Was anyone else disappointed with the A1200?

[hammer]

Originally Posted by babsimov

I'm not saying that this MAC wasn't more powerful than the AGA generation. Just that I don't feel like the DSP inside was used to its full potential. If Apple had not limited the DSP to a single model, perhaps things would have been different on this point. But that's not even certain, since the target audience for the MAC was very different from that of the Falcon or the 1200. Of course, all of this is just a guess on my part.

Quadra 840AV's DSP3210 @ 66 Mhz with 68040 @ 40Mhz and 660AV has DSP3210 @ 55Mhz with 68040 @ 25 Mhz. They are stop gaps to increase the MFLOPS marketing numbers when Intel released Pentium 60/66 in 1993.

They were replaced by PowerMacs.

Originally Posted by babsimov

I read the DSP 3210 specifications a while ago, at the time I initiated discussion with the Falcon community about DSP.
Of course, I am convinced that this DSP would have brought a significant gain to AGA. This is why I sought to learn more about this DSP and also about what a DSP could provide.
So I went to the community that i see the closest to the Amiga community feeling and which was lucky enough to have a DSP as standard, the Falcon community.

Well, Atari fans labeling FP32 as scientific.

AT&T marketed DSP3210 as "3D and multimedia DSP".

FP64 is the scientific FP. FP32 is usually for game and multimedia FP e.g. SSE's or Altivec's quad packed FP32, and PC gaming GpGPUs.

[dreadnought]

Originally Posted by hammer

Amiga has a game console nature with a keyboard, a mouse, and a game controller.

Amiga's graphics chipset transition into the next generation is like a games console, unlike gaming PC's upgradable graphics adapter modularity.

Amiga's nothing like a game console, it's a micrcomputer with advanced fx capabilities. In the Eighties this was normal design, the modularity became popular with the flood of cheap components from Asia later on (and spelled the death toll for all the micros).

[Thorham]

Originally Posted by hammer

FP64 is the scientific FP.

Science uses whatever precision is appropriate, including 16 bit and less (minifloat).

[babsimov]

Originally Posted by hammer

Quadra 840AV's DSP3210 @ 66 Mhz with 68040 @ 40Mhz and 660AV has DSP3210 @ 55Mhz with 68040 @ 25 Mhz. They are stop gaps to increase the MFLOPS marketing numbers when Intel released Pentium 60/66 in 1993.

They were replaced by PowerMacs.

Which therefore explains why the DSP was only used by Apple on this model, if I'm not mistaken.

Well, Atari fans labeling FP32 as scientific.

AT&T marketed DSP3210 as "3D and multimedia DSP".

FP64 is the scientific FP. FP32 is usually for game and multimedia FP e.g. SSE's or Altivec's quad packed FP32, and PC gaming GpGPUs.

Well if we refer to the origin of DSP on Amiga, a fan of astronomy and atmospheric phenomena. Besides, this Amiga user, passionate about science, why did he choose the DSP 32 rather than the DSP 56001 at the time ? Perhaps precisely because this DSP could do floating point calculations well enough for the scientific calculations he wanted to do at home. I would say that the Falcon DSP connoisseurs point of view seems relatively correct about the DSP 3210 at the time. Especially since an Amiga 4000 coupled with several DSP 3210 will be used in the ATH 1000, an ultrasound machine for medical usage. These two areas seem more scientific to me than what is shown with the Falcon DSP.

Their point was above all to say that at the time the choice of the AT&T DSP was questionable, since it was the Motorola 56001 which was the best known in computing. And from their point of view, it was also much better documented for its programming.
At the same frequency, the 56001 develops more MIPS. This is why they explained to me that a 56001 at 32 MHz was equivalent in MIPS to a DSP 3210 at 50/60 MHz. So of course, the DSP 3210 had floating point calculation which the 56001 did not have. I'm not saying they're right about everything, just that some of the things they say seems relatively consistent. And, to my knowledge, these demomakers are the few who have really looked into the best use of a DSP on a computer, I mean to make it do what we didn't expect.
What would be interesting is to know the price at the time of the 56001 in comparison to the DSP 3210 for example.
Here is an article from 1992, from a French magazine which discussed the main points of the DSP. It says that the 56001 was the industry standard at the time.
http://obligement.free.fr/articles/dsp.php


And, as we have seen, the Commodore teams did not really look very far to decide which DSP to use in the Amiga. It seems that without the luck of this Amiga user with his DIY DSP 32 card and the fact that he knew someone related to Commodre, DSP technology would perhaps not even have been considered for the Amiga.

