What would U have done to make the Amiga succeed better?

[Keatah]

Higher resolution Amiga bitmaps also take a lot of chip ram bandwidth "at rest", which can slow down a base model, especially when the color depth is increased.

 

Text modes have a lot of advantages for that era of program interfaces. The Amiga needed more bandwidth and a faster blitter to really make use of its own feature set.

 

And more bandwidth and a better blitter would mean significant redesign.

 

I've often mentioned in other threads that custom chips limit system upgradability and adaptability. While they usually give an initial "WOW" when the architecture is revealed, as time goes on they tend to hold a system's evolution back. This is especially true here when so much of "Amiga" IS the custom chips. Increasing performance in one area likely necessitates a redesign of the entire chipset. Ohh you might get away with a small tweak through a revision, but nothing to take you to the next level.

 

Custom chipsets also complicate the design of accelerators. They have to slow down or buffer to talk to the rest of the machine. That's what contemporary accelerators of the day did. The other option is to replace all the logic of the mainboard into a smaller modern-day daughtercard like the Vampire. The Amiga computer is basically host to a whole new system. Your original hardware is baggage. That's why they're working on a completely standalone Vampire. It's why you have to pull the Kickstart ROM out. It's why I don't consider the Vampire an accelerator. It's a replacement designed to use the IO features your existing mainboard.

 

The MAC had a different philosophy. It was much like the Apple II and PC. There were minimum or no custom chips in those designs. The few that were were not integral to processing information and pushing data around. Consider the Integrated WOZ Machine disk drive controller. Oh yes it was custom alright, but it sat outside the main program loop so to speak. And the GUI? It was all software, all in the firmware ROMs. All pixels and on-screen information were drawn by the 68000. This made upgrades to both hardware and software easier. New more efficient algorithm? Update the ROM. Faster processor? Update it, and leave the firmware mostly untouched. And no redesigning of a centralized blitter need be done.

 

Similar with the PC. In looking at Monochrome/Hercules through CGA/EGA/VGA, these chips were out of the loop. The processor was happy to work with itself and the program in memory. Graphics were optional.

 

When Windows came on the scene, or slightly earlier, PC started to get graphics primitives, but they were loaded with each boot/start, right into RAM. Fine. Point being is that with a standard EGA/VGA framebuffer chip (no acceleration whatsoever), the processor would draw Windows' buttons, sliders, boxes, and all that stuff. Just like the MAC. May not have been fast, but it worked. Windows was just another program running.

 

Sometime in the 1991-1995 range, graphics chips became more sophisticated. They gained knowledge of the Windows' primitives and GUI routines. The main processor, 386/486, whatever, would now call a driver, and the driver would send specialized instructions to the graphics chip to draw a line, or a box, and perhaps fill it in with color X. The main CPU would be only handling a few data bits, the instructions to the graphics card, and the graphics card did the heavy lifting of managing the framebuffer's contents and drawing things to it. Graphics accelerators were known as "Windows Accelerators". Some relied on routines in their own on-board ROM, some had Windows operations hard-coded in the chip itself, others relied on the main CPU to assemble a set of instructions and send it over. And most were a combination of everything.

 

It was cool, because these could work at whatever speed they needed to. They were modular and out of the loop. They had their own RAMDAC. And they even had their own dedicated memory. And in the 1992-1993 time frame they were running at ~100MHz for a SOHO consumer card in the $150 price range.

 

This concept eventually expanded into the 3D graphics chips and the APIs we know today. The 3D chips of today are their own complete computer systems on a card. Many are used for high-end compute projects.

 

While a modern graphics chip, any graphics chip, is custom to the hilt, they all have a standard bus interface. Back then it was ISA - PCI - VLbus - AGP. Today it's PCIe or internal to the processor. And they could be changed out by the user for a faster or more feature-rich model without a complete redesign. My old classic Pentium II rig started out with a 1MB Cirrus Logic GD5422 ISA board borrowed from my 486. I soon updated it to a Riva-128 + 2x Voodoo2, TNT2-Ultra, Geforce-2 GTS Ultra, Geforce-3 Ultra. And finally stopped at a Geforce-4 Ti 4600 Ultra. Each card improving performance significantly. Same motherboard, same hard disks, modem, memory, chipset. Though I did upgrade the processor to a Pentium III, and made other smaller changes.

