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
Performance and Benchmark Results!

Vampire/Apollo PCI Versionpage  1 2 

Michael Borrmann

Posts 140
15 Feb 2018 08:51


Yeah, let's enhance 3.9.

I don't think we can expect much from 4.0 in our 68k camp...


Rod March

Posts 119
15 Feb 2018 12:20


Working from 3.1 sounds good too (or whichever last version isn't part-owned by three competing interests who will never agree!).


Björn Persson

Posts 11
15 Feb 2018 15:05


Rod March wrote:

  Working from 3.1 sounds good too (or whichever last version isn't part-owned by three competing interests who will never agree!).
 

 
  Who owns 3.1?


Mallagan Bellator

Posts 393
16 Feb 2018 14:05


I think OS4 backport sounds great. They could also fix the UAE version for OS4 in a way that it doesn’t need to ”emulate” the 68k instructions on a 68k cpu...

What a pci version of Vamp concerns, ehum... I have a mediator.
Not sure if it would be as good or even better to put a vampire in a pci slot, than onto the expansion port of the A1200, or CPU port of other Amigas.

I know the Mediator was able to increse in speed if you had a SharkPPC card in one pci slot, but those were never released. I wonder if not the Vampire could make something like that happen as well. Imo, that would be awesome!


Harold Joseph Neufeldt

Posts 5
17 Feb 2018 23:12


You know this idea of yours is older than the hills. My business partner and I had a shop which in the late nineties repaired Amigas and Ataris, swell as PCs. One idea that was quite prevalent in those days was to have an Amiga card placed in the PC.This was used to emulate the Amiga on the PC.  It turned out not to be so good an idea. On the better cards, performance was no better than good emulation software. On most cards however, they were nothing but a headache.  You do not want to know what happened to the performance of a PC when some one placed an Amiga card with a 7Mhz. 68000 in a slot with a 66Mhz bus.

Going to the topic at hand let me say this.  The SAM boards,the AmigaOne, and Pegasus are essentially obsolete hardware and will not run 4.1FE software. They will however run 4.1 Classic software. I do not believe Tabor will run theNGFS which is now shipped with 4.1FE  That most likely will use SFS instead. I do not know if the version of NGFS shipped with new X5000s will even run on X1000s.  I know that Tony Wyat developed NGFS on a 460. He said at the time though that the minimum needed to run shipping versions would require at least an X1000. It has been upgraded since I believe.

Back porting OS 4.1 has other pitfalls as well.  Some parts of the Boot sequence are written inPPC assembler.  Much of the software that has been written for 4.1FE will be in much in the same boat. That is that it will require a fairly fast processor torun.

Reading over all the posts, I see only one that makes any sense. Sam Crow's proposal to make a QEmu backend  for the 68080 is by far the best solution. Mind you it will only run4.1 Classic software. Thus it will not run most of the Enhancer software, or much of the new software that is becoming available for 4.1FE.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
18 Feb 2018 07:34


Harold Joseph Neufeldt wrote:

  Back porting OS 4.1 has other pitfalls as well.
  Some parts of the Boot sequence are written in PPC assembler.
 

This should be no hold up.
Many members of the APOLLO Team are IBM developers which did design and develop Power Chips and which are PowerPC assembler experts.
So helping to write some PowerPC ASM code to 68K code is not a problem.
 
 
 
Harold Joseph Neufeldt wrote:

Sam Crow's proposal to make a QEmu backend  for the 68080 is by far the best solution.

Writing a PPC emulator for 68k is actually a funny idea.
The 68080 is a superset of the PowerPC.
All Address modes, all Operations that the PowerPC can do are directly supported on the 68080. This means PPC assembly can be translated 1:1.
 


Mr Niding

Posts 459
18 Feb 2018 09:20


Isnt the main positive with Vampire/Apollo the relative simplicity of the design?
If you read different forums, you see complains about driver support, and compability issues having to be resolved.
Its a nightmare for X developer to find workarounds to acommodate Y developers hardware.

With Vampire, you have a very SET hardware platform. The core is being adjusted/improved, but you have a very transparent and constant hardware config to develop for.

Adding layers of complexity seems counterproductive, and will in the end draw developertime away from the core, and towards accomodating for XYZ 3rd party hardware.

I totally understand the desire to do so from a user point of view, but the end result looks to me to be a developer nightmare.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
18 Feb 2018 10:38


Mr Niding wrote:

Isnt the main positive with Vampire/Apollo the relative simplicity of the design?

To precise this, the APOLLO CORE is internally more complex, more modern and much more advanced than most PowerPC cores.


Mr Niding

Posts 459
18 Feb 2018 11:33


Gunnar von Boehn wrote:

Mr Niding wrote:

  Isnt the main positive with Vampire/Apollo the relative simplicity of the design?
 

  To precise this, the APOLLO CORE is internally more complex, more modern and much more advanced than most PowerPC cores.

Indeed. Just pointing out the "simple" design framework which doesnt invite too many 3rd party addons which results in hardware conflicts/driver development hell, which AOS4 hardware seems to suffer.


Mallagan Bellator

Posts 393
18 Feb 2018 17:49


Gunnar von Boehn wrote:

  Writing a PPC emulator for 68k is actually a funny idea.
  The 68080 is a superset of the PowerPC.
  All Address modes, all Operations that the PowerPC can do are directly supported on the 68080. This means PPC assembly can be translated 1:1.
 

With this in mind I’m only more convinced than ever that it would be best to just port any PPC applications and/or games to the 080 core instead.

Emulation would be a silly thing to do in this case


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
18 Feb 2018 18:47


Mallagan Bellator wrote:

Emulation would be a silly thing to do in this case

How about running MAC OS 9?