But, as I have already said, I do not deny that the DSP 3210 would have been a significant contribution to the AGA and I even hoped that Commodore would then release the DSP card for the 1200.

[sokolovic]

Originally Posted by AestheticDebris

But 241 colours is without any tricks whatsoever, in the most common SNES mode. As I said, practical realities will tend towards duplicating some colours across the various palettes and so forth. So 150+ colours on screen is a more conservative estimate. If you pull out all the stops and use every trick in the book, well then you get Donkey Kong Country level graphics.

And that's part of why the A1200 was so disappointing for gamers. You can argue back and forth over the A500 and MegaDrive, each having different strengths and weaknesses from a graphical perspective. But the A1200, despite being it's contemporary, was such a long way behind what the SNES was capable of.

My point was you cannot just say that 256 colors is the norm on the SNES for games.
This is a theoretical number that the console most of the Time wasn't able to reach because it was costly hardware wise and even financially because of the ROM size.

If the you take the 256 colors number as the standard on the SNES, then you should take the 262000 number as a standard on the AGA.

[Tigerskunk]

Originally Posted by sandruzzo

@hammer

Sorry Pal, but you're hammering this thread with the same exact datas: what's the point?

The point is people on the spectrum posting objective numbers in a thread about the subjective feeling if you felt disappointed by something that happened 31 years ago.

[pandy71]

Originally Posted by hammer

I'm not VCDECIDE.

Well... as you providing detailed numbers then i'm curious how your methodology looks...

[pandy71]

Originally Posted by babsimov

Well if we refer to the origin of DSP on Amiga, a fan of astronomy and atmospheric phenomena. Besides, this Amiga user, passionate about science, why did he choose the DSP 32 rather than the DSP 56001 at the time ? Perhaps precisely because this DSP could do floating point calculations well enough for the scientific calculations he wanted to do at home. I would say that the Falcon DSP connoisseurs point of view seems relatively correct about the DSP 3210 at the time. Especially since an Amiga 4000 coupled with several DSP 3210 will be used in the ATH 1000, an ultrasound machine for medical usage. These two areas seem more scientific to me than what is shown with the Falcon DSP.

Their point was above all to say that at the time the choice of the AT&T DSP was questionable, since it was the Motorola 56001 which was the best known in computing. And from their point of view, it was also much better documented for its programming.
At the same frequency, the 56001 develops more MIPS. This is why they explained to me that a 56001 at 32 MHz was equivalent in MIPS to a DSP 3210 at 50/60 MHz. So of course, the DSP 3210 had floating point calculation which the 56001 did not have. I'm not saying they're right about everything, just that some of the things they say seems relatively consistent. And, to my knowledge, these demomakers are the few who have really looked into the best use of a DSP on a computer, I mean to make it do what we didn't expect.
What would be interesting is to know the price at the time of the 56001 in comparison to the DSP 3210 for example.
Here is an article from 1992, from a French magazine which discussed the main points of the DSP. It says that the 56001 was the industry standard at the time.
http://obligement.free.fr/articles/dsp.php

AT&T DSP was cheapest one and additionally supporting floats - closest competitor offering floats was TMS320C3x family - approx 10..20 times more expensive...
Also strongly disagree with claim that 56k was industrial standard - industrial standard those times was fixed point Texas Instruments TMS320Cxx family, probably also NEC. Float DSP's was never popular from many reasons. Motorola with DSP56k was relatively late on DSP market - 56k was mostly popular in audio applications...

Atari probably selected 56k due audio, perhaps also this can be part combined deal with 68030.

Apple selection of AT&T was probably price driven (similarly to Commodore).
Having 3210 on Amiga board means that for 20$ you had float performance 20..100 higher than 68882.

[Locutus]

Originally Posted by Tigerskunk

The point is people on the spectrum posting objective numbers in a thread about the subjective feeling if you felt disappointed by something that happened 31 years ago.

This is absolutely the greatest summary of this thread ever.

[Promilus]

Originally Posted by pandy71

AT&T DSP was cheapest one and additionally supporting floats - closest competitor offering floats was TMS320C3x family - approx 10..20 times more expensive...
Also strongly disagree with claim that 56k was industrial standard - industrial standard those times was fixed point Texas Instruments TMS320Cxx family, probably also NEC. Float DSP's was never popular from many reasons. Motorola with DSP56k was relatively late on DSP market - 56k was mostly popular in audio applications...

