Was anyone else disappointed with the A1200?

[Cyprian]

Originally Posted by sokolovic

The Falcon was much more expensive than the A1200. There is a reason why nobody bought one back then.

~30k Falcon buyers is a bit more than nobody (A1200 sold in 120k units in comparison vs A500 ~3~4 millions)

Originally Posted by sokolovic

That's interesting btw how Commodore and Atari switched their policy between the ST/A1000 and the A1200/Falcon030.

actually IMO nothing changed in that policy. Atari computers were focused on semi-pro market and for gaming /|\ had a lot of consoles (A2600/A5200/A7800/XEGS/Lynx/Jaguar)

In 1985 Amiga had its advantages in gaming area and Atari had its advantages in semi-pro (and there were areas where they were ahead of the competitors).
and in 1992 was the same.

[sokolovic]

Originally Posted by Cyprian

~30k Falcon buyers is a bit more than nobody (A1200 sold in 120k units in comparison vs A500 ~3~4 millions)

Ok, that's ridiculous. This claims of A1200 poors sales have been several times dismantled here.
You're overexagerrating the Falcon sales and diminushing the A1200 one. There is a reason why Falcon are so rare today, even in comparison to A1200
Yep, nobody bought the Falcon. It was discontinued a year after it was introduced, and it wasn't because Atari went bankrupt. At the same time, the A1200 was selling strongly, at least in Europe, and obviously it didn't killed Commodore. In fact Commodore could'nt even met the demand for it.

You Can also compare software libraries on both machines. The A1200 one isn't as great as it could have been but it is vastly bigger by magnitude than the Falcon one.

That doesn't change the quality of this machine but clearly, Commodore made a better choice for their computer back then, because it actually had some public and attracted publishers.

Edit around 12000 Falcon030 solds and only 20 000 total PCB produced by Atari before discontinuation.
https://comp.sys.atari.st.narkive.co...atistics#post3

Here we have another estimation a bit better but still only 14 000 at best by the end of 1993, just before it was discontinued
https://stcarchiv.de/stc1993/10/inte...ri-bob-gleadow

That's "just" at least more than two times lower than your estimation (which was already low btw).

[AestheticDebris]

Originally Posted by Cyprian

But when you compare it to A1200 or A4000 then Falcon looks very good:

Architecture:
- Amiga: Address range: 24bit; Data bus: 32bit
- Atari: Address range: 24bit; Data bus: 32bit

The Falcon was a terrible design, it's like Atari didn't know what it wanted to be. They crippled it by trying to keep some compatibility with the ST, but not enough to actually justify doing so.

And, FWIW, it's data bus was only 16-bit. Which was an even dumber decision that leaving half of AGA as 16-bit.

[Cyprian]

Originally Posted by AestheticDebris

The Falcon was a terrible design, it's like Atari didn't know what it wanted to be. They crippled it by trying to keep some compatibility with the ST, but not enough to actually justify doing so.

terrible because of what? better video/audio/performance?

Originally Posted by AestheticDebris

And, FWIW, it's data bus was only 16-bit. Which was an even dumber decision that leaving half of AGA as 16-bit.

Falcon bus was full 32bit, e.g. because 16bit color modes need 32MB/s transfer . It is visible on the schematics, the Videl pinout or RAM extension port pinout.

But true is that the Falcon CPU has 16bit access but it was a twice faster than in A1200. As we can see from tests bus bandwidth isn't worst than in A1200/A4000:

RAM Bus Bandwidth for CPU 32bit access:
- Amiga: Read: 4.5MB/s Write: 6.9MB/s (2 chipset cycles 3.5MHz per 32bit) (BusSpeedTest 0.19)
- Atari: Read: 5.4MB/s Write: 6.5MB/s (2 chipset cycles 8MHz per 16bit)

RAM Bus Bandwidth for CPU 16bit access:
- Amiga: Read: 2.2MB/s Write: 3.5MB/s (2 chipset cycles 3.5MHz per 32bit)
- Atari: Read: 5.4MB/s Write: 6.5MB/s (2 chipset cycles 8MHz per 16bit)

[hammer]

Originally Posted by Cyprian

terrible because of what? better video/audio/performance?



