Was anyone else disappointed with the A1200?

[Bruce Abbott]

Originally Posted by hammer

For stock A1200 and CD32,

1. AGA Blitter is still 16 bit like ECS/OCS Blitter.

But due to lower biplane DMA contention the blitter is 60% faster on a high-res 16 color screen.

Meanwhile, existing full 32-bit 386DX CPU-equipped PCs can be upgraded with fast VGA like Trident 8900CL or Tseng Labs ET4000AX and join the PC's 32-bit 386DX/486 and fast VGA target platform install base.

And yet even with an ET4000AX a 40MHz 386DX is no faster than an A1200 with 50MHz 030 when running Doom.

Amiga's game console nature doesn't work for bigbox desktop platforms.

As a former owner of an A3000 with 50MHz 060, RTG graphics card and 32MB RAM, I disagree. Its only problem was lack of compatibility with 'industry standard' software.

2. 14 Mhz 68EC020 @ 14 Mhz performs like a 7 Mhz due to the shared Chip RAM clock cycle with the AGA chipset.

7 MHz, but 32 bit, and the ROMs ran at 14MHz.

A1200/CD32 needs Fast RAM to reduce this shared memory bandwidth design flaw.

Not a design flaw, but a design decision made to keep the price down and compatibility up. Most other Amiga models only came with ChipRAM in the base configuration for the same reason.

68EC020 has a hardware barrel shifter.

Yes, and other instructions that are faster too, and an instruction cache.

3DO alternative has the key original Amiga engineers. 3DO has 50 MB/s sys bus bandwidth.

But the 3DO was expensive and could only play games (of which there weren't many - 253 according to Moby Games). Introductory price on release in October 1993 was US$699.99, by which time the A1200 was selling for US$399 or less.

[Thorham]

Originally Posted by hammer

Amiga's CPU power is bottlenecked by Motorola's 68K dead end e.g. 68060 at 50 Mhz.

I meant why they bothered with an Amiga.

[Bruce Abbott]

Originally Posted by hammer

C65's 256 colors display capable chipset was completed in Dec 1990. R&D effort wasted. This should have been C128.

If you mean the C128 should have had the specs of the C65, that wasn't its goal. The C128 was supposed to be a stopgap until the Amiga matured (ie. until 1987).

The C65 in many ways mimicked the Amiga, with its 4096 color palette, bitplanes, blitter, 16 bit CPU, 'dumb' 3.5" floppy drive etc. It also (unlike the C128) wasn't fully C64 compatible. Unfortunately this meant that yes, it was 'wasted' R&D effort, but mainly because the C64 line had run its course and a 'next generation' machine probably wouldn't have revived interest much. But only a small number of engineers were working on it, and having several 'irons in the fire' provided more flexibility to respond to changes in the market. It's just that the market changed in a direction that didn't favor the C65.

Interestingly Commodore wasn't the only home computer maker who tried to extend a machine's lifespan with similar enhancements. Amstrad's CPC Plus range introduced 4096 colors, hardware scrolling, sprites, DMA sound and a cartridge port. It also appeared very late, in 1990 (the original CPC range debuted in 1984), and was not successful because the CPC had also run its course by then.

AA3000+ AGA was completed around Feb 1991. AA3000+ AGA doesn't have A600's PC-originated IDE and PCMCIA.

Not entirely true. The chipset was not completed - it still had bugs to iron out.

Commodore wasted "more than 6 months" (cited DaveH's statement) on an unreleased ECS-based "A1000Jr" project.

The reason it was ECS based was that the AGA chipset wasn't ready in time. The 'A1000jr' was another stopgap. However this one wasn't accepted by Commodore's own subsidiaries who were expecting an AGA machine, so it morphed into the A4000.

Perhaps if more effort had been put into AGA this wouldn't have happened, but the area to draw on for this would be AAA (which had 15 engineers working on it) not the C65.

Without IDE and PCMCIA, AGA A500 could be launched in Q4 1991 or Q1 1992. ECS A600 is a waste of time.

But the A600 paved the way for the A1200, which is what it 'should' have been from the start. Once again the A600 was a stopgap until AGA was finalized.

Commodore had a "read my lips, no new chips" directive during A3000's R&D phase, hence A3000 is stuck with ECS and 1989's A500 Rev 6A had a 2 MB Chip RAM jumper and ECS Agnus.

Without this directive the A3000 would have been delayed further, which would be not just 'no new chips' but 'no new machine'. Maybe that wouldn't have been such a bad thing, but not what fans imply it meant.

