What spec would the A1200 had launched with if it was up to YOU

[TCD]

Originally Posted by sokolovic

Isn't that the reason of the A400/030 ? A mid range computer between the A1200 and the A4000/040.

I found this:

(West Chester, PA -- March 19, 1993)
Commodore Business Machines,Inc.(U.S.) today announced it is expanding its Advanced Graphics Architecture (AGA) product line with the addition of the new Amiga (R) 4000-030 desktop computer. The Amiga 4000-030 is a lower cost version of the company's Amiga 4000-040 that was introduced in September 1992. The system is expected to be available at Commodore authorized resellers by the end of the month at a Manufacturer's Suggested Retail Price of $2399.00.

That would be $5200 adjusted for inflation in 2024. I wouldn't call that mid-range.

[eXeler0]

Commodore did put the CPU in the 040 on a separate card. So they could have done a similar variant with the 1200? Base model would have tve 020 with unpopulated SIM slot. Then they would offer variants with 28MHz 020 (or maybe the 25MHz 030 they purchased in some volume for the A4000/030 anyway), and a handful of different configs with 1/2/4MB FastRAM, HDDs etc..

[Gorf]

Originally Posted by eXeler0

Commodore did put the CPU in the 040 on a separate card. So they could have done a similar variant with the 1200? Base model would have tve 020 with unpopulated SIM slot. Then they would offer variants with 28MHz 020 (or maybe the 25MHz 030 they purchased in some volume for the A4000/030 anyway), and a handful of different configs with 1/2/4MB FastRAM, HDDs etc..

I think actually the most economical way would have been to have a A1200-like board without CPU in all Amigas - so in the A4000 this would be expanded with better CPU (030, 040), FastRAM, maybe SCSI and ZorroIII slots

in a mid-range AA1000 this would be a faster CPU (030), 1 MB FastRAM, only one ZorroIII on the board to take an option doughter board)

the A1400 with 020 and some FastRAM
the A1200 as it were

All of them sharing the exact same AGA base board, from wedge to tower.

As soon as Akiko or AA+ are released, users can upgrade this base-board.

(living the DSP discussion aside)

[TEG]

Commodore did put the CPU in the 040 on a separate card. So they could have done a similar variant with the 1200? Base model would have tve 020 with unpopulated SIM slot. Then they would offer variants with 28MHz 020 (or maybe the 25MHz 030 they purchased in some volume for the A4000/030 anyway), and a handful of different configs with 1/2/4MB FastRAM, HDDs etc..

I think the problem they had with the A1200 was to output the product ASAP. So engineering and testing the cpu on a board (you have to find the right connector, be sure it cause no interferences and so on...) + the additional cost, that made it wasn't considered.

But for the additional SIM slots I'm on the same opinion. However we add a discussion about it and it require some additional logic not present on the A1200 motherboard and so, some more cost. In addition, it conflict (electronically) with doing trapdoor memory expansion cards. For me, the inexcusable criticism I have, is for Commodore not providing official memory expansion card from day 1.

After all the discussions, the idea was emitted to put the cpu in the trapdoor so it can be changed.

[sokolovic]

Originally Posted by TCD

I found this:

That would be $5200 adjusted for inflation in 2024. I wouldn't call that mid-range.

That's not the price in the announcement in this screenshot from Amiga Format issue 46.
Here we have a "something in the $1000 range". Well not exactly since this is in british pounds.
Note that they are specifically speakîng of bridging the gap between the A1200 AD the A4000/040. And look at the captions of the picture. Mid range computer Indeed !

[eXeler0]

Originally Posted by sokolovic

That's not the price in the announcement in this screenshot from Amiga Format issue 46.
Here we have a "something in the $1000 range". Well not exactly since this is in british pounds.
Note that they are specifically speakîng of bridging the gap between the A1200 AD the A4000/040. And look at the captions of the picture. Mid range computer Indeed !

Just to clear up the numbers a bit:
£999 in 1993 is about £2080 in today's GBP
The exchange rate between GBP and USD at the time of writing is about 1.27, however on March 24th 1993, the exchange rate was 1.48, so £999 was $1478 and £1099 (for the larger HDD version) was $1626
And in today's money, that would mean ~$3200 and ~$3525 today
Inflation for UK & US from 1993 to today is in the same ballpark..

[Bruce Abbott]

Originally Posted by TEG

For me, the inexcusable criticism I have, is for Commodore not providing official memory expansion card from day 1.

