@kolla

Close, but the tape loop is clearly much too long.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@Karlos

Quote:

Karlos wrote: @ppcamiga1 Please correct your dosage and/or medication. You are complaining that the 68K doesn't support a range of fixed function 3D stuff that's the purview of a GPU. That can't make sense, even in your head.

He's conflating 68k (the chip/ISA) with Amiga 68k (the integrated system with a 68k chip/ISA). For all I know, he might even believe that the Amiga custom chipset uses 68k instructions.

He's the master conflator. Just like he conflates anything other than 32-bit big-endian as being for work and therefore a PC. Ignoring the amount of work PowerPC CPUs performed while inside Apple personal computers and servers.

The confirmation bias is strong with this one. For years he trashes anything AGA, and then he latches onto a YouTube video of Doom on a CD32 wherein he thinks it supports his simple world view that the single thing which stood between Commodore and commercial success is "chunky pixels".

_________________
All the way, with 68k

@ppcamiga1

Produce RPi 4B style solution with out-of-order processing PowerPC CPU for under $100 USD.

_________________
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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@cdimauro

Quote:

cdimauro wrote: It wasn't Gates that stated this.

Actually he did. It's just been misquoted as him saying '640k should be enough for anybody'. which isn't the same thing.

Quote:

Anyway, 640kB is fair enough for a 1MB address space: proportionally, Amiga had exactly the same limit (RAM stopped at $A00000. PC's at $A0000).

1 MB, 10 MB - what's the difference? Still the same 'proportionally', eh? No.

Quote:

Do you understand that the first PC was delivered on 1981? Of course it was very very limited: what do you expect? Just check what the market was offering at the time, included Apple and Commodore machines.

Er, that's my point. IBM didn't offer much more than what was already out there.

The Apple II - which the PC was designed to compete against - was launched in 1977. By 1981 the base model was being supplied with 48k RAM.

Commodore introduced the SP9000 'SuperPET' in July 1981. It was actually an 8000 series Pet with a daughterboard containing a 6809 and extra RAM (96k total), designed as a development machine that could connect to a mainframe.

The VIC-20 (also released in 1981) wasn't competition for IBM and they made no attempt to duplicate it. That was soon followed by the C64 in 1982. Then IBM took notice, and produced their own 'home' computer, the PCjr. We all know how well that went.

Quote:

Amiga arrived FOUR years after and, guess what, offered much much more. Clap clap clap...

IBM could have made a 68000 based PC in 1981, but they didn't because they didn't want to compete against themselves. The PC was just supposed to kill demand for the Apple II etc. in the business world where IBM was trying to sell minicomputers.

But it got away from them. Press coverage was overwhelmingly positive. Demand immediately outstripped their expectations by a wide margin, with many customers putting money down on pre-orders before anyone had even seen one. They ramped up production as fast as they could until they were shipping 40,000 per month. In 1983 they sold 750,000. Then the clone makers got interested. By this time the business computer market was all-in on PCs.

If IBM had put a 68000 in the PC I'm sure it would have been similarly received. The Amiga would then have largely overcome one huge hurdle - IBM compatibility. I'm not sure that would have helped though, because then it would just be another PC with different graphics and OS. Still, Radio Shack did very well with the Tandy 1000 which was a clone of the failed PCjr. Perhaps Commodore could have done something similar with the Amiga.

Quote:

BTW, CGA and MDA had a programmable display controller: something which on the Amiga arrived only with the ECS (great "enhancement")...

The PET had it too. So does that mean the Pet was superior to the Amiga?

Of course you conveniently forget to mention the Copper, which gave the Amiga even more programmability. The only limitation is that the blanking and sync pulses were 'hardwired' to ensure NTSC/PAL compatibility. We see this in many other home computers of the time, where different graphics chip variants were used for different regions. This made sense because the composite color signals were generally region-specific too, and international trade wasn't what it is today.

The CGA card might have been 'programmable' via the 6845 CRTC, but the amount of useful programmability was very limited. it didn't have a PAL composite option because the circuit was designed to work only in NTSC. It didn't have the features desired for games such as smooth scrolling, sprites, synchronized color palette changes etc. Most games just used the BIOS settings because playing around with sync settings could blow your monitor.

But of course you knew this (or should have). Equating the 6845 CRTC to ECS or 1 MB RAM to 10 MB are just more of the silly Commodore bashing that abounds here. I wasn't trying to put down the PC for it's design, just stating facts. Whatever IBM put it in would be a hit. So even if the designers didn't realize it, choosing the 8088 was actually a brilliant move. Because while it was a turd, it could be polished.

The Amiga OTOH was already shiny, but harder to make more shiny. This led to the paradox where the Amiga was seen as not innovating enough (compared to the PC) because it started off too good. But home computer lines traditionally didn't innovate much - they were replaced with better but incompatible lines.

The PC was open-ended enough that it could be polished and polished ad infinitum, until it became very shiny indeed. You have to look deep inside a modern PC to see the remains of the original turd. but the main reason it achieved that was the 3 letters IBM. Without their reputation it would have been just another computer, and nobody would have considered polishing it.

That's the real difference between the PC and the Amiga. If Commodore had produced the PC in 1981 - with every feature identical - it would have been a failure. Because it wasn't IBM. By 1985, producing anything that wasn't IBM compatible was very risky, especially in the US. But that didn't faze Commodore, and I'm glad. It might not have been good for them, but it was for (now retro) computer hobbyists, as we got far more interesting machines to play with.

@bhabbott

Quote:

Er, that's my point. IBM didn't offer much more than what was already out there.

1. IBM PC was cloned, hence fulfilling multi-source common platform requirements.

Microcomputers like Commodore's offerings are harder to clone due to CSG custom chips.

Micheal Dell's "Direct from DELL" book made statements on PC's "off-the-self" components in building his PC clones.

2. The "1st killer app" for PC was Lotus 123 which displaced VisiCalc. Platforms with only VisiCalc have followed VisiCalc's decline.

Lotus 123 beaten SuperCalc and VisiCalc.

Lotus 123 standard is one of the major factors in establishing the IBM PC DOS platform standard. MS Excel GUI has to be better than Lotus 123 establishment, not just a #metoo.

Microsoft recycled Mac's GUI use cases which beats Lotus 123 DOS establishment.

https://www.youtube.com/watch?v=PI23HAEN63c
The Rise of Microsoft Excel: Part 1, Mac GUI was an important factor.

Bill Gates views Mac GUI with "powerful" 68000 as a good platform for MS Excel GUI.

Apple's programmer for Mac's Switcher feature worked with MS Excel programmer team.
MS Excel bundles Apple's Mac's Switcher OS update. This Apple Mac OS access advantage tactic for MS was recycled for Windows and MS Office teams.


@bhabbott
Quote:

Commodore introduced the SP9000 'SuperPET' in July 1981. It was actually an 8000 series Pet with a daughterboard containing a 6809 and extra RAM (96k total), designed as a development machine that could connect to a mainframe.

8086 is a full 16-bit CPU with a 16-bit data bus which is a road map upgrade for the 8088 CPU, while MOS/CSG 6502's road map is crippled.

MOS 6502's evolution pace is slow, and the 6809 CPU alternative didn't help the MOS 65xx credibility.

Intel 80286 was released in Feb 1982.

One of the main reasons for the 16bit/32bit 68000 selection is to leapfrog Intel's 16-bit 80286.

https://cdn.arstechnica.net/wp-content/uploads/archive/articles/culture/total-share.media/graph2-1.jpg
In terms of Commodore's C64 unit sales, it was competitive against IBM PC.

SP9000 wasn't a C64 superset upgrade, and SP9000 failed. Commodore's hard product segmentation would influence Jack Tramiel's Atari ST vs MegaST (with Blitter in 1986). Atari STE's 1989 release was too late.

The C64 platform wasn't able to evolve like the PC's partition graphics adapter.

Commodore didn't offer a migration path from C64 to the Amiga.

There was confusion between the C128 and Amiga 1000 road map upgrade path that mirrors Sega's 32X and Saturn debacle.

For desktops, MOS 65xx CPU family is a dead end, hence the reason for ARM's existence.

SuperFX, ARC, and ARM RISC CPU designs are for migrating from the 65xx/65xxx CPU family. Blame COMMODORE management!

For 65816, Western Design Center (WDC) wasn't run like a professional CPU design company.

