tseng's competitors could deliver more, cheaper.

Cirrus died by missing the 3D market, but that's another chapter. Tseng was already brain-dead with the arrival of the ET4000.

[TEG]

Originally Posted by hammer

Commodore tried with "big box" C64 https://en.wikipedia.org/wiki/Commodore_Educator_64

The problem is Commodore configured Educator 64 to be monochrome like the Apple II which is "zero sum" which wouldn't displace the incumbent. Educator 64 lacks the expansion slots of Apple II.

Commodore didn't update 65xx CPU family on par with Intel's 16-bit 8086, 16-bit 80286 (with MMU, 24-bit memory address) and 32-bit 80386. Acorn created ARM to address Commodore's 65xx CPU rubbish R&D road map. Commodore purchased the Amiga as a panic move.

Microsoft displaced incumbent text-based Lotus 123 and Word Perfect with Mac ports of MS Excel 2.2 and Word 2.0 for Windows 2.0.

So they gone half the way. We can read: "The internals of the Educator 64 were refurbished Commodore 64 motherboards and monochromatic green monitors."

Refurbished ?? I guess the idea was to lower the cost.

Anyway it was dedicated to the education, to oppose Apple's plans. And the keyboard is attached in the PET style, something schools wanted to avoid thief. Look like it was not targeted to IBM's plans.

Originally Posted by Gorf

No, that's quite possible and ExpansionPort doublers (or more) from 3rd parties do exist, e.g.:
https://www.go4retro.com/products/x-pander-3/

It is even possible to use two cards at the same time, as long as they are designed for such usage.

Sadly not even Bill Herd thought of this for the C128D...

Thanks for this link. Always something new to discover with Commodore

So yeah, they did not capitalized on their winning C64 brand. I would have market a "C64 PC" with a very similar case of the IBM PC. The C64 was only 40 columns and the max resolution was a bit lower but the price would have been accordingly. And with cards slot you would be able to promise 80 columns, 640x200 soon. This was done with the VIC-II E of the C128. So I think C64 sales would have rocket even more due to the expandability. You buy it now because it's affordable and because you now that in one year you will buy the "professional" 80 columns graphic card.

C64:
320×200, 16 colours

IBM PC CGA:
320×200, 4 colours
640×200, 2 colours, one black

[Gorf]

Originally Posted by TEG

And with cards slot you would be able to promise 80 columns, 640x200 soon. This was done with the VIC-II E of the C128.

I guess you mean the VDC instead?

This is a very different animal - a leftover from the C900 project, that was used for the 80 column mode. But actually it can deliver far higher resolutions:

Up to 752x700 (interlaced) in monochrome graphics mode:
http://forum.6502.org/viewtopic.php?f=4&t=2523

And even 1024x296/592i in character mode (> 120 characters)
https://c-128.freeforums.net/thread/...tion-c128-mode

Also a missed opportunity: an ultra cheap VDC-based gfx-card for the A1000 and A2000

[TEG]

Originally Posted by Gorf

I guess you mean the VDC instead?

My bad, the 8563 yes. Wikipedia do not indicate when it was available. I think it was a bit before the C128 so 1984?

The question is, with a correct focus of the R&D, would it possible to produce A VIC-III in early 1983 (the C64 is August 1982) able to output 640×200, 2 colours to directly compete with CGA without waiting for the 8563?

P.S.
Thanks for the links. More to read...

[hammer]

Originally Posted by Bruce Abbott

Not pointless, it just doesn't provide the speedup that 32-bit RAM in the trapdoor slot does.

A1200/A600's single PCMCIA slot has a use case issue e.g. swapping between network and RAM expansion.

IBM T20 laptop has two PCMCIA slots. DELL Inspiron 5150 laptop has one PCMCIA slot.

IBM T20 laptop has SO-DIMM slots.

Originally Posted by Bruce Abbott

Note that like FastRAM on the A500 etc. it's still faster than ChipRAM for some stuff. One reason for using it was that these cards were already on the market for use with the A600, and were cheaper than A1200 RAM boards (when they became available). If you just wanted more RAM and were happy with the current CPU speed then why not?

Again, A1200/A600's single PCMCIA slot has a use case issue e.g. swapping between network and RAM expansion.

