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.

Are We Just Nostalgic Or Should the Amiga Advance?page  1 2 3 4 5 6 7 8 9 10 11 12 13 

Wawa T

Posts 695
11 Jul 2019 03:12


@Renee Cousins
too many cooks? then there might have been already too many involved with the original. you cant do such a scale project all alone i suppose.

as i said im not trying to force aros upon anybody. im perfectly fine with plain 3.1 or 1.3 if it suits the users needs as is. also a moderate update as thor performs certainly has a value. to each his own.


Steve Ferrell

Posts 424
11 Jul 2019 04:00


@wawa t

I admire the work you're doing and want it to continue.  I'm sorry if my comments have you thinking otherwise.  I'm merely pointing out that AROS 68K has a long way to go....But it also has come a long way since 1995 and I in no way want to discount the blood, sweat and tears that many dedicated developers have put into AROS at no charge.

I think our misunderstanding comes from the perspective of a service-for-hire developer (me) versus a volunteer coder.  Let me explain.

In my world, people often incorrectly assume that the software development clock starts as soon as a coder sits down at the keyboard.  This is simply not true, it starts much earlier for me and consists of 4 phases which are punctuated by several milestones along the way. In my world, the clock starts counting at Phase 1, or by setting a design goal.  In regard to AROS, the design goal was published in 1995.

Phase 2 consists of a resource assessment where the necessary tools (compilers, documentation, skill sets, people) are identified in order to reach the goal(s) identified in Phase 1.

Phase 3 is the budgeting and acquisition of the things identified in phase 2, if they are not already on hand.

Phase 4 begins next, which is the coding.  In the case of AROS 68K it may have taken years to complete phases 1 thru 3, find volunteer coders, and move on to coding, but nonetheless, development began in 1995 with the publishing of the stated design goal.  For volunteer coders, I can see why they would start counting the clock at the coding phase or whenever they entered the project or obtained funds, which is fine too from their perspective since all the resources have to be obtained for free and no one earns a salary.
 
So in my world, AROS development began 1995...for open source devs, well, I'm sure every one of them will come up with a different answer and that's fine too.  What's important is that the work continues and I look forward to the day when we can toss aside OS 3.x and all it's legal problems, but we're not quite there yet.


Renee Cousins
(Apollo Team Member)
Posts 142
11 Jul 2019 04:14


Okay. Why is True Type better than Compugraphic?


Wawa T

Posts 695
11 Jul 2019 04:18


@Steve Ferrell
thats all fine and good, lets not quarrel since when we count development, but we are not in the position to go through all the phases you describe, identifying resources, budget and the like.
the situation is highly unstable, dynamic if you will, like all amiga. people come and go, quarrel, drama, its hard to keep it steady, remain motivated, even on aros. the advantage of the situation is as i said that the source remains, and can be forked if need be.
but if hacking the existing system is a better way, so be it, its already being done, just dont count me in, i cant help with this much.


Renee Cousins
(Apollo Team Member)
Posts 142
11 Jul 2019 04:21


What advantage does HIDD offer above RTG and AHI?


Renee Cousins
(Apollo Team Member)
Posts 142
11 Jul 2019 04:22


How would SMP help M68K?


Renee Cousins
(Apollo Team Member)
Posts 142
11 Jul 2019 04:22


How would process-level memory protection retain any compatibility?


Wawa T

Posts 695
11 Jul 2019 04:23


Renee Cousins wrote:

Okay. Why is True Type better than Compugraphic?

does anybody says its better? tbh i dont really know much about compugraphic. let me bite, true type is a widely used standard and a lot of fonts are available there, it may be used in professional application, where it already surpassed postscript fonts, as far as im aware.
but then, im all for compugraphic if anyone feels it worthy and wants to implement it.


Renee Cousins
(Apollo Team Member)
Posts 142
11 Jul 2019 04:25


How do you expect to overcome the poor code generation of gcc if you insist on remaining "Pure C/C++"? And isn't that a lie anyway? AROS is full of x86 machine code. I guess portability only gets you so far...


Wawa T

Posts 695
11 Jul 2019 04:26


Renee Cousins wrote:

What advantage does HIDD offer above RTG and AHI?

thats what michal or kalamatee could answer, i lack the insight so i take it as given. i guess the modularity is attempted.


Wawa T

Posts 695
11 Jul 2019 04:28


Renee Cousins wrote:

How would SMP help M68K?

i dont think it can be brought to m68k, at least without breaking backwards compatibility. its currently only meant and enabled for x86_64.



Wawa T

Posts 695
11 Jul 2019 04:28


Renee Cousins wrote:

How would process-level memory protection retain any compatibility?

i dont think it would. thats the limitation on aros as well as on genuine amiga os.


Wawa T

Posts 695
11 Jul 2019 04:36


Renee Cousins wrote:

  How do you expect to overcome the poor code generation of gcc if you insist on remaining "Pure C/C++"? And isn't that a lie anyway? AROS is full of x86 machine code. I guess portability only gets you so far...
 

  later gcc generate apparently better code, bebbos 6.5 is almost on pair with 2.9.x which was claimed to be best gcc for amiga for years. it has surpassed vbcc without its limitations as far as i know,
 
  our gcc is not so heavily patched to ensure better maintainability. we are currently using 9.1 with 8.3 and 6.5 as fallback. also clang llvm is there. albeit its not really being used.
 
that said, the critical code, especially in arch section may be asm inlines or in rare cases whole asm sources. mostly low level bare metal stuff where it is necessary, not only x86, m68k as well. portable code is c, in some cases c++ (gallium). i assume sensible combination of these is the key.