A300 suffered a scope creep due to Commodore Germany's demand for IDE which resulted in A600's March 1992 release and A500's cancellation.

Except this 'scope creep' was a good thing. Games were getting larger so a hard drive was becoming desirable, but the A300 had no Zorro expansion port. It couldn't take FastRAM either, which is why they put PCMCIA in it. Both of these features became rather useful to A600 owners, making it worth spending a bit more money on. What the A300 really needed was more 'feature creep' - numeric keyboard, AGA chipset, more powerful CPU and a bus expansion connector - ie. the A1200!

You may think all this parallel development was a waste of time and resources, but this is expected in R&D, especially in an industry which is changing rapidly - as home computing was in the 1980s and 90s. Most fans agree that more R&D was needed for Commodore to meets its targets, but few appreciate the continual financial pressure it was under that made this difficult. Perhaps if Gould and co. had diverted most of their earnings to R&D it might have made the difference if the engineers had understood the urgency and responded accordingly.

By the late 90's things were settling down as the PC asserted its dominance and other architectures couldn't compete due to lack of software compatibility and buyer resistance. Eventually Commodore would have hit this wall anyway, as others did. In an ideal world they might have gotten another 3 or 4 years, but the next generation Amiga would seal its fate just like the C65 and Amstrad CPC Plus did for those lines. Most industries settle into one or two architectures, while the Amiga was always a distant third. Fans may not like it but that's the way things generally go...

[Bruce Abbott]

Originally Posted by hammer

Amiga's CPU power is bottlenecked by Motorola's 68K dead end e.g. 68060 at 50 Mhz.

The 68060 actually 'dead ended' at 75MHz, and had better integer performance than Pentium. Later revision chips have been pushed to over 100MHz, showing that it could have competed throughout the 90s.

Amiga had it's 68060 @ 50 Mhz experience in 1995.

In 1995, Intel had its 3rd-generation classic Pentium improvements i.e. 120 Mhz and 133 Mhz models. Pentium Pro (P6) 150 Mhz was also released in 1995.

But those Intel CPUs were very expensive on release.

Intel's strategy of sorting manufactured CPUs by speed rather than trying to grade them all the same worked because the PC market responded well to it. Some buyers were willing to spend big bucks for the fastest possible CPU, while the majority went for slower cheaper chips. This wouldn't have worked in the Amiga's smaller market.

The reason Motorola dropped the 68k desktop line was not having enough customers. Commodore alone was not buying enough to justify continuing it. Once Apple moved to PPC that was the end.

Well not quite the end. Commodore had plans to produce an 'Amiga on a chip' that would integrate a CPU core and ECS or AGA chipset. FPGAs eventually became powerful enough that even hobbyists could do the same without spending a fortune. The Vampire's 68080 core is equivalent to a 100-200MHz 68060. If put into an ASIC it could be several times faster again.

Software CPU emulation has also proved to be viable, with the Raspberry Pi's ARM CPU getting even higher performance at lower cost. Commodore might been able to do the same with PA RISC, providing a bridge to more powerful CPUs after Motorola stopped making 68k.

For raytracing and compute power, it's either Pentium/Pentium Pro or the RISC hype.

For the most possible compute power yes, but the market for that was very small. OTOH the majority of A1200 RAM expansions boards came with an FPU (or socket for one) because ray tracing was a popular activity on the Amiga. Having the FPU built in made the 040 and 060 more popular too.

Amiga fans didn't need to have the fastest possible machine for their hobby activities, just fast enough to not be too boring. This trend started in the early days when Eric Graham showed his 'Juggler' demo and provided the source code he used to create it. Even though the A1000 only had a 7MHz 68000 and no FPU, fans enjoyed creating their own raytraced images by simply leaving the machine on overnight (I did that!). FPU addons quickly became popular to reduce rendering times. This was a 'killer app' that increased interest in higher-end Amiga systems.

[Cyprian]

Originally Posted by sokolovic

For the 386:
https://en.wikipedia.org/wiki/I386



For the A1200 :
http://amiga.resource.cx/perf/sysspe...cpu&order=mips

68EC020/14MHZ 4.53 MIPS

So yes it is roughly the same, probably a bit better for the A1200 since the figures above are for a 20mhZ 386, not a 16mhz one

EDIT (frome the same WP article) :

I see 1.3 MIPS (chip) / 1.9 MIPS (fast) for A1200 68EC20 14MHz
https://www.ppa.pl/forum/amiga/29637...-s-ram#m417288
https://forum.amiga.org/index.php?topic=27590.15


Regarding memory performance,


A1200 68EC20 14MHz Busspeed test:
Chipram Read: 4.5 MB/s
Chipram Write: 6.9 MB/s


386SX 20MHz Speedsys:
Write: ~15MB/s
Move (read + write): 11MB/s
https://www.vogons.org/viewtopic.php?p=1057325#p1057325

[Cyprian]

Originally Posted by hammer

For stock A1200 and CD32,
2. 14 Mhz 68EC020 @ 14 Mhz performs like a 7 Mhz due to the shared Chip RAM clock cycle with the AGA chipset.