Falcon bus was full 32bit, e.g. because 16bit color modes need 32MB/s transfer . It is visible on the schematics, the Videl pinout or RAM extension port pinout.

But true is that the Falcon CPU has 16bit access but it was a twice faster than in A1200. As we can see from tests bus bandwidth isn't worst than in A1200/A4000:

RAM Bus Bandwidth for CPU 32bit access:
- Amiga: Read: 4.5MB/s Write: 6.9MB/s (2 chipset cycles 3.5MHz per 32bit) (BusSpeedTest 0.19)
- Atari: Read: 5.4MB/s Write: 6.5MB/s (2 chipset cycles 8MHz per 16bit)

RAM Bus Bandwidth for CPU 16bit access:
- Amiga: Read: 2.2MB/s Write: 3.5MB/s (2 chipset cycles 3.5MHz per 32bit)
- Atari: Read: 5.4MB/s Write: 6.5MB/s (2 chipset cycles 8MHz per 16bit)

A1200's CPU max access into Chip RAM is 7.1MB/s.

[Cyprian]

Originally Posted by sokolovic

Ok, that's ridiculous. This claims of A1200 poors sales have been several times dismantled here.
You're overexagerrating the Falcon sales and diminushing the A1200 one.

A1200 sale figures are taken from EAB, maybe it would be worth to create a new thread and clarify that.

Originally Posted by sokolovic

There is a reason why Falcon are so rare today, even in comparison to A1200
Yep, nobody bought the Falcon. It was discontinued a year after it was introduced, and it wasn't because Atari went bankrupt.

Atari unlike Commodore was a small company, actually family company. Their cash flow was much smaller than Commodore's.
That's why they canceled the entire (not only Falcon's one) computer department because they decided to focus on Jaguar. Either way, they sold the rights to Falcon to C-LAB company, which produced it for next few years.

Atari didn't go bankrupt. In 1995, a year later, after Commodore's bankrupcy, Tramiel decided to close the company. He reverse merged Atari with JTS to recover the family money.

Originally Posted by sokolovic

You Can also compare software libraries on both machines. The A1200 one isn't as great as it could have been but it is vastly bigger by magnitude than the Falcon one.

Do you mean application library? Do you have that A1200 application library list?

[hammer]

Originally Posted by Cyprian

~30k Falcon buyers is a bit more than nobody (A1200 sold in 120k units in comparison vs A500 ~3~4 millions)

FALSE.



For UK's AGA platform 1993 Xmas sales: A1200 excess of 160,000 and CD32 has 70,000. Atleast 230,000 units for AGA platform from the UK's XMas sales.

Another 44,000 A1200s for Oct - Dec 1992,
Another 100,000 A1200 for non-Xmas Jan-Aug 1993. Commodore UK is the strongest Amiga AGA market. http://www.bambi-amiga.co.uk/amigahistory/sales.html

UK has at least 374,000 AGA unit sales.

For Germany's AGA platform unit sales, http://www.bambi-amiga.co.uk/amigahistory/sales.html

A1200, 95,500
CD32, 25,000
A4000/040, 3,800
Germany's total: 124,300

From Germany and UK, there's at least 498,300 AGA units.

Another 20,000 units under Escom. AGA platform has an excess of 518,300 units.

Commodore UK, Commodore Germany and Commodore Canada are Commodore's strong divisions. No AGA sales data for Commodore Canada.

My 650,000 AGA units estimate encapsulates the excess 518,300 units.

There's "more than 10,000 Vampires" in modern times.

There's market demand from A500 retro owners for AGA Amigas, hence AGA clones are in development. The ARM CPU is the exit from Motorola's influence and reboot the Amiga platform back into mainstream. Retro is the entry vector for Amiga's return into mainstream.

After PiStorm-Emu68 gained it's modern Pi WiFi WPA2 driver, my A1200 has improved Windows RDP use case. PiStorm-Emu68 with RPi 4B has the fastest 68K accelerator for C= Amiga platform and it gained modern WiFi. This enabled my A1200 to be my daily driver.

