Was anyone else disappointed with the A1200?

[Gorf]

Originally Posted by paul1981

The vast majority of people weren't disappointed then (else they'd reply in here).

The disappointed crowd left the Amiga scene back then and only very few look back and even fewer of these people read this particular thread and probably nobody registers here, just to say "I was disappointed too".

The undeniable truth is: the vast majority of Amiga users left the platform in the early 90s around the time the A1200 got released - many before, because the chipset was so late - and also many after its release, because it did not offer enough.

[paul1981]

The Amiga offered a lot in desktop video. It stuck around throughout the 90's and into 2000's in that role. My memory of PC's all through the 90's and early 2000's was that they couldn't even do simple scrolling (screen savers for example). I suppose these were 'office' machines I saw, but it always fascinated me how they couldn't even scroll text smoothly across the screen without heavy flickering and artefacts.

[Gorf]

Originally Posted by paul1981

That's not my experience. From memory, I'd say a 64 colour AGA Hires (640x256) Workbench is faster than an 8 colour OCS Hires Workbench.

Which still would be quite disappointing.
Consider Moore's Law (with was very true in the 80s and 90s):

The amount of transistors in your average integrated circuit would double every 2.5 years. So between OCS and AGA we have 3 full Moore cycles.
The complexity/capability of a chipset is expected to be 2^3 = 8 times higher.

this would translate to 16 bitplanes being as fast as 2 bitplanes in OSC.

If AGA would have delivered a 65 thousand colour WB in 1992 at the same speed as four colors in 85, I would have no reason to moan.

[paul1981]

The 1200 was a huge upgrade over the 500/600. A 256 colour Hires AGA Workbench screen is still much faster than a Hires OCS 16 Colour Workbench screen.

[Gorf]

Originally Posted by paul1981

A 256 colour Hires AGA Workbench screen is still much faster than a Hires OCS 16 Colour Workbench screen.

No.

[oscar_ates]

Most annoying thing with stock Amiga 1200 was the weak machine unable to play upcoming texture mapping action games and fps genre with decent speeds.

[sokolovic]

Originally Posted by oscar_ates

Most annoying thing with stock Amiga 1200 was the weak machine unable to play upcoming texture mapping action games and fps genre with decent speeds.

Was there many machine available at a £399 price in november 1992 with abilities to play texture mapping action games and FPS genre at a decent speed or is it just another anachronical and unfair judgement made to the Amiga ?

[Gorf]

Originally Posted by sokolovic

Was there many machine available at a £399 price in november 1992 with abilities to play texture mapping action games

The SNES (mode 7) was released in 1991 for $199

[paul1981]

Originally Posted by Gorf

No.

I was comparing the A500/A600 in its 7MHz 68000 form. 256 colour screens are not that slow on the A1200. Not as slow as a 16 colour A500/A600 screen. Hires Workbench screens I'm talking about.

[Gorf]

Originally Posted by paul1981

I was comparing the A500/A600 in its 7MHz 68000 form.

Sure.
in 1992 after 7 years (and 3 Moore cycles) at least the CPU was updated to a model from 1984!
Hurray!

256 colour screens are not that slow on the A1200. Not as slow as a 16 colour A500/A600 screen. Hires Workbench screens I'm talking about.

but the Blitter is still the very same and on a 256 colour AGA WB it has to blit twice as many planes as on a 16 colour OCS WB ... that's why window (re)draw is so slow.
Yes: @256 colours AGA the CPU still gets more access than with 16 colours OCS, but without FastRAM both setups are just unusable.

[Bruce Abbott]

Originally Posted by Gorf

At 640x400 or 640x480 a new PC from 1992 had no problem at 256 colours.
And even if you just got a stupid framebuffer: the gfx-card would not slow down your processing speed, no matter what resolution or depth it supports.
But on a standard A1200 that is the case, because of the lack of FastRAM.

A new PC from 1992 was typically a 386SX with ISA bus video. They were horribly slow in 256 colors. I had a 486SX-25 with 8MB RAM and ISA bus video card. With Windows 95 It was also slow in 256 colors. I also tried Windows 95 on my 386DX-40 with 4MB RAM (a pretty good setup in 1992) in 16 colors. It was slower than an A500.

But hey, let's compare a stock A1200 to the latest high-end PC costing several times the price! PC envy at its finest.