 

This same concept I just described for graphics applies to many other aspects of the PC, sound, modem/telecom, disk interfacing, and more. It was all out on the bus. Any manufacturer could take a ride. And the benefit to the user was they could update their rig piecemeal as budget allowed. In this post-PC era I sometimes miss that sort of tinkering.

 

Tightly integrated custom chips in the Amiga prevented that level of customization, performance, and versatility.

 

---

 

I'm beginning to believe C= execs didn't give a rat's ass about upgradability. I also believe they mis-judged the rate at which computer tech would be changing in the few short years ahead.

 

In light of that I wonder if making an Amiga card for the PC might have prolonged or even advanced the life of the platform? I'm thinking something like the Jaguar development board, or the 3D0 Blaster. Would it have eventually morphed into a standard platform for all future game development?

 

I ask that because before 3D chips came on the market everyone was writing their own software engine for games. Could a properly marketed multi-media card have become a standard instead of the separate Sound/Graphics cards of today? And around 'n' round we go.

[pacman000]

"A500 like product at launch (could sell in Kmart/Sears, etc.)"

 

Atari wasn't able to do that with their cheaper computers; that's why they made the XEGS.

[rpiguy9907]

Commodore management never had a future vision nor thought much about things like upgradability. The engineers had to figure out what to make next they were given no direction.

 

And since the Amiga started life as a game console, very little thought was given to the next generation. Ranger was a hack to add VRAM and more colors to the Amiga, still custom and not scalable.

 

Later on the engineers realized they needed retargetable graphics but it was too late. They made the libraries for it, and probably would have for 3D as well, but it was over already by then...

[Keatah]

The more history I watch about the Amiga the more I realize just how much custom chipsets held the system back. CC essentially made it impossible to upgrade anything without major redesign - which cost money. Nothing was scalable piecemeal.

 

When I was a kid I always thought CC were the end all be all. After all, the VCS, C64, 400/800 had them and they kicked the Apple II's ass in gaming graphics. And I always thought it was dumb that business software never made use of them. I was greener than the front lawn in a thunderstorm!

 

So what gave the Amiga an initial wow factor was the same thing that limited its growth. At one time the Amiga was 4-5 years behind those "boat anchor" PCs! IMHO, anything after the 2000 was little more than putzing around.

[Crazyace]

If the Amiga had shipped with mixed chip and fast ram it might have helped things a bit - the original A1000 could have been 256K fast plus 256K chip - and that would allow the 640x240 16 colour modes to be used without slowing down the cpu at all.

It would have been a lot easier to make new models with faster cpu's - or even move the video circuitry onto a card in the A2000's to more directly compete with the IBM PC

[Zonie]

Open architecture, Like a PC

[Arnuphis]

Focus on A1000/2000/3000/4000 'business box design' - sell it as a business machine only.

VGA output on all models.

Partner with Microsoft on an Office Suite for the platform or have full x86 compatible board option.

 

Failing that:

 

1. Collect all Underpants

2. ?

3. Profit

[Crazyace]

This website is quite interesting http://www.amigahistory.plus.com/sales.html

 

It shows the breakdown as almost 10 to 1 in favour of the cheap A500 model, which just shows how the market had moved on

 

The ST had the foresight to implement the b/w mode - that was a failing of the amiga tying it completely to NTSC - given the price of the A1000 support of a workstation monitor ( No VGA in 1985 ) output with non interlaced 640x480 @ 60Hz would have been amazing - even if it was limited to 4 colours it would still be better than the ST.

 

Maybe the Amiga would have been better as a bare bones machine - a super C64 rather than a new multitasking windowed machine. As well as the games, lot's of the definiing titles such as DPAINT weren't really windowed apps anyway

[rpiguy9907]

As I mentioned in the US they were doomed as s business machine because of their bad relationship with distributors.

 

Also Bill Gates hates Commodore so the chance of MS software on the Amiga was slim. After the multiplan fiasco on the 64 it was over for MS productivity on Commodore outside of BASIC.

 

 

Focus on A1000/2000/3000/4000 'business box design' - sell it as a business machine only.

VGA output on all models.

Partner with Microsoft on an Office Suite for the platform or have full x86 compatible board option.

 

Failing that:

 

1. Collect all Underpants

2. ?