Originally Posted by Bruce Abbott

7 MHz, but 32 bit, and the ROMs ran at 14MHz.

chipsets are clocked at 3.5Mhz and the CPU has access to the chip ram every other chipsets clock. This means that 68EC020 have to wait 8 cycles for each chip memory access.

[CCCP alert]

Doom was coded slowly and properly for x86 and DOS Protected mode advantages, a true port if it was done for A1200 only would be like Dread, so I don't do the Adoom etc comparisons, YMMV.

By Xmas 1993 the A1200 should have been bundled with a cheap bespoke made 2mb exp for £400 due to the 7mhz 020/Chip RAM cock up but by 1993 Gould and Ali had bled Commodore's finances dry vs 1982 half a year up front manic mass production of C64s. No corp. cash = no massive production runs to make this happen. oh well.

[CCCP alert]

Originally Posted by hammer

C65's 256 colors display capable chipset was completed in Dec 1990. R&D effort wasted.

The C65 had a version of compressed screen RAM as an option via split video/colour RAM as well as bitmap screens like Amiga, had a sort of blitter, had min 6 channel audio no worse than MD/PCE artistically and it would sell for the magic £249.99 vs £399.99 4 channel 32 colour A500+ LOL.

A third party R&D company shrank the C64 motherboard to the size of a pack of smokes....hmm I wonder what area of the console market that could be used for in 1990/91 hmmm so tough to think LOL

[sokolovic]

Originally Posted by Cyprian

I see 1.3 MIPS (chip) / 1.9 MIPS (fast) for A1200 68EC20 14MHz
https://www.ppa.pl/forum/amiga/29637...-s-ram#m417288
https://forum.amiga.org/index.php?topic=27590.15


Regarding memory performance,


A1200 68EC20 14MHz Busspeed test:
Chipram Read: 4.5 MB/s
Chipram Write: 6.9 MB/s


386SX 20MHz Speedsys:
Write: ~15MB/s
Move (read + write): 11MB/s
https://www.vogons.org/viewtopic.php?p=1057325#p1057325

Well if you have your figures why did you asked for them ?

Plus, the post you answered was speaking about 68020/14mhz in general. I don't know if the figures on my link about the A1200 are correct but they pretty much match what is listed there.
https://en.wikipedia.org/wiki/Instru...29?wprov=sfla1

I'm not even sure that SysInfo is very reliable anyway. Benchmark on the links I've given to you are made with AIBB or SysSpeed apparently.
https://amiga.resource.cx/performance.html

[Cyprian]

Originally Posted by sokolovic

Well if you have your figures why did you asked for them ?

Plus, the post you answered was speaking about 68020/14mhz in general. I don't know if the figures on my link about the A1200 are correct but they pretty much match what is listed there.
https://en.wikipedia.org/wiki/Instru...29?wprov=sfla1

I'm not even sure that SysInfo is very reliable anyway. Benchmark on the links I've given to you are made with AIBB or SysSpeed apparently.
https://amiga.resource.cx/performance.html

The context was Amiga vs PC, therefore I ask because I'm curious on what basis this statement was based.

We know MIPS means nothing, especially when the platform itself can limit the CPU performance.Therefore before concluding what is better, it would be nice to see the same test benchmark on both platforms.

[hammer]

Originally Posted by CCCP alert

The C65 had a version of compressed screen RAM as an option via split video/colour RAM as well as bitmap screens like Amiga, had a sort of blitter, had min 6 channel audio no worse than MD/PCE artistically and it would sell for the magic £249.99 vs £399.99 4 channel 32 colour A500+ LOL.

A third party R&D company shrank the C64 motherboard to the size of a pack of smokes....hmm I wonder what area of the console market that could be used for in 1990/91 hmmm so tough to think LOL

C65 had no C65 games since Commodore had no 1st party game studio.

R&D effort was wasted. Sega also wasted R&D on the X32 (pushed by Sega America) instead of focusing on Saturn.