Atari probably selected 56k due audio, perhaps also this can be part combined deal with 68030.

Apple selection of AT&T was probably price driven (similarly to Commodore).
Having 3210 on Amiga board means that for 20$ you had float performance 20..100 higher than 68882.

68882 - you compile your 68k with hard floats which goes to 68882 through hardware interface between CPU and FPU. You have nothing to worry, just write your 68k code as usual... Adding DSP to the mix ain't that straightforward and you know it! The flow of the program has to be maintained by 68k and you have to add somewhere DSP code then common memory pool to exchange data. Peak MFLOPS numbers only looks great on paper. Show me actual software which is easy to develop and takes advantage of that? Nothing? We have 600MHz ARM Cortex A7 for few years now. ZZ9000 card. Look how many great apps uses that for acceleration mixed with 68k code. Aaahhhh... exactly.

[Gorf]

Originally Posted by Promilus

We have 600MHz ARM Cortex A7 for few years now. ZZ9000 card. Look how many great apps uses that for acceleration mixed with 68k code. Aaahhhh... exactly.

Strange argument, since we actually do have this kind of acceleration with PPC datatypes for ages or now brand-new on the A600GS with ARM.

This might be a special problem of the ZZ9000 (context switches? not really on the same bus?) or due to it not being widespread to put it mildly.

And not only this:
there IS actually software using the DSP running on the reimplemented (A)A3000+ despite only a handful exist.

[Gorf]

Originally Posted by Bruce Abbott

An A1200 with 14MHz 020 and DSP chip could achieve similar results to putting a 50MHz 030 in it - except the 030 would accelerate everything, not just the few things suited to the DSP. And the same stuff would run on any other Amiga with a fast CPU (even better if its CPU was faster). Furthermore - perhaps more importantly - you could port games from that other platform (the PC) with minimal effort.

A 50MHz 030 did not exist at the time.
You would need to use a 040 @25MHz or @33MHz - which would be factor 5-10 more expensive than 14MHz 020 and DSP combo.

And as for the importance of porting games:
If you are reliant on ports and not longer the primary target, your grave as a platform is already dug out.

Luckily in these years (90-92) the Amiga finally got mostly rid of cheap ST ports and was getting more and more exclusive titles or at least titles that were developed for the Amiga first and ported to other platforms later.
But sadly AGA machines could not hold on to this trend for the reasons discussed in this thread.

[Gorf]

Originally Posted by Amigajay

The A570 drive was planned even before the CDTV was released and Commodore were talking about it and ‘not letting A500 owners down’ with it not coming out, and was shown as the A690 drive at CES 1991 so this theory is not true about the drives.

Thanks for the information.

Announcing an A500 CD upgrade that early might have been also a mistake that did hurt the CDTV sales quite a bit.
Why buy a CDTV if you have already a A500?
Or why not buy a A500 now and get the A570 later to have the best of both worlds?

Was it really worth to appear as a nice company that cares about its A500 customers in this case?

It seems especially odd since the CDTV was not marketed as an Amiga at first - they even tried very hard to avoid the Amiga image in the first year and only resorted to that route after the initial campaign flopped.
So why would you announce an A500 CD upgrade, if the CDTV was supposed to be a totally different animal?

That C= still kept that early promise after the CDTV production was long canned and even the A500(plus) was gone, underlines my claim, that they must have been under some contractual pressure.

Whatever it was, the plan did not really work out either:
only around 20k units of the A570 were ever sold.
(partly because you needed a A500+ with 1MB ChipRam to be really compatible with many CDTV titles)

Also don’t forget another CDTV canned project, the add-on card the ‘AVM’!

I could not find any information if the AVM module (Advanced Video Mode) did exist even as a prototype. It was supposed to support more colours, probably in a similar way HAM-E or DCTV do, and Photo CD compatibility.

[pandy71]

Originally Posted by Promilus

68882 - you compile your 68k with hard floats which goes to 68882 through hardware interface between CPU and FPU. You have nothing to worry, just write your 68k code as usual... Adding DSP to the mix ain't that straightforward and you know it! The flow of the program has to be maintained by 68k and you have to add somewhere DSP code then common memory pool to exchange data. Peak MFLOPS numbers only looks great on paper. Show me actual software which is easy to develop and takes advantage of that? Nothing? We have 600MHz ARM Cortex A7 for few years now. ZZ9000 card. Look how many great apps uses that for acceleration mixed with 68k code. Aaahhhh... exactly.