If AGA would have delivered a 65 thousand colour WB in 1992 at the same speed as four colors in 85, I would have no reason to moan.

As usual, you would only have been satisfied if the Amiga was far ahead of any PC of the day - and still only cost £399.

I bet you still wouldn't be satisfied though. Next thing you would be moaning about is how few titles make use of those 65 thousand colours, and the amount of RAM and hard drive space they use, and how slow they run with only a 68060 CPU (all of which would also be in there of course, because Moore's Law).

Note: Moore's law only refers to the number of transistors in a chip, not their complexity. As chips get more complex they become harder to design. So what about R&D? Between 1985 and 1992 Intel's R&D costs went up 400%, from US$195 million to US$780 million. In order to keep up Commodore probably would have needed a similar increase in R&D funding - which simply wasn't possible given their financial position and the way the market was going.

[Gorf]

Originally Posted by Bruce Abbott

A new PC from 1992 was typically a 386SX with ISA bus video.

We had this already - it depends of course, where you lived on this planet in these years.
I do not doubt your experience in New Zealand as a shop owner. But things were different in Germany (own experience) and USA (conclusion).

I remember very well quite a few friends getting VLB equipped PCs in that period of time - despite me showing them my A3000

The 386SX machines reached their very end of shelf live in 92 - you really can not argue that these were somehow the lates technology! Even down under...

But the AGA machines were the latest and greatest and very best Commodore had to offer at the end of this year - don't you see the mismatch?

As usual, you would only have been satisfied if the Amiga was far ahead of any PC of the day - and still only cost £399.

By now you really should know that this argument does not hold any ground with me. I invested heavily in the Amiga platform and was (still am) the proud owner of an A3000 at this time.
And I continued to use this machine (with upgrades) exclusively until 2001.

Note: Moore's law only refers to the number of transistors in a chip, not their complexity.

you can of course just add a lot of transistors for no reason to your chip - but somehow most companies figured out that this makes no sense and that they need to put these transistors to good use.
So the complexity and capability of chips increased about the same speed and the number of transistors it uses!

Between 1985 and 1992 Intel's R&D costs went up 400%, from US$195 million to US$780 million. In order to keep up Commodore probably would have needed a similar increase in R&D funding

Yes: an increase of 400% in R&D for the Amiga sounds about right.
Maybe a little bit more, considering the meager budget it had in the beginning ...
As I mentioned before: after the purchase of Amiga it should have been "all hand on deck" at Commodore for the new platform.

[paul1981]

Originally Posted by Gorf

Sure.
in 1992 after 7 years (and 3 Moore cycles) at least the CPU was updated to a model from 1984!
Hurray!



but the Blitter is still the very same and on a 256 colour AGA WB it has to blit twice as many planes as on a 16 colour OCS WB ... that's why window (re)draw is so slow.
Yes: @256 colours AGA the CPU still gets more access than with 16 colours OCS, but without FastRAM both setups are just unusable.

Technical specs aside, the A1200 feels much faster in Workbench than an A500/A600. mfilos mentions this (he won't use 16 colour OCS Workbench) :

https://eab.abime.net/showpost.php?p...1&postcount=25

A 64 colour Hires Workbench screen on AGA is pretty fast. Of course, throw in FBlit and a decent CPU upgrade and the same can be said for 256 colour screens. This is my experience and perception.

[Gorf]

Originally Posted by paul1981

A 64 colour Hires Workbench screen on AGA is pretty fast. Of course, throw in FBlit and a decent CPU upgrade and the same can be said for 256 colour screens. This is my experience and perception.

I can not argue with your personal perception.
"Results may vary".

[TEG]

I lived those Amiga 1200 versus PC years in the 90' . I was living with a housemate so sharing our financial resources, we managed to have both.

What quickly appeared, was how it was easy to upgrade the PC. We started with a 386 and a few of memory modules but we were able to add more some months later and then, again a few months later, change the processor because the stuff was widely available, just go to the shop at the corner of the street. And it was easy to open the PC box.

More, there was the choice to bought something cheaper than Intel cpu (Ciryx, AMD).

At the time I was in Ray-tracing and the speed of the PC was great compared to the Amiga. I think we bought a mathematical coprocessor with the 386 to do that. The motherboard was ready to receive one.

It was expandable piece by piece and so doable for a low budget.