3. Profit

[White Flame]

Maybe the Amiga would have been better as a bare bones machine - a super C64 rather than a new multitasking windowed machine. As well as the games, lot's of the definiing titles such as DPAINT weren't really windowed apps anyway

 

Yeah, and they should have called it the Commodore 65.

[zzip]

The more history I watch about the Amiga the more I realize just how much custom chipsets held the system back. CC essentially made it impossible to upgrade anything without major redesign - which cost money. Nothing was scalable piecemeal.

 

When I was a kid I always thought CC were the end all be all. After all, the VCS, C64, 400/800 had them and they kicked the Apple II's ass in gaming graphics. And I always thought it was dumb that business software never made use of them. I was greener than the front lawn in a thunderstorm!

 

So what gave the Amiga an initial wow factor was the same thing that limited its growth. At one time the Amiga was 4-5 years behind those "boat anchor" PCs! IMHO, anything after the 2000 was little more than putzing around.

Not just not upgradable, but the custom solutions never get as widely supported as you'd like because many developers either don't know how to get the most of them, or believe it isn't worth the effort for a relatively small audience.

 

Kind of like the PC fan boys today who spend serious money on top of the line GPUs only to not have developers take full advantage of these cards, because they represent like 5% of the PC games market.

[Crazyace]

 

Yeah, and they should have called it the Commodore 65.

 

 

No, it would have been the Commodore 68K

[roland p]

No, it would have been the Commodore 68K

Wow a whole product-line! 68K20 ... 68K60

[shoestring]

As I mentioned in the US they were doomed as s business machine because of their bad relationship with distributors.

 

Also Bill Gates hates Commodore so the chance of MS software on the Amiga was slim. After the multiplan fiasco on the 64 it was over for MS productivity on Commodore outside of BASIC.

 

 

 

Maybe there was still a small chance. Microsoft wrote Amiga Basic which shipped with the original workbench disks.

[Bill Loguidice]

 

Maybe there was still a small chance. Microsoft wrote Amiga Basic which shipped with the original workbench disks.

 

Having Microsoft productivity apps was still not make or break back then. The productivity app gap for the ST and Amiga was still well beyond just lacking Microsoft support. Big name, contemporary word processing, spreadsheet, and database apps either saw one or two releases with limited updates/support, or no release at all. It was probably a chicken/egg thing. The ST and Amiga would have had to have both sold more and made more in-roads into business to get more attention from these key developers/publishers, and it was going to be difficult to make more in-roads without the software. And it's not like Atari or Commodore really had the money to throw at these publishers to help in bridging the big name app gap, or even really could properly focus on the business markets. They were always up against Macintosh systems and, in particular, IBM PCs and Compatibles, that had relatively big head-starts of 1 - 4 years, no real stigmas to the names, and no particular incentive for those established user bases to move over.

[JamesD]

Bill is right, no software, no sales. That's not just business, it's home as well.

The Amiga didn't even have a decent word processor for more than a year. TextCraft, and GraphiCraft, Transformer, etc... (we referred to the first two as TextCrap and GraphiCrap) that were used to show off the system were all unfinished.
Some version of TextCraft did sell, but it still had issues. They should have given away those titles with the machine as a starter pack, it's not like people could do much with the machine without them when it was introduced.

If you look at my post (#6), the first two items were related to running business software, and the very popular home/school program Printshop.
I sold Amigas for several years, and here were the top questions from potential customers:
Is it IBM compatible?
Does it run Lotus?
Does it run WordPerfect?
Does it have Printshop?
Does it have dbase?
I don't think most customers even knew what some of those were, it was just a checklist they were following from a friend or magazine.

My next item was BASIC. Attract the hobby programmers, and make it easy for them to create impressive looking software.
Hobbyists could fill in some of the blanks in the software market but AmigaBASIC had serious issues.
The most common comment/complaint about programming the Amiga, was that you had to program it in C, which has quite a learning curve.
The states and capitals game 'Great States II' was named as one of the ten best programs of the year in an issue of Amiga World, and it was written in compiled AmgiaBASIC using a system library to load IFF pictures, load and play IFF sounds, and to do color cycling using an interrupt. It even faded the screen in and out to transition between pictures if you wanted to.
The other things AmigaBASIC needed was a tracker library for playing music, and an AEGIS Animator player.
With just those few items, even a BASIC program on the Amiga would look years ahead of anything on the PC.
Those would have also been valuable for people programming in other languages.
The programmer can create fancy content in easy to use software, and can use it without the steep learning curve.