Quote:

IBM could have made a 68000 based PC in 1981, but they didn't because they didn't want to compete against themselves. The PC was just supposed to kill demand for the Apple II etc. in the business world where IBM was trying to sell minicomputers. But it got away from them. Press coverage was overwhelmingly positive. Demand immediately outstripped their expectations by a wide margin, with many customers putting money down on pre-orders before anyone had even seen one. They ramped up production as fast as they could until they were shipping 40,000 per month. In 1983 they sold 750,000. Then the clone makers got interested. By this time the business computer market was all-in on PCs.

MPC 1600 was the 1st PC clone that was released in June 1982 https://en.wikipedia.org/wiki/MPC_1600.

Compaq announced its first product, an IBM PC compatible in November 1982, the Compaq Portable.

Lotus 123 was released in Jan 1983.

The first Phoenix PC ROM BIOS was introduced in May 1984, which enabled OEMs such as Hewlett-Packard, Tandy Corporation, and AT&T Computer Systems to build essentially 100%-compatible PC clones without having to reverse-engineer the PC BIOS themselves as Compaq had.


In 1985, Bill Gates wrote an amazing memo to Apple management. In the memo, he praised the Macintosh for its innovative design, but noted that it had failed to become a standard, like the IBM PC was becoming. He correctly deduced that it was the advent of inexpensive, 100%-compatible clone computers that was propelling the PC ahead, and that any defects in the design of the computer would eventually be remedied by the combined force of the many companies selling PCs and PC add-on products, such as new graphics cards. He proposed a plan, which Microsoft would help bring to fruition, whereby Apple would license their operating system and hardware design to a number of other computer companies. Microsoft had been an early supporter and promoter of the Macintosh, but Gates feared that without compatible machines, it would fail to become a "second standard."

Reference: https://arstechnica.com/features/2005/12/total-share/

Last edited by Hammer on 10-Dec-2024 at 06:48 AM.
Last edited by Hammer on 10-Dec-2024 at 06:41 AM.
Last edited by Hammer on 10-Dec-2024 at 06:41 AM.
Last edited by Hammer on 10-Dec-2024 at 06:41 AM.
Last edited by Hammer on 10-Dec-2024 at 06:34 AM.
Last edited by Hammer on 10-Dec-2024 at 06:21 AM.
Last edited by Hammer on 10-Dec-2024 at 06:20 AM.
Last edited by Hammer on 10-Dec-2024 at 05:20 AM.
Last edited by Hammer on 10-Dec-2024 at 05:18 AM.
Last edited by Hammer on 10-Dec-2024 at 05:05 AM.
Last edited by Hammer on 10-Dec-2024 at 04:57 AM.
Last edited by Hammer on 10-Dec-2024 at 04:53 AM.
Last edited by Hammer on 10-Dec-2024 at 04:49 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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Kronos

Oh I don’t know, I find these threads kinda long winded…

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@Hammer

So what's this got to do with 68k? This is now the general 'History of the Micro Computer' thread!

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@BigD

Quote:

BigD wrote: @Hammer So what's this got to do with 68k? This is now the general 'History of the Micro Computer' thread!

1. Hardware needs to be backed by the "killer apps", not just admiring the CPU design.

2. Motorola wasted its 68K lead and allowed X86 to catch up.

3. 68K lost "price vs performance" to RISC competitors.

4. 68K didn't implement a mix of a fast hardware decoder for common instructions and a microcode engine for less-used instructions.

5. For the A500 type machine, Commodore rejected Motorola's 88000 RISC CPU due to the high cost.

68EC040 silicon is the same as the full 68040. Hence, the BOM cost is the same. Motorola adds premium pricing for MMU and FPU.

Last edited by Hammer on 10-Dec-2024 at 10:44 PM.
Last edited by Hammer on 10-Dec-2024 at 09:21 PM.
Last edited by Hammer on 10-Dec-2024 at 08:43 PM.

_________________
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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@kolla

Quote:

kolla wrote: @Kronos Oh I don’t know, I find these threads kinda long winded…

Come back when there are 68000 replies.

_________________
Doing stupid things for fun...

@Hammer

Quote:

Hammer wrote: 1. Hardware needs to be backed by the "killer apps", not just admiring the CPU design.

'Killer apps' come to any platform with market share. The CPU has very little to do with it.

Quote:

2. Motorola wasted its 68K lead and allowed X86 to catch up.

Motorola and Intel were always neck in neck, with Intel having the lead. The 68000 was better than 8086, but arrived later. The 68020 was better than the 80286, but arrived later. The 68030 was better than the 80386, but arrived after it. The 68040 arrived after the 80486. By this time it was clear that Motorola was having trouble making faster chips. The 68040 was very power hungry and maxed out at 33MHz, while Intel pushed the 486 up to 100MHz.

So Motorola didn't 'waste their lead', because they never really had one and couldn't keep up. But that doesn't mean there was anything 'wrong' with the 68000, it's just that Intel dominated the desktop market due to IBM choosing x86, which meant that any other architectures would fail.

For us it doesn't matter a damn. Commodore pegged out in 1994, but that didn't stop us from putting faster 68k CPUs in our Amigas. Intel CPUs OTOH are completely useless to us - unless stuck inside a PC.

Quote:

3. 68K lost "price vs performance" to RISC competitors.

True, but this appears to have been a process issue, not a CPU design issue. Motorola's RISC 88000 CPU wasn't that great either.

Quote:

4. 68K didn't implement a mix of a fast hardware decoder for common instructions and a microcode engine for less-used instructions.

The more complex the CPU the harder it is to do direct decoding without introducing bugs. That's why Zilog's Z8000 failed. A fully microcoded engine is fast enough, and gets the product to market quicker.

Quote:

5. For the A500 type machine, Commodore rejected Motorola's 88000 RISC CPU due to the high cost.

A500 'type' but not an Amiga.

Quote:

68EC040 silicon is the same as the full 68040. Hence, the BOM cost is the same. Motorola adds premium pricing for MMU and FPU.

68EC040 was specifically designed to be cheaper and lower power. There was also a 68EC040V which was static and ran at 3.3V.

While early LC040's may have been normal 040's with a non-working FPU, the die shots below show it having a quite different layout with no FPU present. 4 of the cache RAM blocks have been moved to take the place of the FPU block.

XC68040RC25B (FPU on left side) . . . . . XC68LC040RC33B (die rotated 90 degrees clockwise relative to 68040 die).

@bhabbott

Quote:

'Killer apps' come to any platform with market share. The CPU has very little to do with it.

Lotus 123 supported 286 FPU before C= A2620's 1988 release which is delayed by Motorola's late 68551 MMU release.


Atari ST needs a special high-resolution monitor that is different from 15 kHz gaming resolution monitors. Needs expensive multi-sync monitors. There's a hard product segmentation between business and gaming use cases. 1990 Atari TT's 256-color platform production run was tiny. Atari TOS platform couldn't create an install base large enough business market.


Similar to Atari ST, the Amiga OCS had a high business resolution A2024 monitor workaround in the 5000 units production run. ECS and AGA need higher-cost multisync monitors since the majority 15 kHz game mode wasn't double-scanned for mass-produced 31 khz monitors. CSG Amber's production run is in the tiny few thousands. Amiga platform couldn't create an install base large enough for the business market. ECS and AGA arrived too late.
Commodore couldn't translate C64's success for the Amiga.


PC VGA (1987) has a double scan line hardware for 15 kHz mode 13h and legacy CGA/EGA which allows for common 31 kHz VGA monitors. VGA cloners made this cheap. Official Windows 2.x gained 2 million install base by 1989. 4 million sold copies for Windows 3.0 for its 1st year. Windows 3.1 has 3 million copies sold in 3 months. Pirated copies allowed Windows to flourish. MS Excel GUI and WinWord dethrone DOS establishment Lotus 123, WordStar, and Word Perfect. Business document automation centered on MS Office.


Apple Mac's monochrome business high-resolution GUI mode is the baseline standard. It had MS Excel GUI version, Aldus PageMaker and Adobe PostScript. Apple developed a 14 million business majority install base by 1994 with users who could spend 1.2 million PowerMacs from 1993 to Jan 1994. Steve Job's return to Apple made sure MS Office would be available for the Mac platform.

"Killer Apps" platform also matters.

Quote:

@bhabbott Motorola and Intel were always neck in neck, with Intel having the lead. The 68000 was better than 8086, but arrived later. The 68020 was better than the 80286, but arrived later. The 68030 was better than the 80386, but arrived after it. The 68040 arrived after the 80486. By this time it was clear that Motorola was having trouble making faster chips.