Renee Cousins
(Apollo Team Member)
Posts 142
11 Jul 2019 04:55


I have a non-ranty question.

Can the original AmigaOS Exec use AROS executables, libraries and devices?

Conversely, can AROS use existing AmigaOS executables, libraries and devices?

If graphics.library, layers.library and the Picasso96 subsystem is the problem, then can we not just replace the implementation in AROS for the real P96 library?


Wawa T

Posts 695
11 Jul 2019 05:15


Renee Cousins wrote:

  I have a non-ranty question.
 
  Can the original AmigaOS Exec use AROS executables, libraries and devices?
 

  yes, as long as the dont rely on aros dependencies and of course if the binaries are in hunk format, which they are being converted to from elf via elf2hunk util when building distfiles.
 
 

  Conversely, can AROS use existing AmigaOS executables, libraries and devices?
 

  yes, of course only amiga-m68k aros. perhaps big endian arm aros will be able to do that as well, in the same manner os4 and morphos do , just on arm instead of ppc.
  for instance as ferells said, you can and sometimes must use closed source amiga devices, libraries or classes available only in binary form. you can replace whole subsystems if you will, such as running classact(reaction) under aros, warpos or substituting zune with mui as olaf did in his distro a while ago.
 
 

  If graphics.library, layers.library and the Picasso96 subsystem is the problem, then can we not just replace the implementation in AROS for the real P96 library?
 

  im not sure. the libs you mention are very much interwoven so this in particular may not be possible. tbh i have not tested this. for instance though i assume you would lose quite a bit of functionality replacing grephics.lib. i think it doesnt make that much sense to do that, it would be better to improve aros counterpart rather than try to hack in an older genuine version of a subsytsem.
 


Wawa T

Posts 695
11 Jul 2019 05:18


btw how would aros make sense if it was not backwards compatible. but it doesnt mean it can be forwards compatible in every case. if some software is using aros exclusive features it would need thease feats implemented on amiga os as well to run. i think its self explanatory.


Wawa T

Posts 695
11 Jul 2019 05:25


for example, as a matter of experiment i have odyssey compiling for m68k. its a huge beast and it doesnt really start yet. it needs a number of stuff unavailable on amiga. pixmann, ttf, cairo a lot of stdc especially math was missing. if i would have to port it all i would never get it compile, but with those parts in place it was doable for a noob like me with some assistance. 


Stefan "Bebbo" Franke

Posts 139
11 Jul 2019 06:31


wawa t wrote:

for example, as a matter of experiment i have odyssey compiling for m68k. its a huge beast and it doesnt really start yet. it needs a number of stuff unavailable on amiga. pixmann, ttf, cairo a lot of stdc especially math was missing. if i would have to port it all i would never get it compile, but with those parts in place it was doable for a noob like me with some assistance. 

use my amiga-gcc and all is fine^^


Gunnar von Boehn
(Apollo Team Member)
Posts 6197
11 Jul 2019 06:53


As we all exchange our wishes maybe I can share my view

For me, AMIGA OS is in its internal design an adorable mix of
- very modern ideas (datatypes, boobsi)
- extern slim and elegant design
  (WB fits on 1 floppy)
- very fast
- not cluttered with endless layer of abstraction
- very upgrade and patch friendly design

What I would like is to see AMIGA OS to continue
and improve based on has it today - keeping its design ideas and keeping its spirit.

I would like to see that the external RTG chunky solutions are merged back into graphics in an AMIGA fashion way and that planar
and chunky screens will more smooth together.




Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
11 Jul 2019 07:13


wawa t wrote:

  i dont think it can be brought to m68k, at least without breaking backwards compatibility. its currently only meant and enabled for x86_64.

(SMP) And there was no need to. Maybe some dual core 080 SOCC would be enabler one day - but we dont have dual core m68k now.

Mr Niding wrote:

        Renees point is not wether we move beyond 020 as baseline performance level, but if the OS require xxx % performance increase to complete the same tasks.
     

     
      In those terms I find AROS closest relative, to old ugly and very basic OS 3.1 install. People tend to forget how OS 3.9,OS 3.1.4 and any jampacked OS 3.x with 32-bit screens, color icons and many OS tweaks eat a lot more RAM and CPU. Here you get most of that on boot, and not at a high price.
    My biggest objection is usually that basic AOS installs are to software wise poor. So I prefer things like AmiKit and AROS Vision.
 
  I do agree it needs more then 030+FAST and even 080-RTG optimizations, but its best chance of OS developing. Even its state is not inspiring - out of box offers less then competitors. Thus AROS Vision and such distros were born. I dont believe MorphOS team is thinking of 080 as new CPU arch :)
     
     
Steve Ferrell wrote:

      1.  ApolloOS is free
     

     
      Freely downloadable, essentially you need legal OS 3.9 to have it for a kind of free. Its not new OS, its specified OS 3.9 distro.
      No AmigaOS like is free, except AROS.
   
   
wawa t wrote:

    for example, as a matter of experiment i have odyssey compiling for m68k. its a huge beast and it doesnt really start yet. it needs a number of stuff unavailable on amiga. pixmann, ttf, cairo a lot of stdc especially math was missing. if i would have to port it all i would never get it compile, but with those parts in place it was doable for a noob like me with some assistance. 
   

   
    Coders, please join. We need that Odyssey AROS m68k :)
 
  @Gunnar
 
  I believe AROS fulfills most of AOS heritage list.

posts 244page  1 2 3 4 5 6 7 8 9 10 11 12 13