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

[Megalomaniac]

My concern is, what was the maximum price that the market could have supported for an A500 successor? The A500 launched at £499 in 1987, which I guess was about £550-600 in 1992 money, but was £399 for most of its life, and PCs had so massively come down in price and gone up in performance in that time. Plus, on the games side you had the SNES and Megadrive, which both did some things the A500 couldn't, as well as having multi-button controllers and cartridges which loaded instantly and were harder to pirate than disks. Even if you launched something for £600 that matched a £1200 PC in performance, would it have been an easy sell? That the PC had industry-standard software making use of it, whereas an Amiga had nothing using AGA, little using a faster processor or extra memory, barely a third of games being hard-drive installable (though most games that had enough disk accessing or swapping to really benefit from it were installable), and maybe a third of existing games not working on a faster processor?

You could probably have changed the spec to 030 2 Chip + 2 Fast and a 40-60Mb hard drive spec and launched at the same time for £600, as it was off-the-shelf hardware that would only affect the variable costs. However, actually improving the graphics or sound hardware (or an improved Workbench) would have needed earlier planning and large-scale fixed-costs investment that would have taken a certain number of sales to break even.

What needed to happen was probably an earlier acceptance that AAA wasn't achievable in a suitable timeframe, an A1200 launch in early 1992 (probably for £600 with the above spec, perhaps alongside the A600 as a wider range review) as an aspirational model, and then going back to AAA with Christmas 1993 or early 1994 in mind - ideally launching that at £600 with the A1200 down to £400ish by then. We saw from later titles like Super Stardust and Slamtilt that the A1200 could beat the 16-bit consoles for action games, and the above spec could just about do Doom style games, and probably most PC games that were feasible on a 4Mb 386.

[roondar]

Main differences I would have liked - assuming budget wasn't an issue and Commodore had a unified R&D plan rather than the fractured approach it seemed to have in reality (but trying to stay within reasonable possibilities for 1992):

• Default to 3.5 inch HDD inside the case and a proper IDE controller for it (though whether or not a drive was included by default is less important to me)
• Make 100% sure that video output supported standard VGA/SVGA monitors at the time without the weird stuff we see today and that using VGA/SVGA modes doesn't slow the machine down to a crawl
• Add a chunky mode to GFX, even if limited to just 320x200x256 colours
• Make absolutely sure that the CPU can read/write Chip Memory not just in 32 bit chunks, but also at reasonable speeds (DMA permitting), rather than the 1/2 speed access it has now
• Redesign the memory controller so that the Blitter can at least do 32 bit reads/writes, preferably 64 bits (i.e. FMODE 4x for blits)

Optionally: add a HD floppy drive and 8 channel sound with programmable stereo separation/L-R position for each channel.

Obviously, all this would also apply to the A4000.

But... This is all hindsight and instead of continuing to dream, I'd rather use what I have today - some of the AGA stuff in development/released recently is quite amazing

[Paul_s]

Originally Posted by Cris1997XX

Ugh, when will all these what-if scenarios stop...?

It's nice to reminiscence once in a while, I do it quite often with Amiga.

But I wish someone out there would create a Premiership Manager for Amiga, or a Theme Park simulator..

You can take the role of Medhi Ali, Irving Gould for example in management or a leading role in engineering such as Dave Haynie... you could create some mutant Amiga that sinks IBM or purchase Apple shares and have Steve Jobs to replace Medhi... buy out Bill Gates and sink Microsoft? Amiga forever!

[Paul_s]

Originally Posted by Bruce Abbott

Ignoring the silly AI crap...

I wouldn't change the hardware at all. What I would do is move the launch date forward to late 1991. This could happen because I would have delivered the AA chipset when Gould was expecting it, in mid 1991.

I would have told Dave Haynie to forget the DSP chip and just concentrate on getting the A3000+ going with AA. The 50 development machines would be sent to major developers in mid 1991, so AA titles could be available on the A1200's launch or soon after.

I would also work closely with 3rd party manufacturers to encourage them to make FastRAM boards and accelerator cards. By mid 1992 you would see fast 030 and 040 accelerator cards with on-board SCSI or IDE for an external CD-ROM drive. RAM boards might also have this interface.