Not an excuse, reasons.

1. The A1200 was a low-end system. Commodore had never produced a FastRAM board to go with a low-end system on launch. The A1000 never got one. The A500 only got it with the A590, released 2 years later. The CDTV and A600 never got FastRAM. So not providing a FastRAM board for the A1200 was both consistent and expected.

2. Commodore made trapdoor expansions for the A1000, A500, A500 Plus and A600, but in all cases 3rd party manufacturers were soon making them cheaper. 3rd parties were making FastRAM boards too, so could quickly develop A1200 boards. Commodore wouldn't be able to compete against them for long. The Microbotics MBX1200 was out in March 1993. The Amitek Hawk (also released in 1993) cost only £99 with a 1MB SIMM installed.

3. Due to ongoing financial difficulties, Commodore was struggling just to get the A1200 itself out in quantity. Any additions would have to wait until they had the cash to afford it (many suppliers were insisting on payment up-front, so they couldn't 'buy now, pay later').

4. The A1200 was already more than twice as fast as the A500. If that wasn't good enough for you...

Amiga fans who didn't think a stock a A1200 was good enough were generally looking at accelerator cards because even with FastRAM the A1200 wouldn't satisfy them. GVP (owned by former Commodore engineer Gerard Bucas) had a 40MHz 030 card out in March 1993.

[Gorf]

Originally Posted by Bruce Abbott

4. The A1200 was already more than twice as fast as the A500. If that wasn't good enough for you...

Or twice as fast as the A1000 or Atari ST in 1985.
After 7 years or over 3 Moore cycles one could expect 8x the computing power.

[lmimmfn]

Originally Posted by Gorf

Just a little bit of FastRAM would have been great anyways.
Or some external cache, or at least Alice not booking every second access slot even if there was nothing to do...

the "one size fits all" approach did not work with the differences in expectations and markets anymore ... so some additional mid-range Amiga would have been needed like the proposed AA1000+



A twice as fast Blitter was planed for AA+.
Of course this would have been a configurable feature like the new screen modes. So compatibility can easily be preserved by different operating modes.



offering an AA1000 or A1400 with such a configuration in addition to the low level model would have solved this problem.
Commodore had no problem offering dozens of different PCs with different mainboards and configurations - why was it so hard to do that for the Amiga just for one mid-range model?

Agreed, the base models should have included fast ram even as a cache.
The blitter and Paula should have been 32bit, even if released after the A1200 the motherboard should have had 32bit access to allow 32bit blitter and Paula to be swapped in later.

Commodore should have focused on getting all AGA chips with 32bit access as a priority than wasting time/effort on AAA, I.e. provide a short stepping stone than a huge jump to AAA.

As mentioned several times in this thread but AGA should have been sooner in later 1989.

Wasn't there something in CU Amiga or Amiga Format that Commodore lost the schematics for the OCS chipset and having to reverse Engineer same!!!

[Bruce Abbott]

Originally Posted by Gorf

Or twice as fast as the A1000 or Atari ST in 1985.
After 7 years or over 3 Moore cycles one could expect 8x the computing power.

For $1285? (price of the A1000 on launch)

BTW Moore's law relates transistor density, not computing power.

CPU... number of transistors
68000 68,000
68020 200,000 (68k x 2.9)
68030 273,000 (68k x 4.0)
68040 1,200,000 (68k x 18)
68060 2,500,000 (68k x 37)

So you are saying the A1200 should have had a 68040 in it? Well it did - except that model was called the A4000.

For Moore's law we should consider the total number of transistors in the machine. I'm not sure about the AGA chipset etc., but the A1200 came with 2MB RAM (~17 million transistors) compared to 256kB in the A1000 - which is a little over 8 times more. Looks like the A1200 tracked Moore's law pretty well!

[Bruce Abbott]

Originally Posted by lmimmfn

The blitter and Paula should have been 32bit, even if released after the A1200 the motherboard should have had 32bit access to allow 32bit blitter and Paula to be swapped in later.

Commodore should have focused on getting all AGA chips with 32bit access as a priority than wasting time/effort on AAA, I.e. provide a short stepping stone than a huge jump to AAA.

AAA was almost finished in late 1993, with production expected in 1994. Just one problem - the estimated cost price was $100, which made it much too expensive for an A1200 class machine.