[hammer]

Originally Posted by CCCP alert

Doom was coded slowly and properly for x86 and DOS Protected mode advantages, a true port if it was done for A1200 only would be like Dread, so I don't do the Adoom etc comparisons, YMMV.

By Xmas 1993 the A1200 should have been bundled with a cheap bespoke made 2mb exp for £400 due to the 7mhz 020/Chip RAM cock up but by 1993 Gould and Ali had bled Commodore's finances dry vs 1982 half a year up front manic mass production of C64s. No corp. cash = no massive production runs to make this happen. oh well.

There was a multi-million dollar fraud by Commodore Netherland's Bernard van Tienen.

Bernard wasted $480,000 on a very high-class brothel.

Bernard was subsequently the subject of subpoenas issued by the Dutch liquidator in the amount of 300 million guilders which is about $147 million USD.

[hammer]

Originally Posted by Cyprian

chipsets are clocked at 3.5Mhz and the CPU has access to the chip ram every other chipsets clock. This means that 68EC020 have to wait 8 cycles for each chip memory access.

Only Lisa's and Alice's fetch bitplane can access up to 28 MB/s as part of 64-bit wide sprites which is 4X larger than OCS/ECS's 16-bit wide sprites.

Besides the 64-bit wide 8-sprite engines from the Lisa, Alice is still stuck with a 16-bit Blitter and the Copper is not even a RISC math co-processor.

Stock A1200's 68EC020 has Sysinfo's 1.35 MIPS. This Sysinfo MIPS value nearly doubles with 32-bit Fast RAM e.g. 2.95 MIPS with A1208 RAM card.

The CPU has a max of 7.1 MB/s Chip RAM access via BusTest Chip benchmark.

14 Mhz 32-bit bus has a 56 MB/s theoretical peak, 2 cycle Fast Page reduces this value to 28 MB/s.

For geometry processing, 68020/68030 at 16 Mhz is nowhere near 16 MIPS.

Dave Haynie's intention for AGA was to be paired with AT&T DSP3210 @ 50 Mhz (13.3 MIPS, 25 MFLOPS FP32) with 32-bit Fast RAM.

[hammer]

Originally Posted by Bruce Abbott

If you mean the C128 should have had the specs of the C65, that wasn't its goal. The C128 was supposed to be a stopgap until the Amiga matured (ie. until 1987).

C128 was like an ECS job i.e. take an aging gaming hardware and add "business" very low color high-resolution modes.

Commodore has C900 R&D and one of its key engineers led the Atari ST project.

C900 is Commodore's 16-bit in-house desktop computer from Commodore Germany. C900 has "business" B/W high-resolution modes.

Originally Posted by Bruce Abbott

The C65 in many ways mimicked the Amiga, with its 4096 color palette, bitplanes, blitter, 16 bit CPU, 'dumb' 3.5" floppy drive etc. It also (unlike the C128) wasn't fully C64 compatible. Unfortunately this meant that yes, it was 'wasted' R&D effort, but mainly because the C64 line had run its course and a 'next generation' machine probably wouldn't have revived interest much. But only a small number of engineers were working on it, and having several 'irons in the fire' provided more flexibility to respond to changes in the market. It's just that the market changed in a direction that didn't favor the C65.

Interestingly Commodore wasn't the only home computer maker who tried to extend a machine's lifespan with similar enhancements. Amstrad's CPC Plus range introduced 4096 colors, hardware scrolling, sprites, DMA sound and a cartridge port. It also appeared very late, in 1990 (the original CPC range debuted in 1984), and was not successful because the CPC had also run its course by then.

Not entirely true. The chipset was not completed - it still had bugs to iron out.

AA3000+ AGA is AmigaOS bootable (cite DaveH).

A3000 had Super Buster bugs that were solved into later A4000's Rev 11. Moving the A3000's SCSI controller from the Ramsey bus into the Zorro III bus, A4091 SCSI card would need A4000's Super Buster Rev 11. One decision led to exposing other bugs. Some early A4000/040 had Super Buster Rev 9.

AGA project has "lost more than 6 months" due to A1000Jr waste of time (cite DaveH) and C65 is resource R&D misdirection.

From Commodore's "read my lips, no new chips" directive, Commodore's workaround for A3000's graphics capability issue is the PC-originated expensive TIGA-based A2410 add-on card which did nothing for the Amiga graphics chipset.

