Things the Amiga didn't get right from Day 1

[Foebane]

Originally Posted by Bruce Abbott

Woz's color 'hack' was actually quite brilliant. 6 colors without any extra video memory, no complex logic or analog circuits required, faster graphics rendering and lower memory requirements than a conventional bitmapped display, no loss of resolution to get color, and compatible with both color and monochrome composite monitors and TVs (with a nice textured appearance in monochrome that also printed well on a dot-matrix printer). It looks crude today, but was a good match to the hardware and expectations of the time.

Except that sort of thing didn't work so well in PAL!

Or didn't the Americans give a stuff about overseas territories at the time? Would be typical of the Yanks to do so.

[Thomas Richter]

Originally Posted by chb

No, most C64s won't show weird colours, as a) the pixel clock is deliberately choosen not to be in sync with the color clock,

This is not a "choice", but a dependency of the underlying TV standard. At NTSC, the color clock is half of the pixel clock (and that is not a choice, it is a "shall"), at PAL, the issue is more complicated, namely 5/8 of the pixel clock.


There is no way for a TV signal to represent red color bars on green background with a hi-res output (try that!).

[Bruce Abbott]

Originally Posted by Foebane

Except that sort of thing didn't work so well in PAL!

Or didn't the Americans give a stuff about overseas territories at the time? Would be typical of the Yanks to do so.

Of course not, why would they? Even the biggest computer company in the World, 'International' Business Machines, couldn't be bothered making a PAL version of their Color Graphics Adapter.

But the reason for the divide between PAL and NTSC is much older. The TV frame frequency was set to the AC mains frequency to reduce the visual effect of hum (which was impossible to eliminate in valve TV sets). In the early days of electrical power distribution many different frequencies were used, often to match the device being powered (eg. 16.66Hz for induction motors, 133Hz for lighting) but in America 60Hz was popularized by Westinghouse while in Europe AEG set the standard at 50Hz.

With modern TVs it isn't a problem anymore, as most support multiple standards. However I have 2 Samsung LCD TVs that do not support 50Hz over HDMI. This is a pity because it means I can't use the Vampire RTG resolution that I want (1024x768). Luckily I have a cheap 26" LCD that does it perfectly.

[Gorf]

I would add: financing

Amiga (the startup) did not get enough money to bring its computer to the market on its own. First they needed additional money from Atari, which then led to the commodore buy-out ....

Some members of the original team learned from the experience and tried a licensing model with the 3DO console. But that did not turn out any better in the end.
(too expensive and a year or two too late... )

With better financing and a working prototype in 1984 a licensing model might have worked out for Amiga, like it did for Microsoft with the MSX - at least for a while.
In the 90s simply every alternative computer platform* died - no matter if it was a "home computer" or a "workstation".

*) Apple only survived with the help of MS

[BastyCDGS]

Not allowing TAS/CAS/CAS2 instructions is really one thing they did wrong from day 1.

After studying multithreading I realized how incredible useful those instructions are really.

They would have allowed, for example, sharing data between PPC and 68k without using locking (mutexes) and therefore without context switches (which were really expensive with >= 1ms duration)!

cas2 for example, allows implementing linked lists and queues without any blocking mechanism like Semaphore.

Have you ever wondered why every 68k accelerator disables the on-board already present CPU? Though it could have been used as a second core?
It's because of this!

[TEG]

Originally Posted by BastyCDGS

Not allowing TAS/CAS/CAS2 instructions is really one thing they did wrong from day 1.

By "not allowing" you mean "not safe due to hardware design" I suppose. I had to find this thread to understand why you said that.

[Foebane]

I'm surprised that the stock Motorola 68000 even supported multithreading back in the 1970s, if that's what this is about. But the Amiga chipsets were designed for the 68000, but single core and single threading only (AFAIK). I'm just guessing here, BTW.

[sandruzzo]

Originally Posted by Arne

It's already been mentioned, but I'd say the Amiga should've been more or less bundled with a good joypad. IIRC they were working on controllers during the secret early development, as a sort of front for the company. Then we all ended up with Tac-2s.

I didn't like Amiga Basic. It was rather unwieldy to use. It would've been cool to boot into a more powerful terminal-basic combo. I wonder how that would've played out, with intuition on disk instead... bad for chipram likely.

Generic (useful) 16 colour palettes is one of my favourite topics. No matter what, one has to make a decision on style and sacrifices though. It's useful to pixel a few things (as many as possible really) as it's incrementally informative. Often I end up changing the palette to accommodate for the new subjects while supporting the old. Generally I'd say the palette should feature not just graphical/popping colours for text (RGBY), but also skin tones and nature colours, and most old palettes don't do those well.

Here's my 16 colour AMOS default palette (WIP), with 16 sprite register colours which are hue offset. It's arranged in 4 colour ramps so blitting tricks can be used to shade/highlight.



I also have a few palettes on my site.

[Thomas Richter]

Originally Posted by BastyCDGS

Have you ever wondered why every 68k accelerator disables the on-board already present CPU? Though it could have been used as a second core?
It's because of this!

Well, this is really over-simplifying the problem. It is not because of the instructions (or rather, this is only the tip of the ice) not being supported by the bus. It is also a matter of the design of the board to release the 68K for working. Actually, on the P5 PPC cards, the 68K and the PPC are working on the same local bus (namely that of the card) and there, in their isolated territories, could have used TAS/CAS - or maybe they even do. As the communication only happens on the same board, and does not go to the outside then, TAS/CAS would have been fine.


The problem is that if for every data exchange you have to wait for one CPU or the other, you loose performance, i.e. every communication from the PPC to the system has to go ping-pong to the 68K is the problem Not exactly how this communication works.


I do not know enough on these boards to say how exactly they worked. In principle, you could get away with bus snooping (both PPC and 68K support this) and read/modify/write instructions, all provided you can design a bus interface for both CPUs that support this. Whether P5 hardware works this way I do not know, but nothing on the Amiga side prevents a TAS/CAS as long as the RMW-instruction does not leave the board.


It becomes problematic if the target of RMW is in chip ram (unsupported) or in Z-II/Z-III RAM.

[Thomas Richter]

Originally Posted by Foebane

I'm surprised that the stock Motorola 68000 even supported multithreading back in the 1970s, if that's what this is about. But the Amiga chipsets were designed for the 68000, but single core and single threading only (AFAIK). I'm just guessing here, BTW.

This is quite orthogonal - multithreading has no impact on chip design. Windows can do multithreading on legacy PC hardware without a problem, and so can AmigaOs on its chipset.



Multi-core(!) has some impact on the bus design, and this is where we have the problem, but multi-core was not realistic back then.

[drHirudo]

Originally Posted by Thomas Richter

Multi-core(!) has some impact on the bus design, and this is where we have the problem, but multi-core was not realistic back then.

Isn't the Amiga multi-core from Day 1?
The 68000 can work independently from the Agnus (Blitter, Copper), the Denise and the Paula. The Paula can do DMA, the Blitter can blit data, while the 68000 does other tasks.

[Thomas Richter]

Hardly. If so, the C64 was "multicore" and the PC was "multicore" as well. I mean SMP, to be precise, and this the current bus design does not allow.

[sparhawk]

Originally Posted by Foebane

I'm surprised that the stock Motorola 68000 even supported multithreading back in the 1970s, if that's what this is about. But the Amiga chipsets were designed for the 68000, but single core and single threading only (AFAIK). I'm just guessing here, BTW.

The M68000 doesn't support multithreading. The Amiga OS does that, but this can be done on any machine. I did it on an 8086 back then, and also on the C64. But of course, if you try to generalize that you run into the problems that modern OSes and MMUs are solving.


So running multiple threads is rather easy to do.

[Overdoc]

Since I had used a C64 for many years before the Amiga, for me the biggest Amiga disappointment from day 1 was that no operating system was built into the machine, not even a minimal one.

Whenever you wanted to do anything, you needed to boot the Workbench disk first.
And even after loading, it needed to load every single Amigados command from disk separately!

So, with a stock Amiga 500 without a second external diskdrive, you practically couldn't do anything except playing games without getting frustrated from permanently swapping disks.

[Thomas Richter]

Originally Posted by Overdoc

Since I had used a C64 for many years before the Amiga, for me the biggest Amiga disappointment from day 1 was that no operating system was built into the machine, not even a minimal one.

Eh, wait. Kickstart is not an operating system?

[Overdoc]

Originally Posted by Thomas Richter

Eh, wait. Kickstart is not an operating system?

Oh, sure, and what a great one:
Is there anything else it can do except waiting for a disk to load?
But seems a little bit poor compared to my 10 years older PET 2001 from 1977.....

[Thomas Richter]

Originally Posted by Overdoc

Oh, sure, and what a great one:
Is there anything else it can do except waiting for a disk to load?

Yes, a lot of things. Drawing windows on your screen, running a shell, looking for expansions, expansion ROMs, playing sound, computing in floating point...

Just in case you didn't know, those things are all in the kickstart, just waiting to be used, just that there is at this point not yet a program using it. The only thing it waits is a boot disk that calls InitResident() on the dos.library. Once that is done, you get already your boot CLI, your boot window, the ROM functions, intuition...

The Amiga is not a toy computer, unlike the C64. They did right *not* shipping it with a bad excuse of a Basic in ROM.

[Overdoc]

Thanks for the explanation, ofcourse I know that the kickstart is an operating system.
But you already answered it yourself: It is just not accessible from the start of the computer!
When you turn on your Amiga, the operating system is not there, it is not complete, yet, und you cannot access and use it until you boot an additional, missing part, which is called 'Workbench disk'.....

And this is the problem which did make everything so tedious. You couldn't do anything with your Amiga 500 until you waited 1 minute to boot the Workbench, and then you became a diskjockey if you wanted to do anything on a disk which was not the Workbench disk.

Developers forgot that it is complete bullshit to boot the operating system from a disk. This is something which only works from a harddisk. But harddisks were expensive as hell back then, and so only few people had them. Most just invested in a second diskdrive, which was mandatory on the Amiga if you wanted to do anything else than gaming.

But you are correct, the Amiga is not a toy computer, it is more a gaming console, than a computer
90% of all Amiga users I know never did anything else than putting a disk in the drive to play games.
This is very different for the C64 users I know, including myself. The C64 gave the possibility do write programs and explore the system right from the start.

I think the Atari ST did it correct, and had it's TOS in ROM, afaik?

[Thomas Richter]

Originally Posted by Overdoc

Thanks for the explanation, ofcourse I know that the kickstart is an operating system.
But you already answered it yourself: It is just not accessible from the start of the computer!

Of course it is accessible, to any code that sits on the boot medium.

Originally Posted by Overdoc

When you turn on your Amiga, the operating system is not there, it is not complete, yet, und you cannot access and use it until you boot an additional, missing part, which is called 'Workbench disk'.....

I'm sorry, but that's really a misconception. The operating system *is* there. What, if not the operating system, is drawing the "Please insert disk" screen? And, no, it does not require the "workbench disk". It requires a program that uses the operating system, in one way or another.

Originally Posted by Overdoc

Developers forgot that it is complete bullshit to boot the operating system from a disk.

Again, you "don't boot the operating system from disk". The operating system sits in the kickstart ROM. Yes, really. The disk contains only a couple of service programs, but you don't need that disk. Hint: Format "df0:", followed by "install df0:", then reset the system, insert the disk.


Voila, you have your operating system all ready, from an empty disk!

Originally Posted by Overdoc

90% of all Amiga users I know never did anything else than putting a disk in the drive to play games.

And, 90% of the C64 users didn't do anything than "LOAD "$",8", followed by "list", so things were considerably simpler for those lazy people.

Originally Posted by Overdoc

This is very different for the C64 users I know, including myself. The C64 gave the possibility do write programs and explore the system right from the start.

So did the Amiga. There was "AmigaBasic".

Originally Posted by Overdoc

I think the Atari ST did it correct, and had it's TOS in ROM, afaik?

You still don't get it. The Amiga has its Kickstart in ROM as well. That's all there, it just needs activation.

[chb]

Originally Posted by Thomas Richter

You still don't get it. The Amiga has its Kickstart in ROM as well. That's all there, it just needs activation.

But a user cannot get into the OS without having a prepared disk. So without that, you can't even list disk contents or format an empty disk. It is IMHO an oversight not to have a minimal, user accessible system running solely from ROM. AFAIK you cannot even enter romwhack without a disk or some hardware modification that provokes a guru - despite the fact that everything is already there in ROM. I fail to see a good reason for that.