1991 was Commodore's best year, with the A500 selling more than ever. But this would not last much longer. Having the A1200 out in late 1991 or even early 1992 would keep the ball rolling while I developed the AA+ plus chipset, to be released in late 1992. This would have all the things people thought should have been in AGA - chunky graphics, faster blitter, more sound channels, HD floppy, buffered serial port. The expansion bus would be running at 14MHz like the A1200, but the CPU to ChipRAM interface would also run at 14MHz rather than 7MHz.

By the time this machine was out numerous 3rd party RAM boards and Accelerator cards would already have been produced for the A1200. These would work with the A1200+ too, so 'early adopters' could trade in their their old machine for a cheap upgrade.

Same, but I would additionally dropped the A500+, 600 and spent the spare cash on AA development. Kept the original 500 going for much longer (or until parts exhausted, much like C64). Put more cash into 4000 and given it similar budget to that of 3000.

Depending on launch of 1200 I'd have put the CD32 on the backburner for a while longer to see how AA pans out until fall of 1993/94.

[TCD]

Originally Posted by Paul_s

Same, but I would additionally dropped the A500+, 600 and spent the spare cash on AA development. Kept the original 500 going for much longer (or until parts exhausted, much like C64).

This.

[vulture]

Personally, I would've added a proper socket where the pins for the fpu are and a simm slot for easy fpu and fast RAM expansions.

Other than those, I'd go for what was suggested about the Budgie chip cache and have a year earlier date of release.

I don't think those would bump the price up much if at all.

[spiff]

eXeler0, do I spot a fellow Armchair general?!

Goal
Doom and street fighter 2 playable

Approach
Similar to the Pentium, just double everything up and stack it with cache. Backwards compatible by running everything at half “steps”

• Sound - two paula with rudimentary stereo panning and 8 channels. 4 channel "HI-FI" mode.
• Graphics -2x Blitter+copper + dsp with rotate/3d instructions with the old being left to do backgrounds. Rudimentary blend/subtract
• CPU/Memory - Whatever Motorola of the shelf is reasonable at that time.

Result
I'd basically just design a shit jaguar. Ports with better music and an additional background. Zero SDK / developer support, crushed by the PlayStation.

Or more realistically, killed by committee. Nothing could have successes with the commodore corporate culture. It was mostly dumb luck that the original Amiga could be scaled down at that time.

*edit* Oh and six months later I'd stick a CD-rom on it and call it the Multimedia Ultra64 ...

Even if you launched something for £600 that matched a £1200 PC in performance, would it have been an easy sell?

This makes too much sense, now that I think about it... Even apple failed at that time, with much larger budget and market.

[d4rk3lf]

Is there any significant speed difference between 020 and 030 with same amount of Mhz?

[BSzili]

If this thread doesn't reach a 150 pages, I'll be very disappointed Prepare your hindsight, here comes the mother of "what if" threads!

[Paul_s]

I know some have mentioned this but just on processor front alone..

68020 at launch was $487 in 1984..

https://www.nytimes.com/1984/06/29/b...-new-chip.html

68000 was subsequently reduced to $15 at time (in article) so made perfect sense to keep for A500/2000 etc in 1987.

68030 was still running at $100 per unit (in large quantities) around 1990, reduced to $50 later... I'd imagine 68020 at the time of 68030 introduction dropped similarly like 68000 in cost so makes perfect sense in lower 1200 model (and to keep cost down for consumer)..

[Cris1997XX]

Alright, whatever. Here's what MY dream Amiga 1200 is like...a 25MHz full 68020, 2MB VRAM for Alice/Lisa and 2MB chipRAM accessible by both the sound chip and the CPU without speed penalties. And of course, a souped up Paula with 8-channel stereo sound at 16bit 44,1KHz quality. No need to sacrifice channels or anything

[Gorf]

Originally Posted by Bruce Abbott

My answer is that I would change the targeted release date to late 1991, a year earlier. It should have happened. With the right person in charge it would have.

I have to fully agree here: earlier is the key!

It this 91 computer would be the 1200 or some 1000plus in a flat desktop at first is less important - key would have been to get the new chipset out as soon as possible.