Then I addressed flaws in the upgrade path of the OS and hardware.
If someone had convinced Carl Sassenrath that home computers would have multiple CPUs, protected memory, and virtual memory within a decade, he probably would have made it possible to extend the OS to do that without breaking anything from the start.

Isolating the chipset from the GUI and APIs might be a bit tougher, but I think an upgrade path could have been established.
Lets face it, Windows had it's share of stops and starts in that department. Backwards compatibility issues, 16 vs 32 vs 64 bit, WinG, a dozen versions of DirectX, etc...

And my CDTV comment... well, that was about inroads into businesses and schools.
Instructional Technology is also a niche market that could have been exploited and the system would have excelled at it.
The CDTV was too expensive to be just a consumer product, but it was inexpensive compared to IBM compatibles configured for that job.

As CD drives dropped in price, support for CD drives could have been added to their other machines.

[Nebulon]

Memory protection from day one (which they almost had and very cost-effectively -- see the book Commodore: The Amiga Years).

 

Bring out an update like AGA earlier (like in 1990).

 

And with that update, fix the AGA bandwidth problem and make sure that its 256-color desktop display mode is fast as hell (along with a much larger maximum capacity for Chip RAM).

 

Move the CPU to the PowerPC series as soon as the PPC became cost-effective and (if the significance of 3D had been more apparent at the time) start boosting the system's 3D capabilities.

 

Skip ECS and the Amiga 3000 and go straight to AGA (or something a bit better than AGA) and get AmigaDOS 2.0 out to users much earlier. AmigaDOS/Workbench 2.0 marked the point at which the Amiga's operating system became a real OS. It's just a shame that it wasn't available earlier.

 

Bring the audio up to 16-bit with at least 8 channels and include a DSP (like in the Atari Falcon).

 

Chunky-pixel support earlier would have helped. Then-again, it was hard to predict what a crazy fad chunky-pixel gaming would turn out to be in the 1990s).

 

Save additional money by skipping the CDTV, A600, and CD32.

 

None of these are big design changes. Nothing radical. Some has to do with timing.

 

In my mind, the Amiga is what it is. I don't expect it to be a huge business machine. Design and entertainment on the other hand...

Edited March 22, 2018 by Nebulon

[JamesD]

The problem with pushing the Amiga as just an entertainment machine, is that the PC became a better entertainment machine very quickly.
Most of your suggestions would make a difference, but AGA would probably need to be in about 1988 or 1989 to counter PC development.
That's about when the first 2D accellerated graphics cards showed up and the Amiga started to fall behind in speed.
By 1992, 3D APIs were hitting the market and the Amiga needed something to counter that. By 1995 hardware 3D acceleration was hitting the market.

The PowerPC didn't come out until 1992. That was a bit late to make a difference in CPU performance by itself.
But with DSPs and multiple CPUs before then, I think they would have been okay.

Ultimately, some sort of new graphics hardware would need to be released by 1992 to speed up software 3D rendering, and full 3D acceleration would need to follow.
I just see too much competition from the PC market for the Amiga to keep up unless they can expand their market share, and that means other markets.

[Keatah]

I never did an honest-to-gods in-depth analysis of Amiga graphics specifications and timings, OCS/ECS/AGA or any other variant, but, it seemed to me that a lowly ISA-based Cirrus Logic 5422 "windows accelerator" from the 1992-1993 era outperformed anything Amiga.

It seemed that 100,000 software packages, games, programs, and everything else ran on one graphics chip and one combination of OSes. That'd be MS-DOS 5, and Windows 3.1. It was a combination that I could migrate into and out of over the course of several years. Unlike the Amiga where the transitions would be rather abrupt. All new hardware and software required for the next OS, like AmigaOS 3.1. All new chipsets, motherboard, computer, required, too. One might argue the same was true of Wintel machines, but I was able to run (barely) Windows 95 on the 486 for a while.

When I upped the 486 to a PII-266, I was able to temporarily take along the graphics card¹, sound card², modem³, hard disk⁴, and a few other tidbits I forget exactly what. But the transition was eased by doing it piecemeal. Later when I could afford to I got proper versions of those old peripherals and returned them to the 486 from which they came. Now I had 2 computers!