Wrong. Motorola was late with the 68551 MMU release and was an optional extra add-on until 68030.

Mac used 68030 from September 1988.

68551 MMU's late release delayed A2620's release.
68551 MMU's late release caused R&D wastage with Commodore's custom MMUs.

PC world doesn't have to deal with Motorola's 68551 MMU being late.
80286 has integrated MMU baseline to run Xenix 286 on commodity PC clones.
80386 has integrated MMU baseline to run Xenix 386 on commodity PC clones.
Both 386SX and 386DX has MMU as standard i.e. Intel didn't compromise the business standard potential.

68K Unix workstation market is fragmented without a common platform standard. Vendor lock-in was the norm for semi-custom 68K workstations.

Motorola wasn't able to create 14 million 68K UNIX workstations! Motorola was unable to turn 68000's success for 68020/68030/68040! In terms of numbers, Motorola's best business desktop effort was the 68K Mac platform.

68K Unix workstations were jumping ship to RISC-based platforms. Mass-produced Unix boxes weren't with 68K!

Like Windows 2.x 386, Xenix 386 enables multiple MS-DOS business applications to run.

For the smart handheld market, Motorola 68000-based Dragon Ball VZ lost to ARMv4T with a fucking MMU, again!!!!!, Motorola tried to compete against ARMv4 with a bad IPC overclocked handheld 68000 CPU core!

Motorola/Freescale deserves to go bust.

Quote:

@bhabbott The 68040 was very power hungry and maxed out at 33MHz, while Intel pushed the 486 up to 100MHz.

68040's 0.8 μm process tech is similar to 1st gen 1993 Pentium 0.8 μm.

68040 reached 40 Mhz during 1993 via Quadra 840av which is bundled with DSP3210 @ 66Mhz (about 33 MFLOPS FP32).

FYI, AMD's 5x86 is based on a 486-based microarchitecture with a 160 Mhz scale. Am5x86-P75 (133 Mhz, 1.6 million transistors) is a poor man's Pentium clone.

In 1992, the Am386-40's price destroyed the Intel 386DX-25 and Motorola 68030-25 prices.

Quote:

@bhabbott, The more complex the CPU the harder it is to do direct decoding without introducing bugs. That's why Zilog's Z8000 failed. A fully microcoded engine is fast enough, and gets the product to market quicker.

Z8000 is not backward compatible with Z80 CP/M.

CP/M fortunes died with VisiCalc's decline.
C= PET fortunes died with VisiCalc's decline.

Quote:

@bhabbott, A500 'type' but not an Amiga.

1. 68K was already EOL during Commodore's 1989. Commodore is aware of October 1991 AIM (PowerPC) alliance's creation.

2. Commodore was planning for non-68K future for the Amiga.

Last edited by Hammer on 15-Dec-2024 at 12:01 PM.
Last edited by Hammer on 15-Dec-2024 at 11:56 AM.
Last edited by Hammer on 15-Dec-2024 at 11:47 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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Hammer wrote: 1. 68K was already EOL during Commodore's 1989. Commodore is aware of October 1991 AIM (PowerPC) alliance's creation. 2. Commodore was planning for non-68K future for the Amiga.

While by 1991 Commodore had somehow wrangled the Amiga into the 3rd major personal computer platform, its business model of trying to have lightning strike twice with a vastly popular cheap computer (C64 redux) meant that its CPU and general components supply was based on primarily inexpensive parts. Which means hand-me-downs.

Apple would jump on the latest Motorola part for a premium business focused computer, and CBM would wait a few years until Motorola got that part down to a quarter of its original costs. Their go-to-market strategy was not based on being the fastest or the best at one specific thing. It was based on bang-for-buck from a balance between the right mix of components at a favourable price. Kind of like Nintendo. They were more interested in profits from volume than creating a high value proposition SKU. The FMCG of computer makers.

But then they buy Amiga. Some nice and esoteric tech which, if the costs can be brought down, could help them have their new volume queen. Meanwhile, the same tech can be used to make high value proposition SKUs. The kind that can do the work of an Apple Mac but at a lower price, and with more flair. And despite so many missteps, they still emerge as a 3rd global computer platform. Problem is, they are now bi-polar: Amiga is an expensive big box AV pro computer thanks to third parties like NewTek, and Amiga is a cheap little wedge gaming all-rounder that kids and teenagers can get for their birthday or Xmas. For some, it's an indispensable tool, for many it is a toy.

Unlike the later Steve Jobs strategy of Pro and Consumer SKUs simplification, Amiga was mostly consumer, and for companies such as NewTek, it was a white label computer platform. Commodore marketing never really developed the Pro market: The 3rd parties did that. Because CBM's focus on a volume SKU created that particular vacuum. If Commodore were around today, I doubt they'd be making the big box SKUs of A-EON. Instead I bet if they had it their way their computers would be sold in blister packs, so the comparison to Raspberry Pi is actually quite apt.

So yeah, CBM would've moved away from 68k because there were no more hand-me downs to be had. Their sights would always be set on the next part that fits their volume product goals.
But again, that doesn't mean that there was anything inherently wrong with 68k. The entire industry looked toward RISC. Not because 68k was bad, but because OEMs everywhere bought into the RISC marketing and that's what they were keen on buying.

This situation kind of reminds me of the Wankel rotary engine. It had more energy efficiency than a crank-shaft internal combustion engine, but outside of a few car models the auto-makers kept spending vast amounts of money in improving the inferior crank-shaft models. The benefit though comes from standardisation. If everyone is making one type of engine, then its easier for people to move from one company to another. Optimisations form one company can inspire optimisations from another, and so on. I sometimes wonder what the world might look like if the rotary engine optimisations received even half the investment of the crank-shaft engines. Of course it's all moot now as electric motors are finally taking over.

Last edited by agami on 16-Dec-2024 at 11:52 PM.
Last edited by agami on 16-Dec-2024 at 11:50 PM.
Last edited by agami on 15-Dec-2024 at 11:50 PM.
Last edited by agami on 15-Dec-2024 at 11:34 PM.
Last edited by agami on 15-Dec-2024 at 11:32 PM.

_________________
All the way, with 68k

@agami

Quote:

While by 1991 Commodore had somehow wrangled the Amiga into the 3rd major personal computer platform, its business model of trying to have lightning strike twice with a vastly popular cheap computer (C64 redux) meant that its CPU and general components supply was based on primarily inexpensive parts. Which means hand-me-downs.

For the US's low-cost gaming market, the NES has displaced the C64, and SNES's 1990 release has obliterated the C64.

C64's bump in sales during 1990-1991 relied on newly opened markets from the Warsaw Pact before SNES's 1992 European market entry.

Like 32X and Saturn, C128 and A500 upgrade paths were confusing for end users.

A500's gaming success relied on pre-SNES European 1992 entry.

A500's US market efforts were countered by SNES's 1991 US entry.

For Commodore's a million production economics of scale, the A500 wasn't ready for business high-resolution GUI mode. A2024 was in a 5000-unit production run scale. This is a similar problem for Atari ST's special high-resolution monitor requirement.

Microsoft's Xbox team learns from Western game platform failures and spends billions of money (from MS Office/OS/Server divisions) on 1st party exclusives games.

In modern times, Elon Musk is using his billions to set up a Western games company.

Quote:

@agami Apple would jump on the latest Motorola part for a premium business focused computer, and CBM would wait a few years until Motorola got that part down to a quarter of its original costs. Their go-to-market strategy was not based on being the fastest or the best at one specific thing. It was based on bang-for-buck from a balance between the right mix of components at a favourable price. Kind of like Nintendo. They were more interested in profits from volume than creating a high value proposition SKU. The FMCG of computer makers.

For million units economic of scale, Amiga OCS wasn't ready for business high-resolution GUI mode baseline standard.

68K CPU doesn't solve the Amiga OCS issue for high-resolution business GUI use case.

Apple Mac's monochrome business high-resolution GUI is its baseline standard that developed into a multi-million install base for business customers and business software vendors.

Commodore didn't use CSG's 2 micron fabs for ECS Denise with scan line doubler feature which could enable Amiga gaming and productivity mode on common 31 kHz PC VGA monitors in a million units economic of scale. It's leveraging strength from one market and spreading to another market.