Originally Posted by Bruce Abbott

The reason it was ECS based was that the AGA chipset wasn't ready in time. The 'A1000jr' was another stopgap. However this one wasn't accepted by Commodore's own subsidiaries who were expecting an AGA machine, so it morphed into the A4000.

Again, the AGA project has "lost more than 6 months" due to ECS A1000 Jr waste of time (cite DaveH) and C65 is resource R&D misdirection.

A3000's R&D phase had a "read my lips, no new chips" directive (cite DaveH).

Originally Posted by Bruce Abbott

Perhaps if more effort had been put into AGA this wouldn't have happened, but the area to draw on for this would be AAA (which had 15 engineers working on it) not the C65.

Meanwhile, key original Amiga engineer leadership in 3DO Company released their 3DO game console in late 1993.

Amiga lost key engineering talent in 1987.

3DO game console has a 2 million install base. 3DO's mistake is a hard-to-program quadrilateral 3D system and the ARM60 CPU @ 12.5 Mhz wasn't strong enough, hence a bad Doom showing.

3DO's quadrilateral 3D mistake is similar to Sega Staurn's quadrilateral 3D mistake. 3DO M2 has triangle based polygon 3D system.

Originally Posted by Bruce Abbott

But the A600 paved the way for the A1200, which is what it 'should' have been from the start. Once again the A600 was a stopgap until AGA was finalized.

Without this directive the A3000 would have been delayed further, which would be not just 'no new chips' but 'no new machine'. Maybe that wouldn't have been such a bad thing, but not what fans imply it meant.

Commodore fired the original Los Gatos Amiga group in 1987. https://en.wikipedia.org/wiki/Amiga_Ranger_Chipset

1987's evolved Amiga chipset with 128 colors and 1024×1024 resolution designed for fast VRAM was replaced by PC-originated expensive TIGA-based A2410 (1024 x 1024 with 256 colors) 1991 add-on card for the A3000.

A3000 has fast 768 KB video memory for Amber Flicker Fixer instead.

Commodore was bleeding proven engineering talent.

Originally Posted by Bruce Abbott

Except this 'scope creep' was a good thing. Games were getting larger so a hard drive was becoming desirable, but the A300 had no Zorro expansion port. It couldn't take FastRAM either, which is why they put PCMCIA in it. Both of these features became rather useful to A600 owners, making it worth spending a bit more money on. What the A300 really needed was more 'feature creep' - numeric keyboard, AGA chipset, more powerful CPU and a bus expansion connector - ie. the A1200!

A600 was a sales flop (cite DavidP). Due to the A600's existence, the A500's cancellation caused a large-scale revenue drop during 1992.

Reminder, the A300 project was supposed to replace C64's price segment.

The higher priced A500 Rev 9 could have A600's surface-mounted chips, PCMCIA, Zorro 1, and IDE. "Amiga 500" has a good reputation in the marketplace.

[hammer]

Originally Posted by Bruce Abbott

But due to lower biplane DMA contention the blitter is 60% faster on a high-res 16 color screen.

16-bit Blitter wasted its 32-bit cycle. Turrican 2 AGA shows its 256-color limitation i.e. needs Fast RAM to enable AGA's potential.

Originally Posted by Bruce Abbott

And yet even with an ET4000AX a 40MHz 386DX is no faster than an A1200 with 50MHz 030 when running Doom.

Reminder, A1200 with 3rd party 68030 @ 50Mhz accelerator is priced around 486SX-25 and 486SX-33Mhz based PC clone. This is for the UK, US, and AU markets.

386DX-33's memory subsystem wasn't stuck in 1985 and clone 486 upgrades were available. I slightly overclocked my 386DX-33 to 40 Mhz.


https://vintageapple.org/pcworld/pdf..._June_1993.pdf
Gateway Party List, Page 72 of 314
4SX-33 with 486-SX 33Mhz, 4MB RAM, 170 MB HDD, Windows Video accelerator 1MB video DRAM, 14-inch monitor for $1494,
4DX-33 with 486-DX 33Mhz, 8MB RAM, 212 MB HDD, Windows Video accelerator 1MB video DRAM, 14-inch monitor for $1895,

Page 128 of 314
Polywell Poly 486-33V with 486SX-33, 4MB of RAM, SVGA 1MB VL-Bus, price: $1250


https://vintageapple.org/pcworld/pdf...ugust_1993.pdf
Gateway Party List, Page 62 of 324
4SX-33 with 486-SX 33Mhz, 4MB RAM, 212MB HDD, Windows Video accelerator 1MB video DRAM, 14-inch monitor for $1495,
4DX-33 with 486-DX 33Mhz, 8MB RAM, 212 MB HDD, Windows Video accelerator 1MB video DRAM, 14-inch monitor for $1795,