And I didn't have to worry about Slow/Fast/Chip RAM, nor OCS/ECS/AGA, nor mixing and matching Workbenches and KickStarts. My endeavors and adventures into Amiga didn't go much beyond KickStart 1.3 and WorkBench 1.3.3 / 34.34 anyway, so..

No doubt there were lots of innovations showcased on the Amiga, and in that way I'm nostalgic for it. Like learning mouse-based graphics through PhotonPaint and DeluxePaint III. Ohh we had similar on the Apple II and Macintosh, but not with thousands of colors.

 

For me it still comes down to ease of migration over time and not any one whiz-bang feature. Another thing was the apparent build quality of the PC machines and peripherals. Machines were metal, peripheral cards had metal brackets. The documentation that came with the stuff was almost as good as the early Apple II manuals. Information was readily available. And it was a simple matter of crossing the street to get from the grocery store to the computer store. PC stuff was EVERYWHERE in the 1990's.

 

 

---

---

 

 

[1]

These "basic" Cirrus Logic 5422 chips were just that, basic. Barely above the VGA and VESA 2.0 common denominator. But most all software was written to work on the standard. They weren't blindingly fast like an ATI or Matrox chip. Or at least that's what the advertising of the day led me to believe. I really hate how advertising tries to make you feel bad if you don't pick "their" choice. As if to say my choice isn't good enough and I could do better. Well fuck'em all!! So.. yeh.. the $179 graphics board lasted me through several years. That's important to a student. The replacement for my 5422 was a Riva-128 which I got in the 1996'ish timeframe. It had 2D/3D capability and a whopping 4MB of onboard RAM + vivo. It was good, because all the software that worked on the 5422 also worked on the 128.

 

 

[2]

I originally started with a ProAudio Spectrum from MediaVision. I had it for a couple weeks or a month. But I returned it because I heard of some incompatibility or something not quite right when it came to SoundBlaster interoperability. And I wanted THE standard anyways.

 

So I eventually got a SB16. And I soon discovered it was the right choice. More games defaulted to its default settings and I didn't have to do anything more with DMA/IRQ settings. Then I picked up the ASP expansion chip for the card, and farted around with speech recognition and other DSP effects. It wasn't much my cup of tea. So I didn't go too far down that rabbit hole. I then got to wondering about the stake connector on the card, what goes there. Ahh-hah! A MIDI card? What's that? Ohh sampled instruments. So I got a daughtercard for it and snapped it in. Awesome! Doom II was now scored by the London Symphony Orchestra. Never mind anything Roland or Gravis. Out of my price league. CL SB16 was good enough for me! And the SB16 also gave me a CD-ROM interface. Cool. I was now able get RedShift, an astronomy program. Among several other "Multi-Media" titles.

 

All in all the SB16 + ASP + WaveBlaster + 1x CD-ROM cost me a little over $525 together. Outrageous those cheaper computer companies would say. Maybe they were right. It was rather expensive. But you know, I was able to migrate that across 2 computers and not worry about software compatibility. Everything used CL SB. So it really cost me $260 per computer.

 

At some later date I upgraded the PII to a PIII. And I decided to return the SB16 + addons back to the 486. I replaced it with an SB64 AWE Gold. Essentially this is the SB16 + ASP chip + WaveBlaster on one card. It was all "SoundBlaster" so it's all good and backward/forward compatibility remained very high. I had no problems.

 

All this card shuffling and borrowing helped keep me in the PC ecosphere. And big business did the same thing. Sounds like a lot of money was spent? Not really, these cards with disks and dox and box would sell well on ebay. But I'm not selling them. They see occasional use from time to time.

 

 

[3]

I had a SupraModem 2400 for the Amiga, it was my first 2400 baud modem. Lightyears faster than The Apple-Cat II and MicroModem II, 1200 and 300 baud respectively. It was sophisticated. It was external. It used a real professional-grade RS-232 serial cable to interface with the computer. I thought I was in the big-leagues with the Apple II. But this.. this.. this was beyond anything I'd ever imagined. I could transfer a complete Apple II diskette in like 5 or 6 minutes, especially with Z-Modem streaming protocol and compression.