Commodore practiced hard gaming vs business market segmentation like Jack Tramiel's product segment approach e.g. C64 vs Super PET, ST vs megaST, STe vs TT.

NVIDIA killed SGI via gaming PC's economies of scale starting with GeForce 256 i.e. hardware T&L benefited CAD, not just Quake 3 NV15 and DIrectX7 games.

Quote:

@agami But then they buy Amiga. Some nice and esoteric tech, which if the costs can be brought down, could help them have their new volume queen. Meanwhile, the same tech can be used to make high value proposition SKUs. The kind that can do the work of an Apple Mac but at a lower price, and with more flair. And despite so many missteps, they still emerge as a 3rd global computer platform.

1. For DTP, Amiga OCS is not ready i.e. A2024's 5000 units production run level.

2. The Amiga OCS platform didn't have credible MS Excel GUI and MS Word GUI in mass numbers i.e. A2024's 5000 units production run level.

3. The A2024 workaround is a bad business decision.

ECS productivity mode was demo'ed on A2000 late in 1988 and management decided to delay the ECS productivity mode feature until A3000's Agnus B 2MB Chip RAM is fixed.

Commodore missed the 1989 window for A500/A2000's ECS Agnus A 1MB and ECS Denise while Microsoft released WinWord for Windows 2.1. Later MS Office dominated the business document automation market.

Unlike IBM's OS/2 Warp, Microsoft ported Doom to Windows 95 as the tip of the spear for the Games for Windows initiative and the precursor for the Xbox team.

Microsoft ported Doom for Windows 3.1's WinG/Win32S as WinDoom. WinDoom (WinG) and Doom95 (DirectDraw) still run on Windows 11.


Quote:

@agami Problem is, they are now bi-polar: Amiga is an expensive big box AV pro computer thanks to third parties like NewTek,

Unlike Apple's QuickDraw Color and QuickTime, NewTek VT doesn't benefit Amiga's general software ecosystem.

NewTek VT's success was in a market niche smaller than Apple's DTP/MS Excel GUI/MS Word GUI business markets. The back office + DTP market is larger than the TV production market.

Commodore's A500 million units economic of scale couldn't get involved with NewTek VT.

NewTek VT wasn't ready for the internet's social media video editing market when PC GPU/iGP card includes video encode/decode acceleration.

Quote:

@agami, and Amiga is a cheap little wedge gaming all-rounder that kids and teenagers can get for their birthday or Xmas. For some, it's an indispensable tool, for many it is a toy.

Amiga OCS wasn't good enough for business high-resolution GUI in mass numbers. Blame Commodore management for R&D mismanagement.

A1000Plus (with AA) was an attempt to mass-produce the desktop Amiga model with one or two Zorro II slots akin to modern gaming PCs with a single AGP slot or a single PEG (PCIe 16X lanes) slot. A1000Plus was canceled.

A modern gaming PC (with optional ECC memory, single PEG slot) is placed between a low-end office PC and a workstation PC (with multiple PEG slots, ECC memory). From Intel's initiative, modern PC laptops/ultrabooks gain external PCIe GPU expansion via USB4 or Thunderbolt 3 and 4. AMD and Intel have direct involvement in the PC platform's product segmentation structure.

Modern wedge desktop computers are today's ultrabooks and laptop PCs, the difference is PC's wedge desktop computers are business resolution capable. A Lenovo ultrabook with AMD HawkPoint APU (e.g. Ryzen 7 8845HS APU) could come with a 3-month Xbox games subscription, it has business-class security and a built-in 120 hz FreeSync 3K display.

Modern wedge desktop computers didn't disappear, Intel obliterated Motorola just as ARMv4T obliterated Motorola/Freescale from the smart handheld devices.

A1000Plus was an attempt to leave behind Jack Tramiel's low-end for gaming vs high-end for business product segmentation model.

$299 priced SteamDeck is a modern A600. SteamDeck is fine for MS PowerPoint presentations.

A wedge computer can be a business computer as long it has the proper hardware for a business use case and is released in a timely manner.

Commodore missed the mass-produced A500 that could display on a mass-produced PC VGA monitor, play Amiga games, and have stable 640x480p business resolution.

Last edited by Hammer on 16-Dec-2024 at 09:56 PM.
Last edited by Hammer on 16-Dec-2024 at 02:35 AM.
Last edited by Hammer on 16-Dec-2024 at 02:29 AM.
Last edited by Hammer on 16-Dec-2024 at 02:26 AM.
Last edited by Hammer on 16-Dec-2024 at 02:22 AM.
Last edited by Hammer on 16-Dec-2024 at 02:15 AM.
Last edited by Hammer on 16-Dec-2024 at 02:00 AM.
Last edited by Hammer on 16-Dec-2024 at 01:33 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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Hammer wrote: Lotus 123 supported 286 FPU before C= A2620's 1988 release which is delayed by Motorola's late 68551 MMU release.

Not sure why you are bringing this up, since I was talking about CPUs not FPUs or MMUs. But if you want to play that game...

Lotus 123 supported the 8087 and 80287 FPUs from version 2.0, which was released in 1985. Motorola's MC68881 FPU was released in 1984. In 1985 it was used by JPL in their Mark IIIfp Hypercube, and Sun Microsystems in their Sun-3 series. IBM used it in the RT PC in 1986.

Before Sculpt 3D was released in 1987 there was little interest in FPUs for the Amiga. Nevertheless by 1986 at least two FPU boards were available. One was a module for the Microbotics Starboard 2 RAM expansion (which I had).

As for Lotus 123, VIP Professional was a close clone of it - released for the Amiga in 1986 with a retail price of US$199.

Quote:

Atari ST needs a special high-resolution monitor that is different from 15 kHz gaming resolution monitors. Needs expensive multi-sync monitors. There's a hard product segmentation between business and gaming use cases.

So you understand why I say that putting a 31kHz mode into the Amiga chipset should not have been a priority - especially since we had flicker fixers from 1988 that allowed you to use a VGA monitor in all screen modes.

Quote:

CSG Amber's production run is in the tiny few thousands. Amiga platform couldn't create an install base large enough for the business market.

The use case for Amber was the A2000. However when the A2320 was released in 1991 there were already several flicker fixers available, which would limit sales. Obviously they wouldn't sell more A2320 boards than there were A2000s.

You say 'business market'. Let's get real here, that market was sewn up in 1981 when IBM released the PC. The A2000 came out in 1987, 6 years later. By that time the PC was firmly entrenched in the business market, leaving only niche markets like desktop video production for anything that wasn't IBM compatible. The Amiga did exploit that market quite well.

As for the installed base, the PC had a huge advantage in the business world not just because it was backed by IBM but also because no one manufacturer defined it - the computer equivalent of the Borg (you will be assimilated!). It was the industry standard in an industry where the standard was vitally important. Any manufacturer with an incompatible platform would be effectively locked out and struggle to survive. This has nothing to do the capabilities of the platform, and everything to do with IBM compatibility. The Amiga didn't have it so goodbye.

So you could argue that 68k was wrong simply because it wasn't x86, and you would be right. But that's not what this discussion is about. The 8088 won despite being a much weaker CPU than the 68000. The 8088 had much lower performance, and yet it attracted the 'killer apps' simply because it was the CPU in the PC.

Quote:

ECS and AGA arrived too late. Commodore couldn't translate C64's success for the Amiga.

The Amiga was successful, bringing in more money than the C64. That wouldn't last forever of course, same as it didn't for all other home computer lines.

Quote:

PC VGA (1987) has a double scan line hardware for 15 kHz mode 13h and legacy CGA/EGA which allows for common 31 kHz VGA monitors. VGA cloners made this cheap. Official Windows 2.x gained 2 million install base by 1989. 4 million sold copies for Windows 3.0 for its 1st year. Windows 3.1 has 3 million copies sold in 3 months. Pirated copies allowed Windows to flourish. MS Excel GUI and WinWord dethrone DOS establishment Lotus 123, WordStar, and Word Perfect. Business document automation centered on MS Office.

Blaa blaa blaa. VGA having 'double scan' in EGA/CGA is irrelevant. It didn't matter what video the PC had in it, that would be the industry standard which developers targeted. Because the PC - whatever it had in it - was it. Any fan who thinks the Amiga would somehow have broken into the business market if only it had a cheap VGA equivalent is deluding themselves.