[Cyprian]

Originally Posted by hammer

A1200's CPU max access into Chip RAM is 7.1MB/s.

are you sure? can you post a screenshot?
Below you can see A1200 BusSpeed test:


And A4000 with A3630 (68EC030@25):

Originally Posted by hammer

For UK's AGA platform has A1200 excess of 160,000
...
For Germany:
A1200, 95,500.

ok, thanks for clarification.
BTW cool page: http://www.bambi-amiga.co.uk/amigahistory/sales.html

[sokolovic]

Originally Posted by Cyprian

.
Do you mean application library? Do you have that A1200 application library list?

No, I mean software library. That includes application but not only, obviously. What would I limite the A1200 library to just apps ? Why don't even narrow it to just music apps that use a DSP to serve your biased PoV ?

[hammer]

Originally Posted by Cyprian

are you sure? can you post a screenshot?
Below you can see A1200 BusSpeed test:


And A4000 with A3630 (68EC030@25):

Amiga 4000/030 BusSpeedTest, using motherboard's 68030 instead of A3630 card.
https://groups.google.com/g/comp.sys...m/aMw2s4s1dWQJ

And the same from an A4000/030 -- again, only testing motherboard memory.

11.Ram Disk:> nuke0:system/c/bustest chip fast
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
fast $077E0000 readw 185.0 ns normal 10.8 * 10^6 byte/s
fast $077E0000 readl 294.2 ns normal 13.6 * 10^6 byte/s
fast $077E0000 readm 272.9 ns normal 14.7 * 10^6 byte/s
fast $077E0000 writew 248.2 ns normal 8.1 * 10^6 byte/s
fast $077E0000 writel 248.2 ns normal 16.1 * 10^6 byte/s
fast $077E0000 writem 209.2 ns normal 19.1 * 10^6 byte/s
chip $00028000 readw 538.9 ns normal 3.7 * 10^6 byte/s
chip $00028000 readl 573.3 ns normal 7.0 * 10^6 byte/s
chip $00028000 readm 629.0 ns normal 6.4 * 10^6 byte/s
chip $00028000 writew 575.0 ns normal 3.5 * 10^6 byte/s
chip $00028000 writel 574.8 ns normal 7.0 * 10^6 byte/s
chip $00028000 writem 573.9 ns normal 7.0 * 10^6 byte/s

--------
7.1 MB/s can be reached when the display is RTG.


Fast RAM equipped A1200/A4000 can have active dual 32-bit bus at the same time i.e. it's "64 bits" under Atari Jaguar marketing standard.

[hammer]

double post.

[Cyprian]

Originally Posted by hammer

Amiga 4000/030 BusSpeedTest, using motherboard's 68030 instead of A3630 card.
https://groups.google.com/g/comp.sys...m/aMw2s4s1dWQJ

And the same from an A4000/030 -- again, only testing motherboard memory.

11.Ram Disk:> nuke0:system/c/bustest chip fast
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
chip $00028000 readw 538.9 ns normal 3.7 * 10^6 byte/s
chip $00028000 readl 573.3 ns normal 7.0 * 10^6 byte/s

chip $00028000 writew 575.0 ns normal 3.5 * 10^6 byte/s
chip $00028000 writel 574.8 ns normal 7.0 * 10^6 byte/s

good to know.
BTW, I wonder why someone would like to use A3630 when the motherboard CPU has better performance.


I've checked your link and see interesting figures for A4000/040:

ere's some results from an A4000/040:
BusSpeedTest 0.19 (mlelstv)   Buffer:     262144 Bytes, Alignment: 32768
========================================================================
memtype   addr       op         cycle     calib         bandwidth

chip      $001B8000  readw     860.9 ns   normal       2.3 * 10^6 byte/s
chip      $001B8000  readl     861.0 ns   normal       4.6 * 10^6 byte/s


chip      $001B8000  writew    863.0 ns   normal       2.3 * 10^6 byte/s
chip      $001B8000  writel    862.7 ns   normal       4.6 * 10^6 byte/s

Originally Posted by hammer

--------
7.1 MB/s can be reached when the display is RTG.

obviously,
in that case there is no need slowdown the CPU by touching ChipRAM

Originally Posted by hammer

Fast RAM equipped A1200/A4000 can have active dual 32-bit bus at the same time i.e. it's "64 bits" under Atari Jaguar marketing standard.