At the A1200 time, Commodore specialized shops, were closing or no more here. So it was a different story. But the Amiga was still great for games and demos, and it was 90% OCS stuff. To say how it was advanced in its first iteration but it was now a too less and too late "all in one" computer.

[sokolovic]

Originally Posted by Gorf

The SNES (mode 7) was released in 1991 for $199

That's ridiculous. This is a scaling and rotating 2D mode (in planar architecture by the way) Not a 3D one. Plus games using full mode 7 like Pilotwings or Mariokart were using a DSP in their cartdrige.

[TCD]

All this energy spent on discussing what-ifs... Amazing

[Bruce Abbott]

Originally Posted by TEG

What quickly appeared, was how it was easy to upgrade the PC. We started with a 386 and a few of memory modules but we were able to add more some months later and then, again a few months later, change the processor because the stuff was widely available, just go to the shop at the corner of the street. And it was easy to open the PC box.

More, there was the choice to bought something cheaper than Intel cpu (Ciryx, AMD).

At the time I was in Ray-tracing and the speed of the PC was great compared to the Amiga. I think we bought a mathematical coprocessor with the 386 to do that. The motherboard was ready to receive one.

I assume it was a 386DX, because you generally couldn't upgrade the CPU on an SX motherboard. All the 386DX systems I remember were ISA bus only, and RAM was only upgradable to perhaps 16MB using 30 pin SIMMs (4 at a time).

In 1992 most 386DXs already had a 40MHz AMD processor (not Intel, the i386 stopped at 33MHz). A Cyrix 486DLC would give a modest speed improvement, but that was it. You weren't going to get 486 or Pentium level performance. You weren't going to get 128MB of RAM or high performance graphics.

3D game requirements quickly outstripped the power of even the fastest PC, soon requiring the purchase of a fancy hardware 3D graphics card as well. Once PC gamers got on that treadmill they never stopped. This phenomenon was mostly confined to PCs - if you purchased an Amiga 1200 or a Sony PlayStation you didn't expect games for it demanding a faster CPU, more RAM and a new graphics chipset.

The A1200 certainly cost more to upgrade, but was a lot cheaper to start with. That was a good thing because games would be designed to work on a stock machine, not just the latest most powerful model. For any genre except texture mapped 3D that was fine (and necessary considering the small user base).

By 1983 a number of RAM boards with FPU socket were out for the A1200. By 1994 several accelerator cards were out that had or could take up to a 50MHz 030 (20% faster than a 386DX-40) with SIMM sockets for up to 128MB of RAM. Some also had a SCSI option. In later years these cards got cheaper, and 040 and 060 cards started to appear. This continued until today when you can get ludicrous speed combined with RTG or 'Super AGA' for a few hundred dollars, proving that there was no limit to how far an A1200 could be expanded.

As for how easy it was to upgrade, I don't think you can beat flipping off the trapdoor hatch and plugging in a card. Putting SIMMs in a PC might have been easy enough, but the CPU was a bit trickier with all those delicate pins. Often you had to remove several cards and push cabling out of the way just to get to the socket. Of course the more you fiddled with the machine the more likely it was that something would go horribly wrong.

[koobo]

Originally Posted by Thomas Richter

You are forgetting something: For your average blit, that just not possible because the blitter is triggered once for each plane involved, and the CPU just sits and busy-waits until the current plane is done before the next plane is triggered. BltBitMap() can, in some cases, for interleaved bitmaps, get away with a single blit. For example, if it just copies data from A to B.

What if all of the graphics operations along with the display architecture had been designed to work with interleaved bitplanes in the first place? Would that have been feasible, how much swoosh would that have brought for the Workbench user?

[Promilus]

This continued until today when you can get ludicrous speed combined with RTG or 'Super AGA' for a few hundred dollars, proving that there was no limit to how far an A1200 could be expanded.

Do you honestly claim A1200 could've been expanded that way by design (original Commodore design)? And 386 PC absolutely could not? RTG and SAGA (and most new and powerful stuff out there) is not reliant on any portion of A1200 architecture. It's off the board... entirely on turbo. Either in FPGA or SoC. I already said that and I guess I have to again - should you create FPGA attached to 386 signals and emulate those, then implement 2 issue pipelined out of order x86 softcore and add pci interface it'll be the same kind of upgrade for those machines as Vampire is. And it is possible. Just not very practical. And that kind of impractical designs are the only thing left for classic 68k Amiga.