Quote:

Apple Mac's monochrome business high-resolution GUI mode is the baseline standard. It had MS Excel GUI version, Aldus PageMaker and Adobe PostScript. Apple developed a 14 million business majority install base by 1994 with users who could spend 1.2 million PowerMacs from 1993 to Jan 1994. Steve Job's return to Apple made sure MS Office would be available for the Mac platform.

By 1995 the Mac was on the way out. It was only getting a small share of the market anyway, but Windows 95 effectively killed it. Apple continued to struggle until they diversified into consumer products (ipod, ipad, iphone). Microsoft supported the Mac as a hedge against being targeted for anti-competitive practices (specifically bundling Windows with new PCs, thus disincentivizing the use of alternatives).

Quote:

"Killer Apps" platform also matters.

That's what I said! But not the CPU - except for code compatibility. The 8088 was objectively worse than the 68000, but it got the 'killer apps' because it was in the PC.

Quote:

Wrong. Motorola was late with the 68551 MMU release and was an optional extra add-on until 68030.

The Amiga and Mac OS's didn't use an MMU. Workstation manufacturers like Sun developed their own 68k MMUs to run Unix. The PC didn't need an MMU either in this time period. So Motorola being 'late' with the 68020 MMU is not important. The only reason Commodore put one in the A2620 was so they could run Unix.

Quote:

Mac used 68030 from September 1988.

In the shockingly expensive Macintosh IIx that almost nobody had. Similarly, almost nobody in the PC world had a 386 at that time.

Quote:

Like Windows 2.x 386, Xenix 386 enables multiple MS-DOS business applications to run.

Xenix wasn't mainstream and wasn't sold by Microsoft to end users - an irrelevance for the vast majority of users and for the PC industry. Guess who was the largest vendor of Xenix systems in 1984? Tandy, with their TRS-80 Model 16 68000-based computer.

Quote:

Motorola/Freescale deserves to go bust.

The hubris is staggering.

Microchip didn't go bust despite having some of the worst CPUs known to Man, because they got in first with MCUs in smart cards etc. It's all about capturing the market. Unfortunately Motorola didn't manage that in the desktop computer market, but they were a very strong second with an excellent product. They didn't 'deserve' to go bankrupt any more than any company did in a highly competitive market.

Quote:

Z8000 is not backward compatible with Z80 CP/M.

It isn't? I never would've figured it! And what does this have to with anything? 8086 wasn't backward compatible with 8080 and 8085 either.

Quote:

1. 68K was already EOL during Commodore's 1989. Commodore is aware of October 1991 AIM (PowerPC) alliance's creation.

What a ridiculous statement. I guess all of us who have a 68040 or 68060 are just imagining them, huh?

Quote:

2. Commodore was planning for non-68K future for the Amiga.

Yes, towards the end when they were flailing around trying to find a new direction. But they were also talking about ASICing the 68k. 30 years later we are still talking about it! Turns out RISC wasn't that great, and 68k ISA isn't as redundant as people thought.

@bhabbott

Quote:

68EC040 was specifically designed to be cheaper and lower power. There was also a 68EC040V which was static and ran at 3.3V. While early LC040's may have been normal 040's with a non-working FPU, the die shots below show it having a quite different layout with no FPU present. 4 of the cache RAM blocks have been moved to take the place of the FPU block. XC68040RC25B (FPU on left side) . . . . . XC68LC040RC33B (die rotated 90 degrees clockwise relative to 68040 die).

https://www.youtube.com/watch?v=atvTV9xU3Cw
Compaq 386SX @ 20Mhz running Xenix 2.3.4

Nearly every PC clone with a 386 CPU has the potential to run Xenix 386. Linux was created on a 386-based PC clone.


-----------------------
http://kpolsson.com/micropro/proc1990.htm
For 1991:

March:
At the International Solid-State Circuits Conference, Intel demonstrates a 100 MHz 486 processor.

June:
Motorola announces that 33 MHz 68040 processors should be shipping in September. LOL

Intel introduces the 50 MHz 486 microprocessor. Speed is 41 MIPS. This new 486 employs 0.8-micron technology.

October:
MIPS Technologies officially introduces the 100 MHz R4000, its 64-bit RISC processor.

-----------------------
For 1992:

February:

Advanced Micro Devices announces the availability of the 40 MHz Am386DX processor, for US$114 in 1000 unit quantities.


March:
MIPS Technologies ships the 100 MHz R4000 processor.

Motorola cancels plans for a 68050 processor, to concentrate work on products that were to follow the 68050, including the "Q" project (likely to emerge as the 68060), and a LP040, a low-power 68040 processor.

Intel introduces the i486DX2 microprocessor, with clock speeds of 50 MHz. Price is US$550 each in quantities of 1000.

May:
Intel produces the first test Pentium processor.

June:
Motorola begins volume shipments of the 33 MHz 68040 processor.

July:
Advanced Micro Devices begins work on a fifth-generation x86 processor (in the class of Intel's Pentium chip).
Intel first demonstrates a system running its P5 (Pentium) processor.

August:
Intel introduces the 66 MHz i486DX2 microprocessor.

September:
At IBM, the first silicon copy of the PowerPC 601 processor is made.
Intel introduces the 33 MHz i486SX microprocessor. Speed is 27 MIPS Price is US$189 each for 1000.

November:
Cyrix announces the Cx486S2/50 processor, its first 486-pin-compatible chip. Initial pricing is US$249 in 1000 unit quantities.

December:
Motorola begins volume shipments of the 40 MHz 68040 processor.



http://kpolsson.com/micropro/proc1993.htm
For 1993:

May 1993, Motorola announces the availability of 40 MHz 68040 processor with the price of US$393 in 1000 unit quantities.

June 1993, Intel adds more 3.3 volt 486 processors to its line: i486SX-33 (for US $171), i486DX-33 (for US$324), and i486DX2-40 (for US $406). Prices are in quantities of 1000.

July 1993, AMD priced Am486SX-33 at US$185 in 1000 unit quantities.

October 1993, TI486SXLC-33, TI486SXL-40, TI486SXLC2-50, TI486SXL2-50, with prices in 1000 unit quantities are, respectively, US$79, US$89, US$110, US$149.

AMD 486DXL-40 processor for US$283 and Am486DX2-66 for US$463 in 1000 unit quantities.

X86 CPU clones provided a wide range of 486 prices.



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

For 1992
68030-25 CQFP = $108.75

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

i386DX-25 PQFP = $103.00

Motorola's 68030-25 $108.75 was sideswiped by AMD's 40 MHz Am386DX for US$114.

Commodore's 1991 cost estimates for 68030-25 was $119.84.

Last edited by Hammer on 17-Dec-2024 at 03:35 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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@bhabbott & Hammer
Your history revisionism is off. Intel was behind Motorola in the 1980s and fighting for their survival. The 68000 created the workstation market and the 16/32-bit embedded market. The 8086 was the 3rd ISA on the low end after the 6502 and Z80 and the 2nd ISA on the high end after the 68000 and were in danger of slipping to number 3 with the later introduction of the NS32k. This left the 8086/8088 the middle ground with a more upgradable ISA than most 8-bit CPUs while retaining some level of hardware and software compatibility at a cheaper price than more advanced designs but this is not a healthy position as exhibited by PPC after x86 took over the high end and ARM the low end CPU markets. Intel's propaganda and marketing kept them in the game as they resorted to spreading lies and using scare tactics about the competition with operation Crush. I talked about this in the following post.

https://amigaworld.net/modules/newbb/viewtopic.php?topic_id=45221&forum=25#870074

The 68000 dominated the 8086.

performance 68000
orthogonality 68000
GP registers 68000
address space 68000
datatypes 68000
code density tie (68k is better than x86 though)
support chips 8086
compatibility 8086
price 8086

Intel was lucky to land the IBM PC contract and they did so by selling at least early 8088 CPUs below cost so the high margin desktop market was far from instantaneous. The C64, Apple II, Nintendo and other 6502 markets were more than competitive in the early 1980s but were divided against the unified and growing 808x/x86 market. The 68k remained competitive into the 1990s despite Motorola redirecting profits to PPC development.

