Overview Features Instructions Performance Forum Downloads Products OrderV4 Reseller Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
The team will post updates and news here

Future of Amiga Platformpage  1 2 3 

Robert Downs

Posts 28
22 Aug 2019 21:55


Gunnar has on many occasions indicated that doing something like this is beyond Vampire's goals.  They want to resurrect the 68K architecture.  Spending time on PPC would detract from the 68K efforts and prove to have little benefit.

It is an interesting thought experiment however.


Olle Haerstedt

Posts 102
22 Aug 2019 23:17


What about adding an expansion port to the Vampire that makes it possible to add a PowerPC CPU, much like A1200 could do in its days?


Steve Ferrell

Posts 397
23 Aug 2019 04:35


Olle Haerstedt wrote:

  What about adding an expansion port to the Vampire that makes it possible to add a PowerPC CPU, much like A1200 could do in its days?
 

 
 
I doubt that such a board would outperform an existing A1200 with a Vampire board.  Remember that Gunnar has highly optimized the 68080 and added 21st century features to a 20th century CPU.  I'd wager that the 68080 softcore would match a PPC softcore in performance, if not better it.
 
The only reason to add a PPC accelerator to an A1200, either FPGA or real silicon, is to run OS4 and I don't think there's enough interest out there from A1200 owners to warrant designing such an accelerator.  I may be wrong though.  There was a lot of interest a while back when some scam artist "appropriated" the Phase 5 name and created a small batch of Cyberstorm PCB's.  Unfortunately the guy was off his rocker and he also forgot to put a ground plane on any of the boards he produced.  It was a total debacle but there was a lot of interest from OS4 users. There's more about the debacle here:  EXTERNAL LINK   

I would guess that a PPC softcore would be of more interest to die hard OS4 fans in the form of a standalone board with a PCI slot for graphics.  It shouldn't be too difficult to replicate an Amiga XE in an FPGA.  The CPU softcore has already been written.  It would probably outperform the Tabor if the Tabor ever sees the light of day!  LOL!
 
 


Eric Gus

Posts 415
23 Aug 2019 06:53


Steve Ferrell wrote:

@eric gus
 
 
As hard as I try, I just can't justify any real reason to own a PPC system in 2019 other than to post on social media that I own one.

 
  There's no justifiable reason to own a Vampire, classic Amiga, or an X1000 either.....Amigas are a retro-hobby and they aren't in competition with any current technologies, but unfortunately we have crackpots who show up here believing otherwise who have delusions about how the Amiga could once again take the world by storm if there were just one or two "small" changes.  That ship sailed and then sank around 1988.
 
 

Well at least with a vampire I can "sorta" marginally justify it in that I can use it to run the old classic games and apps and sundry with a reasonably accurate recreation of the original hardware so I don't have to use my original machines as "daily drivers" and that is at least something more than I can fess up for a PPC system. But ya in the end its a retro hobby system and thats all it will ever really be. The world has just changed far too much.



Chastanier Cclecle

Posts 7
23 Aug 2019 10:05


If you want 68080 + PPC, one possible solution will be:
- Amiga 1200
+ Vampire 1200
+ Mediator 1200
+ Any compatible PPC Card (Sonnet ...etc)
+ (optional) Any compatible PCI GFX card

The limitation will be the connexion beetween Vampire RAM and PPC that will pass trough the zorro II (slow) port. But, the PPC card have its own RAM (and can access PCI GFX RAM) so in most use-case it should work well.

Only question is, is the Vampire 1200 compatible with mediator. I would say there is no reason, but actually nothing is sure until it is not tested ! Wait and see :).



Harold Joseph Neufeldt

Posts 4
20 Sep 2019 00:31


I think I am going to contribute to this discussion in a more general manner. When you are talking about the future of amiga you are talking about many things.

I will start with AmigaOS 3.x. With Thomas Richter and Olaf Barthel completing 3.1.4 and starting 3.2,things seem to be going quite well in that area. I'm not so sure about application software though. While there seems to be a shortage of coders the leadership is there.

What seems to me a better solution is the development of AROS for the platform. I do have a few caveats that need to be considered..  Firstly we need some leadership in place. Secondly, clear goals and aims need  to be in place. There are also some legal hurdles that may have to be dealt with.  This will need to be a long term project that will take several years to complete.