 

It was another time I thought the Amiga was sophisticated. Granted I had some trouble transferring files to and from the Apple II. And that was part and parcel to me not understanding the nuances of a null-modem cable or not using the right terminal program (and settings) on the Amiga. It was at a time when I was looking to transition out of the Apple and get into something more capable. Either way, 2400 baud was the shit! And it was external so I could see the status leds do their thing.

 

But, anyways, I stopped "modeming" on the Apple II right after I got the Amiga 500 more or less. Porn pictures were better, BBSes were bigger, modems were faster, and all that.

 

It was weird, though, for some odd still unknowable reason I never transfered the Supra 2400 to the PC when I got it. It would have been perfectly serviceable. Instead I painstakingly saved up for a 14.4 internal modem from Practical Peripherals. For sure I was now playing with the big boys! 14 thousand baud! Nothing could be faster! And it was a good fit for the 486 and ProComm+. I also migrated it to the P-II later. And like the other bits of hardware, I returned it to the 486 when I got a 56K modem and America Online. All the while, again, keeping the same terminal software and backward/forward compatibility.

 

[4]

Hard disks, what need be said? When I upgraded from 486 to PII I bought along the hard drive for a little while. But this rapidly became unworkable. OS'es were beginning to bloat, my star chart databases (by nature) were getting larger and larger, and I was accumulating hundreds of levels for Doom and Duke and Descent. 200MB in 1998 was rather small. But it worked long enough to get me by.

 

Storage devices are seemingly one class of hardware that remains remarkably compatible in the forward direction. I could pull out the hard disk from the 486 rig and access the data right on this year's laptop without any issues. In fact, using a $10.00 IDE-USB cable I might not even have to unscrew it from its mounts. Conversely in the opposite direction, I can use a modern microSD card on a classic computer with the appropriate peripheral. It's nothing new and nothing impressive. I had seen signs the industry was operating this way since the pre-486 days. Standards were many. Standards were slow to change.

 

We've been stuck on SATA, now, for over a decade. NVME, m2, PCI-e SSD, and that whole gang.. It's just getting started.

[Keatah]

The PowerPC didn't come out until 1992. That was a bit late to make a difference in CPU performance by itself.

But with DSPs and multiple CPUs before then, I think they would have been okay.

 

I will have to disagree there. Unless you mean something else.

 

To make the Amiga more successful, it would need simplification and modularity.

 

Having more main processors and DSP assistants would have been a liability. That asinine Jaguar console had that problem. Many devs complained (or just didn't use to full effect) the the architecture due to complexity. Multiple processors, multiple busses.. What was a developer to do? Spend time on building the game? Or spend time navigating and managing that maze of hardware? The Amiga was the same way.

 

Doom. It was good on a 386/40 - the first uber-popular PC gaming spec. What else was required? A standard run-of-the-mill framebuffer videocard. And a standard run-of-the-mill SoundBlaster card. Neither of these peripherals were permanently locked into the architecture. Nor were they particularly programmable. They didn't have separate programs to run. They didn't need "management". The main program spit the data to them and that was that. Glorified D-to-A converters in a sense.

 

Some of the Amiga's custom chips, while programmable and capable of running their own individual programs, were not complex enough to be significantly useful co-processors. Not in the way an 8087 is. That's a real co-processor. To call Amiga's custom chips co-processors, that's marketing speak. And marketing will tell you anything to make a sale.

 

Amiga "co-processors" weren't complex enough to have their own standardized APIs either. Sure there were Intuition routines, but that was in a constant state of flux. New KickStarts and new Workbenches had to be matched just right, too.

 

What advertising and technical talks touted as advantages were merely extra things requiring babysitting on a set timetable. To change that timetable meant a complete redesign of the machine.

 

JAG has:

https://en.wikipedia.org/wiki/Atari_Jaguar#Processors

Not only does the programmer have to write his/her game. But also a management program to oversee it all. And god knows what else..?

 

Amiga has:

https://en.wikipedia.org/wiki/Amiga#Hardware

 

The PC, similar to the Apple II and MAC had little more than RAM and a Main Processor. This allowed programmers to fully concentrate on the task at hand, whether making a game or an aeronautical simulation. People complained at the time that the PC was crap and slow and hard to work with. Bullshit. You may have had to do more on your own, but the results spoke for themselves. The PC was on solid ground long enough that APIs could be developed. APIs that could be upgraded and re-used in part or full across several generations of machines and software. And everything was modular, yet adhered to a minimal spec.