1970-1975 Intel 4004, 8008/8080 early MPUs with minimal competition
1975-1980 6800, 6502, Z80 emerge and dominate
1980-1985 6502 & Z80 low end, 68k workstation & 16/32 bit embedded markets created
1985-1990 RISC replaces 68k workstation market, rise of 68k desktop & console, 386 game changer
1990-1995 Mot invests 68k profits into PPC, 68k loses desktop but gains in embedded, x86 desktop!
1995-2000 68k loses console market but gains in embedded, PPC dies on desktop
2000-2005 ARM Thumb-2 with 68k like code density finally replacing 68k old silicon for embedded

The Motorola 6800 is often overlooked due to the 6502 success but it was good for how early it was, introduced the single voltage MPU when most MPUs required 2 or 3 different voltages and it made the 6502 possible. The 6800 had the better ISA and the 6809 exhibited good upgradability. Still Motorola decided to create a new 68000 ISA and hit a home run. Intel deserves credit for introducing the first MPU in the 4004 (and first SRAM chip) but their dominance only lasted from late 1971 into the mid 1970s where the competition took over with better MPUs after only about 5 years of dominance. Motorola was responsible for the 6800, 6502 and 68k development which dominated for 30 years and likely could have dominated longer if 68k profits had been invested back into 68k development instead of PPC development. The embedded market was still using 15 year old 68k silicon which was not the case for 15 year old PPC silicon that nobody is using today, except brain dead and propaganda selling A-Eon using embedded PPC SoCs for the desktop where modern tech is more important?

A large flat 32-bit address space was more important than early limited MMUs and FPUs. Commodore was bankrupt before Windows was 32-bit and received preemptive multitasking in 1995 with the released of Windows 95. The AmigaOS was 32-bit with preemptive multitasking in 1985, a whole decade earlier. The late 1985 386 advantages were not fully realized until 1995 Windows so the Amiga had plenty of time to add standard MMU hardware and support. The lack of CPU+chipset performance in the 1990s was a much larger factor in the demise of the Amiga though. Ironically, the 68k Amiga had the early large advantage over the 80286 AT in CPU, chipset and price. I still do not see the PC as the inevitable destiny. The Mac started off worse than the Amiga and incremental upgrades improved the hardware to be better than the Amiga hardware using the same 68k CPUs (Mac 68k base was 68040 with MMU while Amiga base was 68EC020 without MMU). Mac market share losses and the Apple near bankruptcy occurred with PPC.

agami Quote:

While by 1991 Commodore had somehow wrangled the Amiga into the 3rd major personal computer platform, its business model of trying to have lightning strike twice with a vastly popular cheap computer (C64 redux) meant that its CPU and general components supply was based on primarily inexpensive parts. Which means hand-me-downs.

The Commodore marketing department warned upper management early enough and suggested how to take advantage of not only a low end market but high end and peripheral markets. The problem was that the marketing guy knew more about business than upper management. There are usually few low margin volume leaders and more niche high margin businesses but Commodore wanted another C64 low margin high volume PC or bust. Bust!

agami Quote:

So yeah, CBM would've moved away from 68k because there were no more hand-me downs to be had. Their sights would always be set on the next part that fits their volume product goals. But again, that doesn't mean that there was anything inherently wrong with 68k. The entire industry looked toward RISC. Not because 68k was bad, but because OEMs everywhere bought into the RISC marketing and that's what they were keen on buying.

Hombre was a specialized non-Amiga product but could be used as a GPU peripheral for the 68k Amiga. Commodore still had the 68k licensing and 68k SoC on the table even though they were not decisive enough. Motorola was decisive in betting the farm on PPC and taking the 68k (and 88k) off the table which fared no better. The problem with both businesses was upper management failed to understand the technology, failed to have confidence in the engineers and organic projects, underestimated the importance of compatibility and chose the easy development path.

agami Quote:

This situation kind of reminds me of the Wankel rotary engine. It had more energy efficiency than a crank-shaft internal combustion engine, but outside of a few car models the auto-makers kept spending vast amounts of money in improving the inferior crank-shaft models. The benefit though comes from standardisation. If everyone is making one type of engine, then its easier for people to move from one company to another. Optimisations form one company can inspire optimisations from another, and so on. I sometimes wonder what the world might look like if the rotary engine optimisations received even half the investment of the crank-shaft engines. Of course it's all moot now as electric motors are finally taking over.

I own a silver 1993 Mazda RX-7 base model with sequential twin turbo 1.3 liter Wankel engine. It is one of the most unique and funnest to drive mass produced cars ever made. The Wankel engine has advantages and disadvantages compared to a piston engine. In general, energy efficiency as measured by fuel mileage is worse due to inferior thermal efficiency compared to a piston engine. This is because the combustion chamber has more surface area for heat to escape. The Wankel engine is superior in some efficiency metrics.

performance/volume Wankel
performance/weight Wankel
distance/fuel piston

In other words, a Wankel engine can be smaller and lighter than a piston engine of the same displacement. There are twice as many power strokes compared to a 4 stroke piston engine so there is higher performance potential for the size and weight. There are fewer moving parts which improves reliability. Mazda took the Wankel tech the farthest but indeed stopped short as material, coating and manufacturing technology has improved substantially since the 1990s. The apex seals, rotor housings and side/intermediate plates are made of cast iron which is low friction and wears well but is heavy. The engine could easily weigh half of what it does and I can already pick it up with it being about the size of a beer keg. One of the things that got me interested in the Wankel was that my grandpa knew one of the testers of the Wankel engine in the US. One of the early problems was the seals so they burned oil and he stopped at a filling station and had to use the restroom. He told the the filling station attendant to add a court of oil and the attendant opened the hood to find what looked like a transmission with no engine in the days of V-8 engines and was looking around for cameras thinking he was on candid camera. It is indeed difficult to develop the Wankel in a world designed for piston engine standards. The Wankel is often discriminated against even though sometimes unintentionally (for racing leagues, emissions standards, etc.) as the closest equivalent of piston expectations are translated from a piston engine to a Wankel engine. The Wankel is not completely dead because of "electric motors" as a small and lightweight engine can be used in a hybrid for recharging. The Wankel can also use multiple fuels better than piston engines, especially hydrogen which can destroy piston engine valves that the Wankel lacks (and also gives turbos more of an advantage). Your thesis holds that the 68k ISA is like the Wankel engine. They both have huge performance potential from a small/lightweight footprint while neither is likely to be an energy/power efficiency leader. While the Wankel is less energy efficient for startup, idling and low power use, it is competitive for endurance racing conditions where it has performed well. Most x86 cores are bad at idle/sleep and low power processing but are good at power efficiency (performance/W) much like the Wankel energy efficiency in race conditions while the 68k power efficiency no doubt has more potential. Not only was the in-order 68060 much more power efficient than the in-order Pentium but the 68k embedded market grew much faster than the x86 embedded market after 1990 where power is more important than performance. The 68k embedded market volumes kept up with the x86 desktop volumes for a long time and likely would have exceeded them if it had received investment instead of PPC, perhaps by several times like ARM volumes today. Between 68k and ARM, most embedded developers chose 68k for 20 years but when Motorola took away the 68k option and said choose PPC or ARM, developers chose ARM Thumb(-2) licensed from SuperH and copied from the 68000 to achieve the 68k standard in embedded code density.

@all
I am not posting to wade back into this quagmire but I feel like the misinformation has gone too far. I have Christmas stuff to do so do not expect to post much.

Last edited by matthey on 17-Dec-2024 at 03:37 PM.
Last edited by matthey on 17-Dec-2024 at 04:25 AM.

@bhabbott

Quote:

Not sure why you are bringing this up, since I was talking about CPUs not FPUs or MMUs. But if you want to play that game... Lotus 123 supported the 8087 and 80287 FPUs from version 2.0, which was released in 1985. Motorola's MC68881 FPU was released in 1984.

Your "Motorola's MC68881 FPU was released in 1984" is useless without a desktop computer platform.

Lotus 123 2.0 was about 1 year after IBM PC/AT(286)'s 1984 release.

Your statement is useless for Amiga 500's large economics of scale context.

Quote:

@bhabbott, In 1985 it was used by JPL in their Mark IIIfp Hypercube,

Mark IIIfp Hypercube has a tiny minority install base.

Your statement is useless for Amiga 500's large economics of scale context.

JPL's Mark IIIfp Hypercube is not Amiga software compatible.

Quote:

@bhabbott, and Sun Microsystems in their Sun-3 series.

Sun-3 has a tiny minority install base.

Your argument position is useless for Amiga 500's large economics of scale context.

Sun-3 is not Amiga software compatible.

Quote:

@bhabbott, IBM used it in the RT PC in 1986.