Page 292 of 324
From Comtrade
VESA Local Bus WinMax with 32-Bit VL-Bus Video Accelerator 1MB, 486DX2 66 Mhz, 210 MB HDD, 4MB RAM, Price: $1795

https://vintageapple.org/pcworld/pdf...tober_1993.pdf
October 1993, Page 13 of 354,
ALR Inc, Model 1 has a Pentium 60-based PC for $2495.



https://archive.org/details/amiga-wo...ge/n7/mode/2up
Amigaworld, October 1993, Page 66 of 104
Amiga 4000/040 @ 25Mhz for $2299
Amiga 4000/030 @ 25Mhz for $1599

Page 82 of 104
M1230X's 68030 @ 50Mhz has $349
1942 Monitor has $389
A1200 with 85MB HDD has $624
A1200 with 130MB HDD has $724

The Commodore solution is beaten by the Gateway solution.

Target sales period: XMas of 1993 Q4. 1993 XMas sales period was Commodore's last chance.

Beyond 1992, the Amiga platform doesn't have "bang per buck" and a new "32-bit 3D" gaming experience to attract new users at 1990 to 1991 levels.

Originally Posted by Bruce Abbott

As a former owner of an A3000 with 50MHz 060, RTG graphics card and 32MB RAM, I disagree. Its only problem was lack of compatibility with 'industry standard' software.

Reminder, Amiga's 68060 @ 50Mhz experience started around 1995 with CyberGraphicsX RTG.

For my A3000 during early 1996, I price quoted for Phase 5 CyberStorm 060 and CyberGraphics 64 (S3 Trio 64), a new build Pentium 150/S3 Trio 64UV based PC has superior bang for the buck and absolute performance. Phase 5's tax is a major factor despite 68060 @ 50Mhz being sold at half the price of Pentium equivalent.

During 1996, I estimated 68060 @ 50Mhz had inferior Quake experience compared to Pentium's 166 Mhz target.

I slightly overclocked my Pentium 150 Mhz into 166 Mhz via 60 Mhz to 66 Mhz FSB jumper. In 1997, I purchased my RIVA 128 due to my university's workstation PC's RIVA 128 fleet deployment.


Pentium PC's 1995 vs Amiga's 1995
From https://archive.org/details/EA1995/E...e/n69/mode/2up
Page 70 of 148
Ritron Computers System in the Australian state of Victoria
Pentium 90-based PC has $1945 including tax or $1595 without tax.
Includes: 1.44FDD, 8MB RAM, 540MB HDD, 256KB L2 cache, PCI video card, keyboard, 14-inch SVGA monitor.

For November 1995, Pentium 90 with PCI motherboard combo has a $799 asking price with tax.

https://archive.org/details/EA1995/E...e/n67/mode/2up
December 1995 Xmas month. Page 88 of 68.
Ritron Computers System in the Australian state of Victoria
Pentium 90-based PC has $1872 including tax or $1535 without tax.
Includes: 1.44FDD, 8MB RAM, 540MB HDD, 256KB L2 cache, PCI video card, keyboard, 14-inch SVGA monitor.

Multimedia upgrade kit with CD-ROM 16-bit sound card, speakers, MS Encarta/Works/Money/Golf/Dangerous Creatures
CD-ROM 2X = $329
CD-ROM 4X = $399
CD-ROM 6X = $649


https://archive.org/details/Australi...ont_Studios_AU
The Amiga returns in the Australian 1995 market
From Australian Amiga Review, Nov 1995
Page 2, Cybervision 64 (S3 trio 64U), $1099 AUD

Page 8,
DKB Mongoose 030 @ 50Mhz, 882 @ 50Mhz, 4MB RAM = $869 AUD
GVP 40Mhz 68040, 4MB RAM = $1399 AUD

Page 34
Amiga 1200, please call. Hint: final price and stock issues.
DKB Mongoose 030 @ 50Mhz, 882 @ 50Mhz = $599 AUD (needs Fast RAM)

Page 82
Warp Engine 040 40Mhz = $2299 AUD
A1200HD = $1245 AUD
A4000T/040 = $4945 AUD
A4000T/060 @ 50 Mhz = $5445 AUD

A1200HD's $1245 + DKB Mongoose 030/882 @ 50Mhz with 4MB RAM's $869 = $2,114 AUD.