 

Heh "Minimal Spec." advertisers loved to twist that one around didn't they?

[zzip]

I never did an honest-to-gods in-depth analysis of Amiga graphics specifications and timings, OCS/ECS/AGA or any other variant, but, it seemed to me that a lowly ISA-based Cirrus Logic 5422 "windows accelerator" from the 1992-1993 era outperformed anything Amiga.

 

 

ISA was slow and not great for graphics. Systems with ISA-based graphics tended to have lots of screen tearing. It wasn't until the local bus connections (VLB and PCI) became available in the early 90s did the PC really start to excel in graphics performance

[JamesD]

I can't believe someone just used the Jaguar as an example of anything. The Jag is nothing like an Amiga.
The MAC has never been a huge gaming machine, so it's not a very good example either.

Look, even applications that use multi-core CPUs require management of what each core is doing, yet that is where we are today.
But multiple CPUs aren't any worse than multiple cores, and the Amiga would have been years ahead of the PC getting multiple CPUs or cores.
The DSP is certainly a special case, but you don't have to use it if you don't need it.
Modern graphics cards have GPUs that do what DSPs do, but they specifically target 3D graphics.
I'm not suggesting an 80s DSP could render 3D video in real time, but I worked on an embedded system that had a 68020 and a DSP back in the 80s.
It would stomp on a PC for mathematical calculations, and the company producing that box didn't replace it with a PC for a decade.
Having a DSP built in would have jumped the Amiga way ahead of the PC for rendering animations, processing audio, and for scientific applications.

By the time the PCI buss came out in 1992, the faster x86 CPUs could drive the graphics display faster than an Amiga and the battle was mostly over.
3D cards were the final nail in the coffin in 1995. PC's were multi-tasking, were faster, had better games...
Any advances that would keep the Amiga competitive longer had to start in the 80s.
Multiple CPUs and DSPs would have been a good start. Quickly transitioning to the PowerPC in 92 would have been a good move given Motorola's failings with 680x0 upgrades.
Once the OS could handle multiple CPUs, that would carry over the to PowerPC.

The AmigaOS was tied to closely to the Amiga's custom chipset. I've stated that from the beginning.
APIs that could hide the combination of software and hardware use are not that difficult.
With the Amiga chipset, you'd just have more in software than with a more advanced chipset.
The problem is, you have to design the software with that in mind from the start or you break compatibility with existing applications.

Edited March 23, 2018 by JamesD

[Keatah]

I'm content to view the Jaguar an overcomplicated mess that developers believe to be a waste of time. And leave it at that.

 

In the first 10-12 or so years the PC was out, I don't know how much of the stuff was pre-planned. On the surface NONE of it seemed planned. The roadmap was being made realtime it seemed. Things moved fast, and nothing seemed guided by one central authority. New advancements were happening every month in some aspect. Graphics, sound, storage, memory, bus, cpu..

 

I recall one of my Gateway literature packs had 386 - 486 - P-Pro - Pentium machines all listed in the same catalog. ISA - VL-Bus - PCI - EISA were all in there too. Quite a range.

 

The first 3D graphics chip from Nvidia used curved surfaces and quadratics. It was practically an experiment to see if they could establish a standard. Didn't work. And it had a proprietary API. To their credit they quickly backed out and went the traditional polygon route we still have today.

 

3DFx experimented with 3D passthrough connectivity, for cost reasons, and the kludge worked. Voodoo also used Glide - another proprietary API.

 

And same thing with Rendition's Vérité 1000. But it was 2D/3D in one chip - and would be the way forward, except for the proprietary API again.

 

All these tricks and hardware variations were being thrown at the wall to see what would stick. None really did. There was OpenGL - which Nvidia and 3DFx were reluctant to support. But Nvidia eventually did.

 

All these things were being tried out. None of this was happening on any other platform. Consumers and multiple companies were spending the money to wade through it all. Too big for one company to handle. And it is an essential part of the process.

[Nebulon]

 

Having more main processors and DSP assistants would have been a liability. That asinine Jaguar console had that problem. Many devs complained (or just didn't use to full effect) the the architecture due to complexity. Multiple processors, multiple busses.. What was a developer to do? Spend time on building the game? Or spend time navigating and managing that maze of hardware? The Amiga was the same way.