1. Approximately 23,000 RTs were sold over its lifetime, with some [b]4,000 going into IBM's development and sales organizations. Pick OS sales accounted for about 4,000 units. https://en.wikipedia.org/wiki/IBM_RT_PC

That's Atari Falcon-level failure i.e. 10,000 to 15,000 unit sales.


2. Compaq released its 386 PC clone in 1986. 387 is backward compatible with 287 and runs 1985-era Lotus 123 2.0.

https://en.wikipedia.org/wiki/Compaq_Deskpro_386
During the second quarter of 1987, Compaq sold 90,000 units of the Deskpro 386.

https://www.upi.com/Archives/1989/02/23/Compaq-Computer-Corp-with-1988-sales-of-21-billion/8063604213200/
In 1988, Compaq reached $2 billion USD in revenue sales.

"We're seeing the benefits of having led the industry into the new generation of personal computers based on the Intel 386 chip" - Eckhard Pfeiffer, senior vice president of international operations for Compaq, 1988.

Your argument does NOT have the weight of large economics of scale.

Your argument position is useless for Amiga 500's large economics of scale context.

Quote:

@bhabbott, So you understand why I say that putting a 31kHz mode into the Amiga chipset should not have been a priority - especially since we had flicker fixers from 1988 that allowed you to use a VGA monitor in all screen modes.

Your argument does NOT have the weight of large economics of scale.

Your argument position is useless for Amiga 500's large economics of scale context.

Quote:

@bhabbott, The use case for Amber was the A2000. However when the A2320 was released in 1991 there were already several flicker fixers available, which would limit sales. Obviously they wouldn't sell more A2320 boards than there were A2000s.

Without ECS Productivity Mode, deinterlacing 640x400i or 640x512i would require an extra frame buffer.

Amiga market's best addon vendor is $35 million in revenue is GVP.

Your argument does NOT have the weight of large economics of scale.

Apple Mac's high business resolution install base was in the multi-million e.g. 14 million by 1994.

Your argument position is useless for Amiga 500's large economics of scale context.

Quote:

@bhabbott, You say 'business market'. Let's get real here, that market was sewn up in 1981 when IBM released the PC. The A2000 came out in 1987, 6 years later. By that time the PC was firmly entrenched in the business market, leaving only niche markets like desktop video production for anything that wasn't IBM compatible. The Amiga did exploit that market quite well. As for the installed base, the PC had a huge advantage in the business world not just because it was backed by IBM but also because no one manufacturer defined it - the computer equivalent of the Borg (you will be assimilated!). It was the industry standard in an industry where the standard was vitally important. Any manufacturer with an incompatible platform would be effectively locked out and struggle to survive. This has nothing to do the capabilities of the platform, and everything to do with IBM compatibility. The Amiga didn't have it so goodbye. So you could argue that 68k was wrong simply because it wasn't x86, and you would be right. But that's not what this discussion is about. The 8088 won despite being a much weaker CPU than the 68000. The 8088 had much lower performance, and yet it attracted the 'killer apps' simply because it was the CPU in the PC.

Blaa blaa blaa.

1. For MS Excel/WinWord, Microsoft has to battle the same "killer apps" DOS establishment.

2. Apple Mac's high business resolution install base was in the multi-millions e.g. 14 million by 1994.

3. For large economics of scale context, Amiga OCS wasn't ready for stable high-business resolution.

A2000-CR (B2000) wasn't cost-reduced like in Far East Taiwanese PC clones.

A2000-CR didn't use mass-produced "A500" SBC as an Amiga chipset card.

Commodore Canada/Amitech's A2200-1 and A2200-2 used mass-produced CD32 SBC board with Agent secondary board that provided an A1200 CPU edge connector, A3000/A4000 CPU slot, and Zorro II/III slots. Commodore Canada/Amitech's action was too late since their 65,000 CD32 SBC was locked up in the Philippines warehouse.

For about $1500 USD, Commodore Canada/Amitech's A2200-2 has 40Mhz 68030/68882 vs Commodore Germany's A4000 with 68030/68882 @ 25Mhz.


Quote:

@bhabbott, The Amiga was successful, bringing in more money than the C64. That wouldn't last forever of course, same as it didn't for all other home computer lines.

Amiga was successful in games and video toaster-related niches.

When the Amiga gaming scene collapsed, it took out the Amiga desktop application market with it.

Amiga is not a Mac.

Quote:

@bhabbott, VGA having 'double scan' in EGA/CGA is irrelevant.

It mattered since VG's Mode 13h and Mode X used VGA's double scan line feature.

VGA's double scan line feature for VGA ModeX/Mde 13h was resued for EGA/CGA legacy.

VGA is only firmware API level compatible with CGA.

Quote:

@bhabbott, It didn't matter what video the PC had in it, that would be the industry standard which developers targeted. Because the PC - whatever it had in it - was it. Any fan who thinks the Amiga would somehow have broken into the business market if only it had a cheap VGA equivalent is deluding themselves.

You have forgotten Apple Mac's high business resolution install base was in the multi-millions e.g. 14 million by 1994.

Mac's userbase was able to purchase 1.2 million PowerMacs from 1993 to Jan 1994. Clowns from Amiga NG PPC camp assumed Amiga is like Mac.

Quote:

@bhabbott, By 1995 the Mac was on the way out. It was only getting a small share of the market anyway, but Windows 95 effectively killed it. Apple continued to struggle until they diversified into consumer products (ipod, ipad, iphone). Microsoft supported the Mac as a hedge against being targeted for anti-competitive practices (specifically bundling Windows with new PCs, thus disincentivizing the use of alternatives).

In 1996, Steve Jobs returned to Apple, stabilized the Mac platform, simplified Mac product segmentation, and released the "killer OS" i.e. Mac OS X which was released in 2001.

Steve Jobs negotiated with Bill Gates on continuing the MS Office Mac edition.

Mac OS X is the foundation OS for iPhone, iPad, and iPod.

Leadership direction matters.

Quote:

@bhabbott, That's what I said! But not the CPU - except for code compatibility. The 8088 was objectively worse than the 68000, but it got the 'killer apps' because it was in the PC.

For 1981-era cost reduction issues, IBM wanted an 8-bit bus with a 16-bit ALU-equipped CPU. 68008 was late.

NEC-98 and other PC clones used the full 16-bit 8086 CPU.

IBM was dragging its heels with the PC/AT 286 1984 release which later bite them in the ass when Compaq released the 386-based PC in 1986. IBM wasn't able to hold on to the VGA monopoly when ET3000AX was released in Dec 19987.

Commodore's C64 was competing against IBM on unit sales numbers up to 1985.

Commodore didn't provide a seamless upgrade from C64 to the Amiga.

Quote:

@bhabbott, The Amiga and Mac OS's didn't use an MMU.

Steve Jobs was developing NextSTEP which used PMMU which is the foundation for the "killer OS" MacOS X.

Steve Jobs's return to Apple benefited Mac since MacOS X's 2001 release easily countered Microsoft's full Windows NT migration i.e. Windows XP (NT5.1).

Steve Jobs's NextStep team is effectively the Windows NT team for Apple.

Quote:

@bhabbott, Workstation manufacturers like Sun developed their own 68k MMUs to run Unix.

Your argument position is useless for Amiga 500's large economics of scale context.


Quote:

@bhabbott, The PC didn't need an MMU either in this time period.

386 MMU is needed for Xenix 386 (for business PC servers), Windows 2.x 386, and Windows 3.1 386's virtual memory.

Without 386 MMU, NEC's X86 clones died.

MS's Windows NT replaced MS's Xenix.

Quote:

@bhabbott, So Motorola being 'late' with the 68020 MMU is not important. The only reason Commodore put one in the A2620 was so they could run Unix.

386 MMU is needed for Xenix 386 (for business PC servers), Windows 2.x 386, and Windows 3.1 386's virtual memory. Xenix 386 can run multiple DOS applications.

Without 386 MMU, NEC's X86 clones died.

MS's Windows NT replaced MS's Xenix.

PMMU is important to prepare the platform for "big iron" influenced OS switch e.g. VMS influenced Windows NT and BSD/Mach/Darwin MacOS X.

Any 386 PC clone has the potential to run MS/SCO Xenix 386, MS Windows 2.x 386, MS Windows 3.1 386, and IBM OS/2 2.0.