As mentioned several times in this thread but AGA should have been sooner in later 1989.

That wasn't going to happen in this universe.

Wasn't there something in CU Amiga or Amiga Format that Commodore lost the schematics for the OCS chipset and having to reverse Engineer same!!!

I remember reading that, but Brian Bagnall didn't mention it in his books so apparently it wasn't that much of a big deal. According to Bagnall the engineers developed the 'hi-res' Agnus/Denise chipset from the tapeout provided to them by Jay Miner in January 1987. Perhaps that is when some of the original chipset schematics got lost?

The engineers must have been confident that they had enough info because sample 'hi-res' chips were expected in May and production units in July 1987. However due to compatibility issues it still wasn't ready in September. In the mean time IBM had released VGA, and the engineers were concerned that a monochrome hi-res display wasn't going to be enough. They were right, but their efforts to improve the existing design were not sufficient. At that point they should have scrapped it and started a new initiative to deal with the VGA threat. But engineers working on the hi-res chipset wanted to just tweak it to get a few more colors. That is how we ended up with ECS.

This was a big mistake, and it was all on the engineers. Problem is the A500 had just been released and the path it would take was not clear to them. They thought the Amiga needed to compete with workstations and business computers, not games machines. They also didn't seem to appreciate the issues with displaying both 15kHz and 31kHz screens on the same monitor.

For flicker-free hi-res a scan converter could do the job until the chipset could be made fast enough to do everything at VGA scan rates. That's what Hedley Davis proposed, and what the A3000 got (Amber). By the time they got the A3000 out the market had already found a solution with 'flicker fixers' for both the A2000 and A500. ECS hi-res was therefore redundant - a complete waste of time.

The engineers should have forgotten about putting hi-res in ECS, and instead concentrated on getting 256 color chunky pixel lo-res into the chipset. But they didn't because they weren't in touch with what most fans wanted - better games, not boring productivity stuff!

[Promilus]

Of course Moore's law relates to transistor density. It is supposed to DOUBLE each 24 months. Which - if followed - would result (assuming we ignore A1000 completely) 030-based amiga home computers in 91. One year before Falcon. And not to mention serious improvements to chipset. And obviously that 2MB chipram as well. Instead the only home computer presented by Commodore in '91 was A500Plus soon replaced by A600 and then A1200 at the very end of '92. That's exactly what happens when the price is your only concern... you fall behind.

[sokolovic]

Originally Posted by eXeler0

Just to clear up the numbers a bit:
£999 in 1993 is about £2080 in today's GBP
The exchange rate between GBP and USD at the time of writing is about 1.27, however on March 24th 1993, the exchange rate was 1.48, so £999 was $1478 and £1099 (for the larger HDD version) was $1626
And in today's money, that would mean ~$3200 and ~$3525 today
Inflation for UK & US from 1993 to today is in the same ballpark..

In that case the A1200HD as it was officially labelled by C is the closest to a mid range computer since its price probably fitted your expectations.

[Gorf]

Originally Posted by Bruce Abbott

For $1285? (price of the A1000 on launch)

I knew this would come, that's why I included the Atari ST in the comparison.
Commodore wanted to position the A1000 clearly above the C128, which is part of the price tag it got initially.

BTW Moore's law relates transistor density, not computing power.

That is why I carefully chose the term "Moore cycle".
Higher transistor density also means finer structures, witch means faster gates, which again means faster clock rates (at least in that time period)

So you are saying the A1200 should have had a 68040 in it?

No, but a 020@28MHz with cache or FastRAM would have delivered an 8x speedup compared to the original hardware.

[eXeler0]

Originally Posted by lmimmfn

Agreed, the base models should have included fast ram even as a cache.
The blitter and Paula should have been 32bit, even if released after the A1200 the motherboard should have had 32bit access to allow 32bit blitter and Paula to be swapped in later.

Commodore should have focused on getting all AGA chips with 32bit access as a priority than wasting time/effort on AAA, I.e. provide a short stepping stone than a huge jump to AAA.

As mentioned several times in this thread but AGA should have been sooner in 1989

Yes, thats basically what I meant when asking the question about going brute force and get faster to the finished product vs some 5 year long development cycle (AAA)…

Doubling bandwidth, clockspeed, making sure memory subsystem wasnt a bottleneck. Stuff like that doesnt take 3, 4 or 5 years to do. Full 32-bit AGA should have been possible for A3000.