Mallagan Bellator

Posts 393
20 Feb 2018 08:49


Gunnar von Boehn wrote:

Mallagan Bellator wrote:

  Emulation would be a silly thing to do in this case
 

  How about running MAC OS 9?

I’ve never really been much into Mac, so I don’t know if it runs on 68K or PPC, but emulating Mac OS of any kind is ofcourse cool if you can. There are quite some games that I would love to play on Amiga, that exists on Mac, like Diablo and Diablo II for example. However, if those were ported to Amiga instead, to 080, that would ofcourse be alot better. Everyone knows that emulating a CPU takes more time from your CPU than the original CPU would need to perform the tasks needed. I know you said the Apollo core has benefits that the PPC has, but the CPUs are still different. Most PPC CPUs run at about 200 Mhz at least, and Apollo core at about 80 Mhz, right? Yes, it has improved instructions, and is faster than the 060 by quite a bit, and you shouldn’t stare yourself blind on the Mhz value. This is all true, but 80 Mhz vs 200 is quite a difference, even in this regard. Don’t get me wrong, I love the Apollo core and the whole idea behind it, but making an emulator for an 80Mhz PCU to emulate a 200+Mhz CPU, how much fun could you really have with it? Would it really be useful for ”performance”? If your answer to this is positive feedback, and you think it would really be worth your or someone elses time, then sure

I’m just thinking realism here.
Not to mention, 68K is, as we all know, CISC, while PPC is RISC. That just HAS to make things even more complicated. If the Apollo core would be ASIC today and run in Ghz, I’d say sure, that’s a great idea. And I believe it is a great idea... for a future project. I believe in the core!


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
20 Feb 2018 09:00


Mallagan Bellator wrote:

  There are quite some games that I would love to play on Amiga, that exists on Mac, like Diablo and Diablo II for example. However, if those were ported to Amiga instead, to 080, that would of course be alot better.
 

Porting games would be nice.
But of these games the sources are not public.
So porting them is not possible.
Also if you port 1 game then you have 1 game, but if you write an Emulator like for example the NEO-GEO emulator then you get 100,200, or 300 games at once.

Regarding Emulation speed.
The 68080 is in many programs / benchmarks / application performing a lot better than 200/300 Mhz PPC.

Of course emulation will cost something but with a good JIT compiler reaching a reasonable speed should be possible.

You have to mind that the "low-end" PPC MAC started with 60MHz - their speed should be not a problem to reach. And in the future like you mentioned there will be faster 68080 versions.


Mallagan Bellator

Posts 393
21 Feb 2018 19:57


I had no idea the low ends started at 60Mhz, that’s cool. Well, if it can be optimized to a reasonable speed, of course. I know you’re right about the amount of games that would be available through emulation, of course. And the 080 reaching the performance level of a 300Mhz PPC is cool, but at emulating the PPC? I mean, if you can make an emulator to run Mac OS 9, or what version you said, as if it was really running on a real 300Mhz PPC, let me be the first to know, and I’ll gladly be the first one to take my hat of for you.

I mean, it’s obviously already off, for all the stuff that all of you guys in the team have done, but you know what I mean.

Looking at the UAE emulator running turrican (yes, they are different processors, but still)... you know the rest already, I don’t need to say it. You practically already said it yourself


Harold Joseph Neufeldt

Posts 5
18 Mar 2018 22:47


Hi Bjorn

AT this point ownership pf OS 3.1 is up in the air. Up to this point it has been assumed that Amiga Inc. owns Amiga OS with Hyperion having effective control of the OS because of the agreement with Amiga Inc. Cloanto however, is challenging this state of affairs.  They could do this in part because Amiga Inc. failed to defend their trademarks earlier last year. There is now a lawsuit going on. It appears that this case is actually going to trial.    This is very unusual in these types of cases.  It means that Cloanto thinks that they will wind up with all the marbles.  When you consider that Cloanto has pretty well cleaned up in the trademark wars in North America Hyperion is pretty much of a wounded duck, whose end is near.  There is more info on AmigaWorld.  Number6 is posting info as he gets it.  It took me quite a while to get some relevant info together. I am pretty sure this case will go to trial unless Cloanto blinks, which seems highly unlikely. I should also point out that at this time, there is no point in speculating what this means for the Amiga community. That is what seems to be the case of OS3.1 at this time

Harold


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
18 Mar 2018 23:16


Gunnar von Boehn wrote:

    How about running MAC OS 9?
   

   
    That would enable few more titles unavail to 68k MacOS and would be great, as in fact it is "top of the line MacOS Classic" (whose clean design I prefer to MacOS X candy look).
 
  Example is "ClassZila" ported Gecko to MacOS 9
  EXTERNAL LINK 
  Which would solve the browser dilemma?
 
 
Harold Joseph Neufeldt wrote:

  I am pretty sure this case will go to trial unless Cloanto blinks, which seems highly unlikely. I should also point out that at this time, there is no point in speculating what this means for the Amiga community. That is what seems to be the case of OS3.1 at this time
    Harold
 

 
  Its relevant to announced Hyperions OS 3.1+ (dont know how to call it as OS 3.2 existed in beta for Boxer/Walker) but even if that does not happen, wont be a big thing. OS4 is much someone elses work, and I doubt OS will reach pre-purchased OS 4.2 with my x1000. Tower57 is great game, and LibreOffice would aid its usability, but then again it might end as good as Timberwolf. Not much faith there. Chance of AmigaOS 4.1.8 for 68k is even thinner.
 
  However, whatever happens there, it does not affect Coffin / EmuTOS / AROS 68k and other Vampire related efforts. But is another real nail in coffin in Amiga story.

posts 36page  1 2