Jag has a real 64bit bus not marketing one, and only BFG9060 or TF1260 comes close to its bus performance.

[hammer]

Originally Posted by Cyprian

You can't use HAM in Workbench or inside your own GUI window.
Regarding games - I'm not really interested in this area, but I know only one game preview (shown here this year - really nice screen) with HAM but sprites there are static.

For AGA, HAM driver can be used for Workbench GUI e.g. https://aminet.net/package/driver/moni/WBHacksAGA

Amiga's HAM is considered to be lossy color compression. Modern PC GpGPU includes lossy color compression modes, when extracting every additional performance, a hardware cheat is used.

[hammer]

Originally Posted by Cyprian

good to know.
BTW, I wonder why someone would like to use A3630 when the motherboard CPU has better performance.

EC030 variant on A3630 card with A4000 was a cost-cutting from Commodore.

Originally Posted by Cyprian

I've checked your link and see interesting figures for A4000/040:

ere's some results from an A4000/040:
BusSpeedTest 0.19 (mlelstv)   Buffer:     262144 Bytes, Alignment: 32768
========================================================================
memtype   addr       op         cycle     calib         bandwidth

chip      $001B8000  readw     860.9 ns   normal       2.3 * 10^6 byte/s
chip      $001B8000  readl     861.0 ns   normal       4.6 * 10^6 byte/s


chip      $001B8000  writew    863.0 ns   normal       2.3 * 10^6 byte/s
chip      $001B8000  writel    862.7 ns   normal       4.6 * 10^6 byte/s

obviously,
in that case there is no need slowdown the CPU by touching ChipRAM

Most game consoles (e.g. 3DO, Saturn, PS1, SNES, Mega Drive) and gaming PC competitors in the 1990s have discrete memory pools.

For performance with gaming, the CPU and GPU should stay in their own lanes.

PS4 Pro restored discrete 1 GB DDR3 and 8 GB DDR5 memory pools. PS4's UMA experience wasn't ideal for pure gaming performance.

PS5 has 512 MB DDR4 and 16 GB GDDDR6 discrete memory pools.

UMA is not for focused gaming performance.

Xbox 360 had 10 MB EDRAM and 512 MB GDDR3 discrete memory pools.

Originally Posted by Cyprian

Jag has a real 64bit bus not marketing one, and only BFG9060 or TF1260 comes close to its bus performance.

Note that PS1 has a discrete system and video memory buses.

PS; I have both TF1260 and PiStorm32. The other users have Vampires.

[Cyprian]

Originally Posted by hammer

EC030 variant on A3630 card with A4000 was a cost-cutting from Commodore.

ok

Originally Posted by hammer

PS; I have both TF1260 and PiStorm32. The other users have Vampires.

PiStorm and Vampire are great

Originally Posted by hammer

Note that PS1 has a discrete system and video memory bus.

PS1 is a different era (3 years after Jag, A1200 and Falcon) and yes, it had a way better architecture than previous consoles like Jaguar or CD32.
It had 32MHz 32bit bus, it is 132MB/s for the CPU: vs 4.5/6.9MB/s in A1200/CD32 or 5.4/6.5MB/s in Falcon

[hammer]

Originally Posted by Cyprian

ok


PiStorm and Vampire are great



PS1 is a different era (3 years after Jag, A1200 and Falcon) and yes, it had a way better architecture than previous consoles like Jaguar or CD32.
It had 32MHz 32bit bus, it is 132MB/s for the CPU: vs 4.5/6.9MB/s in A1200 or 5.4/6.5MB/s in Falcon

PS1 was released in Q4 1994 in Japan.
Jag was released in Nov 1993.

3DO has discrete memory pools.

Jag uses four 16-bit 80 ns fast page memory chips i.e. it's not SRAM cache quality.

A1200's CPU link with Chip RAM is only a fraction of total Chip RAM bandwidth dictated by Budgie i.e. Lisa has up to 28 MB/s.