Yep, true, you compile code and at best you get 0.2MFLOPS - probably not enough even to think about some real time processing...
Quickly after introducing DSP you can expect software using it especially if DSP is standard HW present in every machine (uncommon thing for 68881/68882 yet software was offered).
Compare apples with apples - if ZZ9000 became standard HW present in every Amiga and it will cost 25$ then you will get software especially in 90's.
It has prefect sense to write computationally intense code with DSP on mind - vendors usually offers set of libraries and highly optimized code - like inverse DCT 8x8 to decode jpeg so everyone, every application can quickly become beneficial from such things.
Nowadays try convince people to write code Zynq ARM where your target is probably few users (how many ZZ9000 was sold - 10? 100?).
Compare apples with apples.
Having DSP in Amiga will be highly beneficial even if performance rarely hit half of peak- still this will be like 100 times faster than 68882.

[hammer]

Originally Posted by Thorham

Science uses whatever precision is appropriate, including 16 bit and less (minifloat).

AT&T's, PowerPC Altivec, AMD's and NVIDIA's FP32 vs FP64 product positioning is consistent. Atari fans are not consistent.

[hammer]

Originally Posted by Promilus

68882 - you compile your 68k with hard floats which goes to 68882 through hardware interface between CPU and FPU. You have nothing to worry, just write your 68k code as usual... Adding DSP to the mix ain't that straightforward and you know it! The flow of the program has to be maintained by 68k and you have to add somewhere DSP code then common memory pool to exchange data. Peak MFLOPS numbers only looks great on paper. Show me actual software which is easy to develop and takes advantage of that? Nothing? We have 600MHz ARM Cortex A7 for few years now. ZZ9000 card. Look how many great apps uses that for acceleration mixed with 68k code. Aaahhhh... exactly.

You can't compare the two eras when the Amiga in 1991-to-1993 had many paid mainstream game developers.

One of those paid mainstream game studios can reverse engineer PC Doom for SNES port and still using an Amiga as a dev workstation for the SNES. SuperFX was co-designed from ex-Amiga 3D game developers in the UK!

The difference is Nintendo supported game developers with actions, not just words. Nintendo is 100 percent focused on its target market. After removing the original Los Gatos Amiga team, Commodore was a master of none.

Non-mainstream Amiga era needed PC's Doom source code in 1997.

It's well known that Psygnosis wanted an upgraded CD32 with minimal cost increase before selling themselves to Sony's PS1.

Intel has a mastery over CPU that is attached to clone PC platform with an industry leading distribution channels. SSE is SIMD extension that is supported megacorps like Intel and MS (e.g. DirectX6.0 geometry pipeline). PC world did both X86 translation in hardware acceleration and SIMD extensions.

--------
Emu68 has quad-core(four threads) soft 68K extensions beyond increasing the single 68K thread. Emu68 exposes quad-core ARM CPUs for quad-core 68K. It needs new software for quad-core 68K usage. New features will need new or modified software.

[hammer]

Originally Posted by sokolovic

My point was you cannot just say that 256 colors is the norm on the SNES for games.
This is a theoretical number that the console most of the Time wasn't able to reach because it was costly hardware wise and even financially because of the ROM size.

If the you take the 256 colors number as the standard on the SNES, then you should take the 262000 number as a standard on the AGA.

SNES can display 256 colors without cheating and it was needed for certain 2.5D games.

AGA's HAM8 has its lossy color edge problem and the same 8 bit-planes load on stock A1200's Chip RAM.

[lmimmfn]

Originally Posted by Tigerskunk

The point is people on the spectrum posting objective numbers in a thread about the subjective feeling if you felt disappointed by something that happened 31 years ago.

Amen brother.

Instead of a thread about a few opinions on what could be improved on A1200 release, its going around in circles with huge datasets of PC info that noone cares for and unhelpful replies of "PC envy" that I don't think any Amiga owner had in '92

[dreadnought]

Sure, this thread is incredibly silly, repetitive, full of overblown drama, and all the possible logical fallacies you can imagine. But it's also the forum's equivalent of a soap opera, a guilty pleasure I undeniably enjoy checking on now and then (and put some more wood on the fire too, of course).

So, leave us alone, haters, and go back to you other perfectly reasonable and sensible threads

[Thorham]

Originally Posted by lmimmfn

"PC envy" that I don't think any Amiga owner had in '92

Indeed.