Quote:

@bhabbott, Xenix wasn't mainstream and wasn't sold by Microsoft to end users - an irrelevance for the vast majority of users and for the PC industry. Guess who was the largest vendor of Xenix systems in 1984? Tandy, with their TRS-80 Model 16 68000-based computer.

FYI, Tandy's Xenix 68K is from MS.

Xenix 386 is important for backend business X86 PC servers and it runs x86 DOS "killer apps" establishment. Windows NT's primary purpose is to replace MS's Xenix.

TRS-80 Model Xenix 68K does NOT run DOS "killer app" establishment. The same failure for Windows NT 4.0 for PowerPC and MIPS editions.

Steve Jobs' 68K Mac GUI and next-gen GUI killer apps didn't execute a #metoo.


Quote:

@bhabbott, It isn't? I never would've figured it! And what does this have to with anything? 8086 wasn't backward compatible with 8080 and 8085 either.

1. Z8000 killed the Z80 CP/M platform's road map, just as 68K Amiga killed C64. Hint: bombed back to ground zero.

2. Z80 CP/M died along with VisiCalc's demise. Z80 CP/M platform didn't have a strong display standard like IBM's CGA, EGA and VGA.

3. Digital Research (DR) focused on GEM as their "new future" over CP/M.

Commodore releasing Z80 CP/M capable C128 is a joke when Atari ST has DR's new GEM desktop OS.

4. Digital Research insulted PC clones like Compaq by locking down GEM for genuine IBM PCs and demanding porting fees. Compaq responded by co-funding Windows 2.0 386 with Microsoft (with MS Excel team) and Intel (for 386). Other PC clones followed Compaq's leadership, hence "Wintel" was born.

MS/SCO Xenix 386 supported Compaq 386.


Quote:

@bhabbott, What a ridiculous statement. I guess all of us who have a 68040 or 68060 are just imagining them, huh?

Read the fucking book i.e. Commodore The Final Years.

The public hint was the AIM Alliance formation in 1991.

Last edited by Hammer on 17-Dec-2024 at 06:23 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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@matthey

Quote:

Your history revisionism is off. Intel was behind Motorola in the 1980s and fighting for their survival.

Wrong, Motorola was behind on standardized and integrated PMMU.

Microsoft's Xenix team hated custom MMUs.

For OEM system integrators like Compaq, 1985 era 386 has PMMU as a single-chip solution while 68020/68551 was a two-chip solution. 68030 was late.

Commodore The Final Years book includes a rant against Motorola's late 68551 MMU.

For smart handheld devices, Dragon Ball VZ doesn't include MMU while ARMv4T includes MMU. Repeated mistakes from Motorola/Freescale camp.

Commodore wasted R&D on a custom MMU for 68000. C= custom MMU for 68000 was slow.

Commodore wasted R&D on a custom MMU for 68020 which delayed A2620's release.

Quote:

@matthey The 68000 created the workstation market and the 16/32-bit embedded market. The 8086 was the 3rd ISA on the low end after the 6502 and Z80 and the 2nd ISA on the high end after the 68000 and were in danger of slipping to number 3 with the later introduction of the NS32k. This left the 8086/8088 the middle ground with a more upgradable ISA than most 8-bit CPUs while retaining some level of hardware and software compatibility at a cheaper price than more advanced designs but this is not a healthy position as exhibited by PPC after x86 took over the high end and ARM the low end CPU markets. Intel's propaganda and marketing kept them in the game as they resorted to spreading lies and using scare tactics about the competition with operation Crush. I talked about this in the following post.

For the 1981 era cost reduction methods, IBM wanted an 8-bit external bus with a 16-bit ALU-equipped CPU i.e. 8088 for the design win.

Motorola's 68008 was late.

NEC used 8086 as the baseline standard for PC-98. NEC has 640x480p with 16 colors and a 4096 color palette single-chip solution in 1981.

IBM PGC (PGA) was released in 1984 with 640x480 with 256 colors and a 4096 color palette.

Amiga's 4096 color palette was contemporary with NEC PC-98 and IBM PGC.

There are 18 million NEC PC-98 desktops before it changed to a normal Wintel PC. Japan's market is larger than the UK and West Germany combined.

For 1981, Japan had advanced graphics technology.

IBM VGA and SVGA clones have beaten the PC-98 display standard.

Last edited by Hammer on 17-Dec-2024 at 06:50 AM.
Last edited by Hammer on 17-Dec-2024 at 06:38 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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@matthey

Quote:

The Commodore marketing department warned upper management early enough and suggested how to take advantage of not only a low end market but high end and peripheral markets.

Henri Rubin's wasted R&D on A2000's peripherals and did not focus on the Amiga's core graphics chipset business.

Henri Rubin turned Commodore into an oversized GVP peripherals company.

Jeff Frank's role in the Commodore PC group is to take Commodore Germany's PC designs and cost-reduce them.

For A2000-CR/B2000, the Amiga team at West Chester executed the cost reduction measures on Commodore Germany's 1986 A2000.

Notice the pattern.

West Chester's A2000-CR didn't exploit Commodore's "Far East" A500 cost reduction and large economics of scale e.g. building a common Amiga multimedia SBC in mass numbers.

A common Amiga SBC would like CD32 which can scale to big box Amiga i.e. https://bigbookofamigahardware.com/bboah/product.aspx?id=1829
Commodore Canada's A2200 initiative is too late.

For a similar $1500 USD asking price, A2200-2's 68030/68882 @ 40 Mhz has improved bang per buck when compared to Commodore Germany's A4000's 68030/68882 @ 25Mhz.

Quote:

@matthey The problem was that the marketing guy knew more about business than upper management. There are usually few low margin volume leaders and more niche high margin businesses but Commodore wanted another C64 low margin high volume PC or bust. Bust!

Again, Amiga is not the Mac.

Major Unix vendors from HP and SUN have ECC memory designs which is a step above Xenix 386 PC builds.

When Pentium Pro(1995) gained ECC memory as part of Intel's platform standard, it was sunset for Big Iron Unix RISC. Lintel pushed many Big Iron RISC into oblivion.

Big iron RISC didn't survive the Mhz war between AMD and Intel. Meanwhile, ARM found a safe space in the smart mobile market where it has beaten MIPS, X86 Atom, and SuperH.

SGI was killed by NVIDIA's PC Master Race's large economies of scale.

IBM's POWER is the remaining big iron RISC and IBM has mostly switched to custom big business software services and RHEL. MS also dominates this space with Office 365/Azure.

https://youtu.be/CtnX1EJHbC0?t=292
For the professional workstation market in 1990, Steve Jobs revealed shipments for Sun which are numbered in 40,000 unit range from 50,000 market size. Steve Jobs courted major 3rd party software vendors for NextStep, hence it had the best 3rd party software library for a Unix distribution while it's a barren wasteland on C= AMIX.

In the 1990s, Microsoft also gained custom business document automation via MS Office/MS Visual Studio's strength. Windows NT/MS Office/MS Visual xx countered NextSTEP.

Selling a plain Unix workstation without a good middleware framework is unwise.

Commodore's AmigaOS team is not Steve Job's NextSTEP team.

Microsoft and Commodore are aware of NextSTEP, hint: Workbench 2.0's and Windows 95's 3D metal GUI designs.

Commodore licensed REXX from IBM for automation with different 3rd party applications.

Evans & Sutherland (ex-graphics workstation vendor) is still alive with $9.359 million in Y2020 revenue. Evans & Sutherland aided Namco's System 22 arcade board that powered Ridge Racer which aided Sony's PS1 development.

1 million A600 for Q4 1992 was a major mistake.

Last edited by Hammer on 17-Dec-2024 at 09:27 AM.
Last edited by Hammer on 17-Dec-2024 at 09:24 AM.
Last edited by Hammer on 17-Dec-2024 at 09:20 AM.
Last edited by Hammer on 17-Dec-2024 at 09:04 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 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Commodore licensed REXX from IBM for automation with different 3rd party applications.

This nonsense popped up in the late 90s and doesn’t make any sense - IBM had nothing to do with ARexx, so why would CBM license REXX from IBM? Btw, the “full� myth is that IBM got “the looks� of OS 2.x in return from CBM to use with OS/2, which also doesn’t make any sense as CBM got that look from (as you point out) NeXT as well as Motif. There used to be three articles around claiming this CBM/IBM nonsense, all pointing at each other as source.

Last edited by kolla on 17-Dec-2024 at 09:52 AM.
Last edited by kolla on 17-Dec-2024 at 09:48 AM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC