Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Information about the Apollo CPU and FPU.

Why Was the Amiga Such a Success?page  1 2 

Gunnar von Boehn
(Apollo Team Member)
Posts 6196
31 Jan 2023 17:48


Samuel Crow wrote:

Gunnar von Boehn wrote:
Please help me understand what software I today miss out.
  Which Software would I be able to use if the V4 would have e.g 64 GB of memory?

  Being able to run your VHDL compiler or System Verilog for anyone else so-inclined, on your existing Amiga-compatible computer would be nice, wouldn't it?  Of course you still use a PC to do that.  Most VHDL compilers are not ported to Amiga.  On GitHub there is an organization called the "Chips Alliance" to develop FPGA cores with open-source software.  SiFive is backing the effort.  I'm pretty sure the compilers for that are 64-bit only.

And for this you want to break compatibility with all Amiga software?

Does the software compile for Amiga OS?
How long do you think you will need to port it?



Samuel Crow

Posts 424
31 Jan 2023 18:02


Gunnar von Boehn wrote:
And for this you want to break compatibility with all Amiga software?

I'm typing this on my 64-bit Haiku system right now.  Multithreaded and large software runs on it quite nicely.  It has all the features of AmigaOS but is based on the design of the BeBox.  BeBox was originally PPC.  The version I'm using is x86-64 though.  Choose your path.
Gunnar von Boehn wrote:
Does the software compile for Amiga OS?
How long do you think you will need to port it?

It's open-source and I think it's written in Rust.  Rust compiles to WebAssembly and W2C2 compiles WebAssembly to the 1989 version of C.  All that's needed is the WebAssembly WASI runtimes.


Tommo Noorduin

Posts 130
31 Jan 2023 23:40


The 68k instruction_code is ready for 64 bit and more if that comes to that.
Address relative or pc relative code does not care if it is 32 bit, 64 bit or more.
Like the adda.w that is sign extended and added to whatever size the address might be.


Samuel Crow

Posts 424
01 Feb 2023 02:53


Tommo Noorduin wrote:

The 68k instruction_code is ready for 64 bit and more if that comes to that.
  Address relative or pc relative code does not care if it is 32 bit, 64 bit or more.
  Like the adda.w that is sign extended and added to whatever size the address might be.

Unfortunately, the library format uses absolute addresses all over the place.  The jump table entries are hardwired to 6 bytes; 2 for the opcode and 4 for the 32-bit address.  The reason they can't be fully position-independent is that the original 68000 and 68010 had only 16 bit relative addresses.  It wasn't until the 68020 that that grew to 32-bit position independence.

That means that the new operating system will need to either enforce 68020 as minimum and avoid 68000 16-bit code except for a backward compatibility layer like hosted AROS.


Gunnar von Boehn
(Apollo Team Member)
Posts 6196
01 Feb 2023 08:44


As you know our goal is to revive the 68k Amiga.
 
This means we want Amigas with nice to program 68K CPU together with clever and elegant Amiga chipset.
And our systems has to be compatible to the existing thousands of Amiga games and demos and with the existing Amiga OS.
 
Our systems come with 512 MegaByte of memory - this is one thousand times more than what the Amiga 500 came with.
 
We all know that switching the OS to 64bit addresses would break compatibility and also these new binaries would be fatter and slower. Neither of this is in our opinion desirable.
 
Let me give some more technical information to help illustrate this:
 
ALU:
A CPU will process data in its ALU.
A wider ALU enables the CPU to process more data in one instruction = this means a wider ALU makes the CPU faster
 
REGISTER:
A CPU does hold data in register.
Wider register can hold more data and help to feed the ALU.
Wider Register help to make the CPU faster.
 
BUSWIDTH:
A CPU does load or write data from a memory bus.
A wider memory bus allows the CPU to load more data per cycle.
Wider memory bus allows the CPU to be faster.
 
ADDRESS
On the plus side a Wider address mode enables to plug in more than 4GB of memory, but on the negative side it will makes all programs fatter and it will makes them also slower. But most important: A 64bit mode will break compatibility with all existing software.
 
 
As you see there are 4 areas in a CPU where it can be 64bit.
3 of these areas bring only benefits without any drawback.
And in all these 3 areas our APOLLO 68080 uses 64bit already.
 
One area brings a lot of drawbacks and breaks with existing software. This is something that we don't want.
I'm sure you understand this.
 
I look forward to see more from you.
Please also  help to revive Amiga and program something for the people.
How about coding a nice game?
 


Gunnar von Boehn
(Apollo Team Member)
Posts 6196
01 Feb 2023 09:47


Steffen Häuser wrote:

I do not see the 68080 as "enemy" as you see the PPC. 

   
I don't see PowerPC as enemy.
What I don't like is if people are lied into their face.
And I see this happened a lot in PowerPC marketing - IBM marketing is also very guilty of this.
   