I will now talk a little about OS4.x. The situation here is actually quite complex. A few weeks ago there was quite a little kerfuffle onAmigaWorld about the disappearance of the AmigaOne brand.  Most of the people who posted missed the main point. The reason the brand disappeared is that there will be no more AmigaOne computers made.  There is now no reason to advertise the brand.  You will still be able to buy X5000s because they have quite a number still in inventory.(This was the case before I started writing this post)  They have not manufactured any boards since ultrvarisys was acquired by a major British company.  Wait a minute! What about Tabor?  My belief is that it was cancelled some weeks ago but it hasn't been announced yet.  If it was to go into production they would not have cancelled the brand.  It still could go into production as a pure Linux machine but I think this is highly unlikely. If it were to go into production as an AmigaOne A-Eon would likely owe Hyperion some $125,000USD. It appears the PPC era in Amigaland is over. One major problem is thatOS4 cannot be ported to advanced PPC processors. It will only run on obsolete processors which are EOL.

On the positive side, software development is going ahead full speed. Enhancer 2.0 is on the way.  I should have begun by saying that A-Eon is developing three modules Enhancer, a graphics module and a multimedia package. These are all going ahead full steam.  The good news is that they all will be back ported to3.x. The bad news is that will not happen today or tomorrow. One problem being that none of the modules will run properly on Vampire.

I should say a word or two about Hyperion___whether I want to  or not.  In my humble opinion they are now irrelevant . Other than the revenue they are getting from the OS 3 updates they don't have any.  They aren't in any position to do any real development.

In conclusion, I will say that Amiga has a future. How bright it is only time will tell.  I've omitted a great deal of what I originally wanted to say, which is probably a good idea. I see that some people have commented on a couple of projects I wanted to comment on. I'll do that when I reply to their posts.



Gunnar von Boehn
(Apollo Team Member)
Posts 4846
20 Sep 2019 06:41


Thank you Harold, for your excellent post.
You perfectly summed up the whole situation.


Eric Gus

Posts 415
20 Sep 2019 06:45


@harold, while interesting I personally dont think or classify the PPC/NG machines as "Amigas" proper.. they are "amiga like" and inspired by Amiga/AmigaOS but they are for me just not Amiga. I have never paid much attention to the NG systems and more so once I came back to the amiga community and took a look at the price that was being asked for them.. yea no.. its strayed too far from the classic amiga and they are frankly too expensive and too technologically too far behind to be placed with current/modern day desktops..


Gunnar von Boehn
(Apollo Team Member)
Posts 4846
20 Sep 2019 07:43


Steve Ferrell wrote:

  People will buy PPC FPGA's for the same reasons that people buy Vampires..
 

 
 
PPC Softcores are not new.
I myself have worked on a PPC Softcore many years ago for IBM.
 
IBM did actually compared its own IBM PPC Softcore with APOLLO-Core in several benchmarks - and APOLLO did win most of these benchmarks!
 
This is no wonder at all and very easy to understand.
A Softcore in an FPGA is limited by the technology.
This is valid for both the PPC and for 68080,
and both cores reached in the IBM comparison similar MHz numbers.
 
Now is it a fact that a PPC needs in general more instructions to do the same work compared to an 68K.
 

Lets make this more clear with numbers:
Example 1:


  ADDI.L #$123456,$02223344
1 instruction on 68K, the PPC needs 7 instructions to do the same.
 

Example 2:


  ADD.L D0,(a0)
1 instruction on 68K, the PPC needs 3 instructions to do the same.

 
Example 3:


  ADDi.L #11223344,D0
 
1 instruction on 68K, the PPC needs 3 instructions to do the same.
 
 
 
Example 4:

  ADD.L D1,D0
1 instruction on 68K, the PPC needs 1 instructions to do the same.
 
 
As you see in average the PPC will need a lot more instructions to do the same amount of work as the 68K.
 

If both 68080 and PPC softcore run at roughly the same clockrate, and both execute roughly the same amount of instructions - then 68080 will naturally be a lot faster - as it can do with the same amount of instructions a lot more work.
 
This is just natural and very clear to understand.

As a ballpark estimate you can say that todays PPC softcore will reach about 50% of the speed of APOLLO 68080.
 
 
Will 50% of the speed of APOLLO be enough to run your OS 4 software?
 