[Bruce Abbott]

Originally Posted by Cyprian

Falcon bus was full 32bit, e.g. because 16bit color modes need 32MB/s transfer . It is visible on the schematics, the Videl pinout or RAM extension port pinout.

But true is that the Falcon CPU has 16bit access but it was a twice faster than in A1200. As we can see from tests bus bandwidth isn't worst than in A1200/A4000:

RAM Bus Bandwidth for CPU 32bit access:
- Amiga: Read: 4.5MB/s Write: 6.9MB/s (2 chipset cycles 3.5MHz per 32bit) (BusSpeedTest 0.19)
- Atari: Read: 5.4MB/s Write: 6.5MB/s (2 chipset cycles 8MHz per 16bit)

RAM Bus Bandwidth for CPU 16bit access:
- Amiga: Read: 2.2MB/s Write: 3.5MB/s (2 chipset cycles 3.5MHz per 32bit)
- Atari: Read: 5.4MB/s Write: 6.5MB/s (2 chipset cycles 8MHz per 16bit)

So what? The fact is that the Falcon was crippled in a stupider way than the A1200, which didn't escape the notice of fans.

All low-end Amigas were supplied with ChipRAM only to keep the price down. The AGA chipset runs at the same 3.58MHz speed as OCS for compatibility, but the CPU is not so limited. The A1200 runs ROM code at full speed, and RAM too with an expansion board. Put a CPU on the expansion board and there's no limit.

Originally Posted by AestheticDebris

FWIW, it's data bus was only 16-bit. Which was an even dumber decision that leaving half of AGA as 16-bit.

FWIW, leaving 'half of' AGA as 16-bit was not a 'dumb' decision - it was a necessary decision. AGA was late enough as it was. The dumb decision was pouring resources into a chipset that was beyond their capabilities (AAA). They should have started with AGA, then improved on that in the next round.

[hammer]

Originally Posted by Cyprian

ok
PS1 is a different era (3 years after Jag, A1200 and Falcon) and yes, it had a way better architecture than previous consoles like Jaguar or CD32.
It had 32MHz 32bit bus, it is 132MB/s for the CPU: vs 4.5/6.9MB/s in A1200/CD32 or 5.4/6.5MB/s in Falcon

A1200's CPU-to-Chip RAM link still delivers 320x200 256 colors at >50 fps.

For example
[ Show youtube player ]
Star Wars Dark Forces 68K port running on Amiga 1200's AGA raster display with PiStorm32-Lite Emu68 CPU accelerator.

This Star Wars Dark Forces 68K port runs on open-source The Force Engine which is intended to be run on modern PCs.

A1200's Budgie has baked in Fast RAM memory controller support i.e. missing DRAM chips and the intended DSP3210.

[Cyprian]

Originally Posted by Bruce Abbott

So what? The fact is that the Falcon was crippled in a stupider way than the A1200, which didn't escape the notice of fans.

crippled by what? better video, audio, bus performance, cpu performace or dsp performance?

[hammer]

Originally Posted by Bruce Abbott

So what? The fact is that the Falcon was crippled in a stupider way than the A1200, which didn't escape the notice of fans.

All low-end Amigas were supplied with ChipRAM only to keep the price down. The AGA chipset runs at the same 3.58MHz speed as OCS for compatibility, but the CPU is not so limited. The A1200 runs ROM code at full speed, and RAM too with an expansion board. Put a CPU on the expansion board and there's no limit.

FWIW, leaving 'half of' AGA as 16-bit was not a 'dumb' decision - it was a necessary decision. AGA was late enough as it was. The dumb decision was pouring resources into a chipset that was beyond their capabilities (AAA). They should have started with AGA, then improved on that in the next round.

AGA itself wasn't that late, but Gayle (PCMCIA and IDE mandates) and Budgie (PCMCIA link with Gayle) were late.

The original intent for AGA's object manipulator was $20 DSP3210 with Fast RAM.

A1200's Budgie has Fast RAM controller support baked in which is missing on A500's Zorro I edge connector. Lew Eggebrecht intended to restore the original AGA+DSP3210 bundle and Lew made sure the $20 argument is focused on.

Bill "IBM PCJr" Sydnes made A1200 to be another JR.