I'm a PowerPC expect and I perfectly know where the strength and were also the weaknesses from PowerPC are.
I used to work for IBM in PowerPC design and development. Actually most of our core team are IBM PowerPC CPU engineers.
We know what strength a PowerPC has and what not.
   
Being PowerPC experts we can clearly identify when people are lied to and when false claims are made.
   
Lets look at some examples:
   
"Its RISC so its faster"
This is complete wrong. RISC is a trade off which first of all makes it need more instructions to do the same amount of work.
So it primarily makes it less performend.
   
   
"Its true that PowerPC ASM is not nice to code but no one needs ASM coding today. Compilers are good enough"
I can tell you that I spend years at IBM coding PowerPC assembly.
And I wrote several tens of thousands lines of PowerPC assembly - because compilers are not good enough. So this is again total nonsense
 
"Yes, PowerPC does not support this feature in Hardware, as its better to do this in software."
Generally its always better to solve things in hardware. But the sad truth is hardware development is under extreme time pressure today. Companies want to bring out new models quickly, and they want to less money for hardware developers. This means hardware developers are under time pressure and often have not enough man power - to do a 100% work
So in reality often feature that the hardware engineers wanted to have in hardware - have to be left away /dropped to bring the product out earlier. Doing things in software is very often the "emergency solution" and never was the goal.
When someone claims that "doing it in software" was good/or on purpose - mind this might be market people just lying to you!
 
   
What I don't like is when people are lied in their face.
 
PowerPC and 68K are design choices. Both have advantages and disadvantages.
 
Like Tractors, Motorbikes, and Caravans are design choices - with their own advantages and disadvantages.
   
I think people need to get more information to understand these design choices better.


Cyprian K

Posts 26
05 Feb 2023 16:40


Steffen Häuser wrote:

  - Color Depth - 320x240 - x256 in EU - with 32 colors, or 64 colors with EHB was LIGHTYEARS ahead of the concurrence. EGA graphics ? Or whatever ST offered ? Inferior.

well, I'm not going to start flame-war but in 1985 Atari had a few advantages, which could be seen as useless from game players point of view, but from power user side were a killer features.
E.g. it was high resolution, high refresh rate VGA video mode. Similar feature appeared in Amiga 5 years later - ECS 50/60Hz only vs 71Hz in Atari. In additional, /|\ was bundled with SM124 - high quality, high-res paper-white monitor. Atari with that monitor and with 1MB RAM had lower price than competition with 512kB and without monitor.
I have SM124, C= 1802 and /|\ SC1435 on my desk now, and SM124 with mono mode is still way better.

Regarding Power-PC in my opinion it was a dead end, due to lack of 68k compatibility. Would be cool to have 68080 in '90. That probably would save Commodore and Atari world.

Regarding 32bit vs 64bit address space, there are application which needs a lot of memor on Atari (DTP or graphics manipulation) but at that moment 2GB RAM address range (31bit because of 32th bit means error) is more than enough also in Atari world.
Maybe porting FireFox browser would change that needs.


Adam A

Posts 130
12 Feb 2023 18:59


Cyprian K wrote:

Steffen Häuser wrote:

    - Color Depth - 320x240 - x256 in EU - with 32 colors, or 64 colors with EHB was LIGHTYEARS ahead of the concurrence. EGA graphics ? Or whatever ST offered ? Inferior.
 

 
  well, I'm not going to start flame-war but in 1985 Atari had a few advantages, which could be seen as useless from game players point of view, but from power user side were a killer features.
  E.g. it was high resolution, high refresh rate VGA video mode. Similar feature appeared in Amiga 5 years later - ECS 50/60Hz only vs 71Hz in Atari. In additional, /|\ was bundled with SM124 - high quality, high-res paper-white monitor. Atari with that monitor and with 1MB RAM had lower price than competition with 512kB and without monitor.
  I have SM124, C= 1802 and /|\ SC1435 on my desk now, and SM124 with mono mode is still way better.
 
 
  Regarding Power-PC in my opinion it was a dead end, due to lack of 68k compatibility. Would be cool to have 68080 in '90. That probably would save Commodore and Atari world.
 
  Regarding 32bit vs 64bit address space, there are application which needs a lot of memor on Atari (DTP or graphics manipulation) but at that moment 2GB RAM address range (31bit because of 32th bit means error) is more than enough also in Atari world.
  Maybe porting FireFox browser would change that needs.

For a gamer i think amiga had the edge when it comes to audio and was cheaper (around $100?)


Nick Fellows

Posts 176
12 Feb 2023 19:47


No mention of the operating system, which, when it came out offered full colour WIMP desktop with pre-emptive multitasking, and a host of content creation software like Dpaint , Lightwave. Amiga was much more than just the hardware.