You also need to mind that FPGA cost a lot of money.
This means adding an PPC Softcore to Vampire is NOT free but will mean adding $100 of more FPGA space on the card!
 
A much cheaper option could be to run a PPC Emulator on the 68080.
A good JIT Emulator running on 68080 might reach the same speed as the PPC Softcore  - but will cost not FPPA space - so this would be much cheaper!

Also you need to ask yourself the question - is the software you want to run available in source code?
Old Amiga games are NOT available in source code.
But for OS4 Apps the source should all be available today.
So compiling for 68K might often be a good alternative.


Richard Gatineau

Posts 58
20 Sep 2019 08:03


I agree that PPC emulator on 68080 is a good solution for peoples that require to continue to run some PPC applications. (at less the time to port them to 68080 if needed)

PPC is the past and ARM will be a better choice today. Now, the 68080 is the best alternative in the Amiga's spirit, and for its retro-compatibility.


Gunnar von Boehn
(Apollo Team Member)
Posts 4846
20 Sep 2019 08:18


With ARM is the same situation.
And ARM CPU is clock by clock really a lot weaker than 68080.

If we ever reach the goal to make 68080 ASICS then we will dance around all ARM CPUs.



Gunnar von Boehn
(Apollo Team Member)
Posts 4846
20 Sep 2019 08:33


Richard Gatineau wrote:

I agree that PPC emulator on 68080 is a good solution for peoples that require to continue to run some PPC applications. (at less the time to port them to 68080 if needed)

Please help me to understand the situation better:
I was not aware that there is much important PPC software at all.

Can you name some Applications are really "important" and are "PPC" only?




Thellier Alain

Posts 118
20 Sep 2019 09:18


>require to continue to run some PPC applications.

If you wanna still use the programs you used on a A4000 + ppc card (warpos or OS4) then most of them are running on WinUAE-ppc
So no need for a new vampire ppc emulator

For more recents OS4 programs (games) that use Warp3D,Compositing or Nova then a ppc emulator will be not enough are those somponents are not emulated in WinUAE(or as software only...)
  or implemented in Vampire...

 


Olaf Schoenweiss

Posts 647
20 Sep 2019 11:13


AROS is open source. You can today branch the sources and concentrate on Vampire

You do not even need to ask

The only problem is to find developers


Gunnar von Boehn
(Apollo Team Member)
Posts 4846
20 Sep 2019 11:28


Olaf Schoenweiss wrote:

AROS is open source. You can today branch the sources and concentrate on Vampire

You are right AROS is open source, and this could be done.
Or fixes for 68K could nicely be put back into the AROS main.

Olaf how about discuss some of this in detail?
Would you mind to meet on IRC today or phone?


Olaf Schoenweiss

Posts 647
20 Sep 2019 11:30


I am not a system programmer and cannot do that


Gunnar von Boehn
(Apollo Team Member)
Posts 4846
20 Sep 2019 11:33


Olaf Schoenweiss wrote:

I am not a system programmer and cannot do that

This is OK. We have developers in the Apollo-team and I think you can help plan / organize and prepare the distro.


Andy Hearn

Posts 318
20 Sep 2019 12:00


i've personally only ever run amiga PPC stuff on the "classic" environment of both A1200 and A4000. i still have all my A1200 PPC stuff. but i doubt i'll ever fire it up again once i get my hands on the 1200 variant of the Vampire.

I can sum up all of my amiga PPC experience with one phrase:-
an absolute effing ball ache.

I'm done with PPC. this is consigned to history in my mind.
long live 68k!


Wawa T

Posts 695
21 Sep 2019 15:51


Gunnar, you still better talk to nick . olaf may provide you with some user experience as to what needs to be done but he has never contribited code and i doubt he knows the source. I do a bit. But you better ask who knows best. Ideall have people join our slack.


Gunnar von Boehn
(Apollo Team Member)
Posts 4846
22 Sep 2019 08:13


wawa t wrote:

Ideall have people join our slack.

AROS works on V4-SA.
We have compiled it and it runs.
 
There are a few things were it can be polished.
Maybe the fastest way forward would be for today to "Fork" AROS Git,  brush up all we can as fast as possible and then provide the fixes back into AROS main.
 

posts 58page  1 2 3