Above Pentium 90Mhz in 1995 the PC platform has Pentium 100Mhz, 120Mhz, 133 Mhz, and Pentium Pro 150 Mhz.

For November 1995, the Pentium 90 with PCI motherboard combo has a $799 AUD asking price with sales tax which is compatible with my existing 386DX-33 PC clone.

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

For running PC ports, the only reason why I prefer my retro A1200 with PiStorm is WIMP UI and Paula audio is guaranteed which minimizes PC DOS's audio configuration issues and command line interface.

Originally Posted by Bruce Abbott

7 MHz, but 32 bit, and the ROMs ran at 14MHz.

Lisa has a 14 Mhz clock speed synchronized from 28 Mhz. https://www.amigawiki.org/doku.php?id=enarts:lisa

2 cycle Fast Page's net result is 7 Mhz 32 bit effective which is about 28 MB/s.

Originally Posted by Bruce Abbott

Not a design flaw, but a design decision made to keep the price down and compatibility up. Most other Amiga models only came with ChipRAM in the base configuration for the same reason.

You still defend Commodore's management decisions. LOL. Look at the results...

Stock A1200 didn't deliver a "new 32-bit" 3D gaming experience, hence it competes against SNES's late 16-bit gaming experiences.

From 1993 to 1994, mid-priced gaming PC Doom and its clones delivered a "new 32-bit" 3D gaming experience above SNES's late 16-bit gaming experiences.

Originally Posted by Bruce Abbott

Yes, and other instructions that are faster too, and an instruction cache.

But the 3DO was expensive and could only play games (of which there weren't many - 253 according to Moby Games). Introductory price on release in October 1993 was US$699.99, by which time the A1200 was selling for US$399 or less.

3DO's install base is about 2 million which is more than doubles AGA's under 1 million install base.

Minus inflation, 3DO's $699.99 USD in 1993 mirrors A500's 1987 $699 USD price introduction. 3DO is just a game console instead of a desktop computer.

3DO's quadrilateral-based 3D system mistake is similar to Sega Saturn's quadrilateral 3D system mistake.

Sony's PS1 is a strong competitor with a triangle-based 3D system.

In 1995, 3DO M2's triangle 3D system was demo'ed.

[Bruce Abbott]

Originally Posted by Cyprian

chipsets are clocked at 3.5Mhz and the CPU has access to the chip ram every other chipsets clock. This means that 68EC020 have to wait 8 cycles for each chip memory access.

This doesn't mean anything. Real-world tests show the bandwidth is almost exactly double when running in FastRAM or from ROM, and code tests show a similar difference. So the 020 is effectively running at 7MHz in ChipRAM, and 14MHz in FastRAM and ROM.

There was a multi-million dollar fraud by Commodore Netherland's Bernard van Tienen.

Bernard wasted $480,000 on a very high-class brothel.

There was fraud in the UK too.

Sadly this the kind of thing tends to happen in large organizations. I worked as a telephone exchange technician in the NZ Post Office / Telecoms for 15 years, and saw a bit of it. Worst case I remember was a postmaster who requisitioned a new car and took it home for personal use while the government paid all the costs (and he would have gotten away with it too if wasn't for that meddling temp filling in while his secretary was on leave).

Originally Posted by CCCP alert

Doom was coded slowly and properly for x86 and DOS Protected mode advantages, a true port if it was done for A1200 only would be like Dread, so I don't do the Adoom etc comparisons, YMMV.

Doom was coded for 32 bit 386 protected mode and compiled with Watcom C (which was much faster than other C compilers), with some critical sections coded in optimized x86 asm. An expert coder might be able to do better with 100% optimized asm, but nobody's going to do that. A lot of Amiga games were coded mainly in C too, as were most apps. This makes Doom a good game to use for a comparison of CPU performance, because most of the code is generic and not dependent on specific hardware.

But enough talk, here are the numbers:-

Doom
i386 SX25 (WDC ISA VGA) 3.4 fps
Am386 SX33 (ALi M1217) 3.86 fps
Am386 SX40 (ALi M1217) 4.68 fps

Doom Attack
Amiga A1200 14MHz 020 8Mb Fast (Blitter 020 c2p) 3.6 fps
Amiga CD32 14MHz 020 8MB Fast (Akiko c2p) 5.8 fps

This shows that an A1200 with FastRAM is slightly faster than a 25MHz 386SX despite the significant overhead of c2p. The CD32 is a better comparison because Akiko c2p has much lower overhead. It is 70% faster than a 25MHz 386SX and 24% faster than a 40MHz 386SX.