Cyprian K

Posts 26
13 Feb 2023 00:12


Adam  A wrote:

  For a gamer i think amiga had the edge when it comes to audio and was cheaper (around $100?)
 

1040ST with 1MB RAM and highres monitor was cheaper than A500 with 512kB RAM without monitor.

nick fellows wrote:

No mention of the operating system, which, when it came out offered full colour WIMP desktop with pre-emptive multitasking, and a host of content creation software like Dpaint , Lightwave. Amiga was much more than just the hardware.

for me it is matter of taste

anyway, until last week I thought the Macintosh was a niche product in the 80's and the 90's, I it seems that Apple sold more that Atari ST and all Amiga models counted together. From Wiki:  "Between 1984 and 1989, Apple had sold one million Macs. It sold 10 million over the following five years."




Adam A

Posts 130
13 Feb 2023 00:46


Cyprian K wrote:

Adam  A wrote:

    For a gamer i think amiga had the edge when it comes to audio and was cheaper (around $100?)
 

  1040ST with 1MB RAM and highres monitor was cheaper than A500 with 512kB RAM without monitor.
 

Are you sure? I just googled it and found that Atari 1040ST was at $799 with monochrome monitor while amiga was $699 in the US. 
 



Gunnar von Boehn
(Apollo Team Member)
Posts 6196
13 Feb 2023 14:31


nick fellows wrote:

a host of content creation software like Dpaint , Lightwave. Amiga was much more than just the hardware.

 
Yes for sure software is important
 
But is it not true that Hardware attracted software?
I think the Amiga features enabled the software.
 
Would have been any effort to compile Dpaint for Atari?
I think it would have been no effort.
 
 
Dpaint was nice on Amiga because Amiga supported 64 colors on screen - while Atari could at max show 16 colors.
 
And the palette on Amiga offered 12 bit colors = 4096 colors while Atari only offered 9 bit.
This means the Amiga colors were simply 8 times richer.
 
Yes Dpaint is a nice paint program.
But did not the Amiga hardware enabled it to be show its potential?
 
Could have been DPaint on PC in EGA GFX with fixed 16 colors a similar success?
I think NO. For sure never would that been such a success.

Amiga Tracker were nice for making music.
Of course here the PAULA Amiga DMA audio channels enabled these trackers to be good on Amiga.




Cyprian K

Posts 26
28 Feb 2023 23:32


Adam  A wrote:

 
Cyprian K wrote:

 
Adam  A wrote:

      For a gamer i think amiga had the edge when it comes to audio and was cheaper (around $100?)
   

    1040ST with 1MB RAM and highres monitor was cheaper than A500 with 512kB RAM without monitor.
   

 
  Are you sure? I just googled it and found that Atari 1040ST was at $799 with monochrome monitor while amiga was $699 in the US. 
   
 
 

 
  It turned out that my memory fails me a bit, (I was sure that my 1040 STfm with SM124 was a bit cheaper than A500, maybe prices in Poland were a bit different...). Anyway below are the prices from  ST/Amiga Format Jan 1989:
  EXTERNAL LINK 
  A500  £369,95 (Page 38)
  520STfm  £269,95 (Page 38)
  SM124/5  £119 (Page 30)
  1040STfm  £419 (Page 30)
 


Örjan Olofsson

Posts 17
06 Mar 2023 12:32


Hi Gunnar!

In an earlier post you mentioned increasing the number of DMA
sound channels from 8 to 16/32, would that include all Vampires,
and would it suffice to make a core update once that work is completed?




Gunnar von Boehn
(Apollo Team Member)
Posts 6196
06 Mar 2023 12:38


Örjan Olofsson wrote:

  Hi Gunnar!
 
  In an earlier post you mentioned increasing the number of DMA
  sound channels from 8 to 16/32, would that include all Vampires,
  and would it suffice to make a core update once that work is completed?
 
 

 
 
Yes all our cards with Audio = all V4 based cards. (The V4-Standalone,V4-Firebird,V4-Icedrake,V4-Manticore)
Have been updated to provide 16 Amiga hardware DMA channels.
After update the FPGA core, you have 16 channel.

There are also already first games making use of the 16 Amiga audio channel and Amiga music players using them.

In my opinion this is huge improvement - and makes game porting and now game coding much easier.
 


Örjan Olofsson

Posts 17
25 Mar 2023 19:00


Gunnar von Boehn wrote:

  Yes all our cards with Audio = all V4 based cards. (The V4-
  Standalone,V4-Firebird,V4-Icedrake,V4-Manticore)
  Have been updated to provide 16 Amiga hardware DMA channels.
  After update the FPGA core, you have 16 channel.
 

That's great,it still says 8 channels in the shop that's why I asked :)

posts 36page  1 2