Originally Posted by Bruce Abbott

The other reason was that it put you in good Company with the Atari Falcon and Mac LC, both of which had 16-bit RAM (source for the goose...).

Macintosh LC was a 1990 released product.

32-bit ALU equipped CPU with 16-bit external bus is like 386SX PC.

Originally Posted by Bruce Abbott

What more could you want? Well perhaps not being crippled by a 16-bit bus, more than just a single graphics resolution, some hardware acceleration like blitter and sprites etc., broadcast standard PAL/NTSC video output with genlocking, and 4 channel sound with independent playback frequencies that can use the whole 2MB instead of only precious VRAM. A proper multitasking OS would be nice too.

Macintosh's DTP and WYSIWYG office market is larger than the niche video genlock market.



From https://groups.google.com/g/comp.sys.../c/V5H6hA1TGLg
Date: Jun 16, 1993

"The largest PCMCIA SRAM card is 4MB and costs around $1000"

This is expensive for 4MB RAM!!!!!

Actually that is not really THAT cheap for PCMCIA SRAM. 128K at $29
comes out to roughly 23 cents/Kb. You can get a 512K card for $89,
which is 17 cents/Kb; a 1Mb card for $159, which is 16 cents /Kb; and
a 2Mb card for $289 which is 15 cents/Kb

2MB card for $289 ?? This is USD. This is NOT suitable for building an A1200 config that competes against 386DX-16 with 32-bit system RAM.


From Amiga_World_Vol_09_11_1993_Nov.pdf page 94 of 100, in USD.

For A1200
Microbotics Expansion Board for the A1200 bare bone = $139
Fast RAM IC 2 MB = $119
Total: $258 USD.

Baseboard 1208 bare bone = $124 .99
Baseboard 1208 with 2 MB, 32-bit Fast RAM = $239.99

For A500
Supraram 500 RX with 2 MB RAM = $164


3DO's MADAM (Agnus/Alice equivalent) can access both 2MB 80 ns access FP DRAM and 1 MB 20 ns serial access VRAM. MADAM's UMA has a different memory bandwidth.
3DO's CLIO (Denise/Lisa equivalent) can access 1 MB 20 ns serial VRAM.

3DO solves the expensive UMA 2 MB to 3 MB Chip VRAM problem.

Xbox Series X's UMA has different memory bandwidth that is dependant on memory address range i.e. 6GB with 192 bits bus and 10 GB with 320 bits bus.

[hammer]

Originally Posted by TEG

So they gone half the way. We can read: "The internals of the Educator 64 were refurbished Commodore 64 motherboards and monochromatic green monitors."

Refurbished ?? I guess the idea was to lower the cost.

Anyway it was dedicated to the education, to oppose Apple's plans. And the keyboard is attached in the PET style, something schools wanted to avoid thief. Look like it was not targeted to IBM's plans.

"Zero-sum" matching wouldn't displace the incumbent.

https://archive.computerhistory.org/...-05-01-acc.pdf
Page 119 of 981

For 1992
68000-12 = $5.5
68EC020-16 PQFP = $16.06,
68EC020-25 PQFP = $19.99,

68EC030-25 PQFP = $35.94
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


Motorola's "zero-sum" price match 68030-25 against Intel's 386DX-25, meanwhile AMD breaks the status quo with Am386-40.

Originally Posted by TEG

Thanks for this link. Always something new to discover with Commodore

So yeah, they did not capitalized on their winning C64 brand. I would have market a "C64 PC" with a very similar case of the IBM PC. The C64 was only 40 columns and the max resolution was a bit lower but the price would have been accordingly. And with cards slot you would be able to promise 80 columns, 640x200 soon. This was done with the VIC-II E of the C128. So I think C64 sales would have rocket even more due to the expandability. You buy it now because it's affordable and because you now that in one year you will buy the "professional" 80 columns graphic card.

Refer to C128's "zero sum" high-resolution text mode for business.

Microsoft and Apple breaks the status quo with MS Excel and MS Word for Mac and Windows 2.x products against incumbents like high-resolution text-based Lotus 123, Word Star, and Word Perfect.

Amiga's Word Perfect 4.x and 5.x ports are from the MS-DOS text version, hence they wouldn't displace the incumbent.