[sandruzzo]

How much more performance can A1200(blitter, copper, cpu...) gain from just adding FastMem?

[AestheticDebris]

Originally Posted by Bruce Abbott

But enough talk, here are the numbers:-

Doom
i386 SX25 (WDC ISA VGA) 3.4 fps
Am386 SX33 (ALi M1217) 3.86 fps
Am386 SX40 (ALi M1217) 4.68 fps

Doom Attack
Amiga A1200 14MHz 020 8Mb Fast (Blitter 020 c2p) 3.6 fps
Amiga CD32 14MHz 020 8MB Fast (Akiko c2p) 5.8 fps

This shows that an A1200 with FastRAM is slightly faster than a 25MHz 386SX despite the significant overhead of c2p. The CD32 is a better comparison because Akiko c2p has much lower overhead. It is 70% faster than a 25MHz 386SX and 24% faster than a 40MHz 386SX.

Is that the right link for the PC figures? Just takes me to a generic benchmark page with no mention of Doom, not those figures.

Without a clear comparison of how they were configured, I don't think you can really draw many conclusions. The Doom Attack figures, for example, have the audio turned off. Which might be fairer if you're solely focusing on rendering speed, but not if you're concerned with how the game runs in a playable state. Level of detail, window sizes, pixel resolution etc also all contribute and make it hard to compare unless like for like.

[hammer]

Originally Posted by Bruce Abbott

This doesn't mean anything. Real-world tests show the bandwidth is almost exactly double when running in FastRAM or from ROM, and code tests show a similar difference. So the 020 is effectively running at 7MHz in ChipRAM, and 14MHz in FastRAM and ROM.

There was fraud in the UK too.

Sadly this the kind of thing tends to happen in large organizations. I worked as a telephone exchange technician in the NZ Post Office / Telecoms for 15 years, and saw a bit of it. Worst case I remember was a postmaster who requisitioned a new car and took it home for personal use while the government paid all the costs (and he would have gotten away with it too if wasn't for that meddling temp filling in while his secretary was on leave).

Doom was coded for 32 bit 386 protected mode and compiled with Watcom C (which was much faster than other C compilers), with some critical sections coded in optimized x86 asm. An expert coder might be able to do better with 100% optimized asm, but nobody's going to do that. A lot of Amiga games were coded mainly in C too, as were most apps. This makes Doom a good game to use for a comparison of CPU performance, because most of the code is generic and not dependent on specific hardware.

But enough talk, here are the numbers:-

Doom
i386 SX25 (WDC ISA VGA) 3.4 fps
Am386 SX33 (ALi M1217) 3.86 fps
Am386 SX40 (ALi M1217) 4.68 fps

Doom Attack
Amiga A1200 14MHz 020 8Mb Fast (Blitter 020 c2p) 3.6 fps
Amiga CD32 14MHz 020 8MB Fast (Akiko c2p) 5.8 fps

This shows that an A1200 with FastRAM is slightly faster than a 25MHz 386SX despite the significant overhead of c2p. The CD32 is a better comparison because Akiko c2p has much lower overhead. It is 70% faster than a 25MHz 386SX and 24% faster than a 40MHz 386SX.

Not an apple-to-apple comparison when your cited A1200/CD32 has 32-bit Fast RAM and the CPU has a 14 Mhz 32-bit front side bus or 28 Mhz 16-bit equivalent.

The comparable spec PC is 386DX-16 or 386DX-20 with 32-bit front-side bus and ET4000AX or Trident 8900CL.

386SX has a 16-bit front-side bus.

[hammer]

Originally Posted by sandruzzo

How much more performance can A1200(blitter, copper, cpu...) gain from just adding FastMem?

With a matching 32-bit Fast RAM, 14 Mhz 68EC020's 32-bit hardware barrel shifters can operate at their full potential. A1200's 68EC020 is rated at 16 Mhz and a minor 20 Mhz overclock is an easy reach.

Reducing the shared Chip RAM bus bottleneck allows Turrican 2 AGA (with 256 colors VGA art assets) to run at 50 Hz.

1 MB Fast RAM + 2 MB Chip RAM equipped A1200/CD32 allows basic engine and art asset sharing with 3 MB RAM equipped 3DO and a lower gap with 4 MB RAM equipped 386DX PCs. The extra 1 MB Fast RAM is the foot-in-door entry point for CD32/A1200.

In the PC world, AIB graphics vendors sell factory "out-of-the-box" overclocked variants. Ignore the embedded CPU market mindset.