[Photon]

Well what made it the most meh was no upgrade to sound or Blitter. And just 1 chunky screen mode would have been sweet. As for the no fastmem in this machine either, well you could get some pretty easily either via PCMCIA or internal.

So 8 panned sound channels, doublespeed on the Blitter, and 1 lores chunky mode would have felt like a real upgrade.

Instead we got full RGB variants of the bitplane based modes. Looks good, and the A1200 is a great platform to build on.

[Gorf]

Originally Posted by Samurai_Crow

The Budgie bus controller should have been a cache controller for AGA. Adding page-fetch look-ahead buffers to the Copper and Blitter DMAs would have as much as quadrupled their bandwidth, allowing their clock speeds to be quadrupled as well. This would have boosted the Copper-chunky hack to 320×240 using overscan on NTSC with 12 bits per pixel color depth.

Such a mode would be nice indeed - but it would saturate the bus pretty much, wouldn't it? So not much memory access left for the CPU. So it is not really a suitably candidate for early 3D games. A real 8bit chunky mode in addition to that would still be a good thing.
(which could be archived pretty easy be moving the Akiko logic into Alice along with some buffers)

[8bitbubsy]

Bundled fastmem and an improved Paula would be very welcome for an Amiga released in '92. It's almost silly it didn't have any of that.

[Cris1997XX]

Originally Posted by 8bitbubsy

Bundled fastmem and an improved Paula would be very welcome for an Amiga released in '92. It's almost silly it didn't have any of that.

Yeah, a lot of new machines were coming out in the early 90s and they had more advanced PCM synthetizers (SNES, Neo Geo, Atari Jaguar, Sega CD, etc.) Heck, the FM Towns from 1989 had a Ricoh RF5c68

[StevenJGore]

Originally Posted by Photon

As for the no fastmem in this machine either, well you could get some pretty easily either via PCMCIA...

I thought PCMCIA memory expansions halved the speed of the A1200?

[Korban]

2MB fast ram and maybe either a 25hz 68020 (or ec variant) or 28mhz '30.
In an ideal world it should have been able to compete with a mid to higher end 386 with the 4000 competing with 486's.
Granted in some aspects even the stock a1200 we received could compete with a 486, but in other areas it was heavily humbled by even a mid to higher end 386.
Being able to compete with a 386 in all areas, not just its strengths should have been a target/goal by Commodore.

Yes, the higher spec, extra RAM, and/or sockets on a board would have increased price, but it would have let the price remain at 399 pounds and not dropped to 299 pretty quickly.
That extra hundred pounds would have covered the improvements.

Sidebar, but just noticed I remember the prices in uk pounds, despite being Australian.
I blame CU Amiga and Amiga Format

[Gorf]

Originally Posted by d4rk3lf

Is there any significant speed difference between 020 and 030 with same amount of Mhz?

Depending on what you are doing:
If your algorithm fits in the larger instruction- and data-caches of the 030 you can have a large improvement.

Paolo Cattani explained the cashes are the reason why his Virtual GP needs an 030.

Originally Posted by eXeler0

CPU: Motorola 68EC030 at 25 MHz
Motivation: A balance between cost and performance. The 68EC030 is a cost-reduced version of the 68030, lacking a MMU (Memory Management Unit), which isn't crucial for a consumer-level machine.

I disagree here with The Oracle:

A full 030 with MMU would have been much better.
Even if AmigaOS can not utilize most of the MMU functionality, tools like Mungwall or Enforcer improve stability quite a bit, making the Amiga look less like a toy.
Commodore could have also released a stable virtual memory library this way, allowing software that needs a ton of RAM to swap.

BSDs and Minix where already available and Linux was on the rise:
being able to install UNIX clones on your Amiga would have attracted more developers, like Matthew Dillon (Amiga "Dice", later DragonFly BSD)

Having an affordable alternative to 386-PCs for BDS/Linux development could have changed the landscape of early OSS.

(I myself installed 68k-Linux alongside AmigaOS on my A3000 as soon as I could get my hands on it...)

[Gorf]

Originally Posted by StevenJGore

I thought PCMCIA memory expansions halved the speed of the A1200?

Since it is 16bit it surely did ...
PCMCIA memory works fine for the A600 but is a terrible idea for the A1200.