 

Amiga "co-processors" weren't complex enough to have their own standardized APIs either. Sure there were Intuition routines, but that was in a constant state of flux. New KickStarts and new Workbenches had to be matched just right, too.

 

 

According to people who have programmed the Amiga (and according to Joe Decuir himself), the Amiga is actually not hard to program. It isn't a SEGA Saturn situation by any means.

As for the DSP.... The intent is to use it for audio. And that would essentially be the same scenario as the Soundblaster Live card and its DSP. Nothing taxing on the programmer in either case. It's pretty straight-forward for people who write audio apps.

On the topic of co-processors, the Agnus really is a co-processor. It doesn't just have to process graphics data; it can process any data that's fed to it and it can do it independently of the CPU.

You make a good point about the Cirrus Logic card as far as static graphics and chunky-pixel scaling goes. It's true, Amiga users had been asking Commodore to please give them non-interlaced 1024x768 (or at least 800x600) with 256 colors -- since at least 1990. But of course that fell on deaf ears. And that was the PC's edge (along with decent polygon rendering speeds starting around 1991). However, as Zzip mentioned, scrolling on a PC in 1992 was abysmal.

I used to work at a store configuring, selling, and supporting PCs throughout the late 1980s and into the early 2000s. I'd spend my days supporting Windows and then I'd go home and use my Amiga 500 (and later 4000). It was a good opportunity to compare and contrast the two platforms).

Serial transfer speeds on a 68030-equipped Amiga edged out the 16550A UART on the PC (not to mention the 8250). 115,000 bps was not a problem. In fact the Amiga's serial transfer rating has a theoretical max speed of close to a million bits per second when error-correction is disabled).

And while the Amiga bus may sound complex, it's actually quite elegant. That's part of the machine's ability to smoothly multitask and even provides a system of plug-and-play that works (whereas it took M$ almost two years to iron the bugs out of Windows 95's plug and play implementation).

People talk about the Soundblaster card a lot. Many of them seem to forget that the first one was 8-bit monophonic and that the SB16 had terrible DACs.

Having said that, you hit the "year everything changed" dead on. 1992/1993 was the beginning of the death knell for the Amiga. That was when the PC started taking the lead in certain areas. However, it wasn't until about Windows 95 came out that the PC had a real GUI that could somewhat compete with AmigaDOS 3. Even at that, an Amiga 4000/040 could multi-task the doors off of a PC (try putting a CD in the drive of a PC right up into the PIII era and observe the sudden inability to so much as move the mouse pointer).

Also as Zzip mentioned, PCI was the big advancement on the PC.

It was when the PCI-based 3D accelerator cards came out (along with the Sound Blaster Live card) that I finally had to walk away from the Amiga. Those advancements -- coupled with Windows NT/2000 -- were just too much power to ignore.

And setting the time machine back to the 1980s again, it's hard to compete against cheap clones that offer enough business performance on easy-to-read displays. Even the Atari ST tried to take them on with their high-resolution productivity mode. By rights they should have been able to make a significant dent in the PC market. Again though, it's the clones -- an important advantage in leverage that the Mac, ST, and Amiga did not have. The platform with the most machines in place wins the devs and sets the standards.

[zzip]

(try putting a CD in the drive of a PC right up into the PIII era and observe the sudden inability to so much as move the mouse pointer).

To be fair, that's Windows. Other PC OSes didn't have that issue

 

 

And setting the time machine back to the 1980s again, it's hard to compete against cheap clones that offer enough business performance on easy-to-read displays. Even the Atari ST tried to take them on with their high-resolution productivity mode. By rights they should have been able to make a significant dent in the PC market. Again though, it's the clones -- an important advantage in leverage that the Mac, ST, and Amiga did not have. The platform with the most machines in place wins the devs and sets the standards.

And this is why I think IBM's decision to make the PC an open platform is what ultimately killed the Amiga (and ST). My understanding is the only reason IBM did this was they thought personal computers were a passing fad, so they really didn't care about the PC market much, but still wanted a presence. They came around to view this as a mistake and tried to "correct" it by making the PS/2 more proprietary, but by then it was too late.

 

So in a world where the PC clone market doesn't exist, Amiga and Atari have a chance to compete on price and own the low-end. Instead the competition and economies-of-scale in clones meant it was only a matter of time until they exceeded the proprietary Amiga and ST on price and performance.