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!

Atari ST On Vampire 600 V2page  1 2 

OneSTone O2o

Posts 159
05 Apr 2018 18:49


Interesting would also be to get this emulation running on vampirized ST... - but only as much as it can run 68000 software at 8 Mhz on original ST/STE speed.
 
You should add VT-x commands to Apollo core for virtual machines... (yeah, like newer Intel CPUs do)

And then, porting UAE to vampirized Amiga & ST as well, but also just as much as it behaves exactly like a stock A500.


Eric Gus

Posts 477
05 Apr 2018 19:44


oneSTone o2o wrote:

 
  And then, porting UAE to vampirized Amiga & ST as well, but also just as much as it behaves exactly like a stock A500.
 

 
  I dont know about ST but UAE for amiga 68k already exists .. (I dont think its optimized for vampire though)

  EXTERNAL LINK 
  VT-x would be nice (and I fully agree with you) but at this point Id hope they dont change the core too much and "feature lock" it so developers can start working in earnest without the base cpu changing on them.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
05 Apr 2018 20:03


eric gus wrote:

Id hope they dont change the core too much and "feature lock" it so developers can start working in earnest without the base cpu changing on them.

This is already the case, actually was the case also before.
The core does is not changing.


Marlon Beijer

Posts 182
06 Apr 2018 10:40


eric gus wrote:
Id hope they dont change the core too much and "feature lock" it so developers can start working in earnest without the base cpu changing on them.

Nothing gets removed, only new things are added. There's nothing preventing developers from doing any work for the Vampire. Hell, I ported over Milkytracker which requires FPU, because of the way it's programmed, before the FPU was even mentioned for Vampire v2. And now it's using AMMX optimized instructions to gain performance, besides the FPU.


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
06 Apr 2018 13:29


eric gus wrote:

  VT-x would be nice (and I fully agree with you) but at this point Id hope they dont change the core too much and "feature lock" it so developers can start working in earnest without the base cpu changing on them.

At present state, ApolloCore is quite done. AMMX instructions are standardized, as well as 080 ones, FPU will improve with v4 further. Rest is pure adding cache, higher frequency and supporting expansions, USB, LAN, nothing that should stand in coders ways. It might get a bit better explained and more coding tools added I suppose.

Adding virtualization would be great for all emulators and virtual machines.



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
06 Apr 2018 13:43


Vojin Vidanovic wrote:

At present state, ApolloCore is quite done. AMMX instructions are standardized, as well as 080 ones, FPU will improve with v4 further.

 
Lets precise this.
 
* APOLLO 68080 supports all instruction of 68000 and 68040 and 68060.
This means normal 68K code will run good on it.
 
* APOLLO 68080 is Super-Scalar.
This means with "normal" written Super-Scalar code performance can be over 150 MIPS.
 
* APOLLO 68080 does supports ALL FPU instructions of 68882/68040/68060
This means you can just write normal FPU code.
No new instructions are needed to be learned.
 
* APOLLO 68080 does support AMMX2 which is a very powerful SIMD enhancement.
AMMX2 instructions are very powerful and flexible and are well documented. All AMMX2 instructions will also be supported in the future.
 
 
The APOLLO CPU architecture is very stable.
  2-4 instructions can be executed per cycle peak.
  16 byte decoding per cycle peak.
  8 byte memory read per cycle peak.
  8 byte memory write per cycle peak.
  The general behavior of the CPU is unchanged since day one.
 
 
If you wrote good code for APOLLO a year ago,
then this code will run good today too,
and will also run good in the future.
 
  APOLLO is absolutely not a changing target.
 
 


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
06 Apr 2018 16:03


Gunnar von Boehn wrote:
 
    Lets precise this.
  ...
  APOLLO is absolutely not a changing target. 
 

 
  Thanks for de-classifying on of the most "hailed" anti-arguments.
 
  To the best of my not-coders knowledge this means:
 
  a) Generic 68k code will work
  b) 68k Amiga code will work unless it uses MMU or AGA (whereby last will change with core3)
  c) Optimized 080-AMMX approach needs using VASM or Linux cross compiling, where new 080 instructions, AMMX can be used to reach previously impossible on 060.
 
  What about v4 full FPU?
 
  I assume all past FPU 68k code will work, but similarly, some tricks or newer compiler will be needed to use new FPU and precision to max.
 
 
Marlon Beijer wrote:

  Nothing gets removed, only new things are added. There's nothing preventing developers from doing any work for the Vampire. Hell, I ported over Milkytracker  ... And now it's using AMMX optimized instructions to gain performance, besides the FPU.
 

 
  Thanks for your example Marlon.
 
  I assume v1.01 or v1.02 supports MMX? Cant see it in commits.
 
  Milky Tracker GitHub Amiga Releases Page
  EXTERNAL LINK   


Bart Mikulski

Posts 101
09 May 2018 22:30


How do I run HDD format games in STonAmiga?
 
  I tried a few, whenever I run RUNME.TOS or another *.TOS, it crashes. I haven't managed to run a single game this way.
 
  I have tried a few in PRG format, they work.

EDIT: OK, I managed to run 2 out of 10 games in HDD format through *.TOS files.


Eric Gus

Posts 477
10 May 2018 07:36


Has anyone figured out how to "insert" a floppy disk image into STonAmiga ?


Bart Mikulski

Posts 101
12 May 2018 14:00


eric gus wrote:

Has anyone figured out how to "insert" a floppy disk image into STonAmiga ?

No, I haven't.

What is your games loading success ratio?

Are you experiencing the problems like mine here:

CLICK HERE 
?

Maybe you could upload a package of games that you have no problem loading, I would love to check them on my system...




Marlon Beijer

Posts 182
12 May 2018 15:04


Vojin Vidanovic wrote:

Gunnar von Boehn wrote:
 
    Lets precise this.
    ...
    APOLLO is absolutely not a changing target. 
   

   
    Thanks for de-classifying on of the most "hailed" anti-arguments.
   
    To the best of my not-coders knowledge this means:
   
    a) Generic 68k code will work
    b) 68k Amiga code will work unless it uses MMU or AGA (whereby last will change with core3)
    c) Optimized 080-AMMX approach needs using VASM or Linux cross compiling, where new 080 instructions, AMMX can be used to reach previously impossible on 060.
   
    What about v4 full FPU?
   
    I assume all past FPU 68k code will work, but similarly, some tricks or newer compiler will be needed to use new FPU and precision to max.
 
 
Marlon Beijer wrote:

    Nothing gets removed, only new things are added. There's nothing preventing developers from doing any work for the Vampire. Hell, I ported over Milkytracker  ... And now it's using AMMX optimized instructions to gain performance, besides the FPU.
 

 
  Thanks for your example Marlon.
 
  I assume v1.01 or v1.02 supports MMX? Cant see it in commits.
 
  Milky Tracker GitHub Amiga Releases Page
  EXTERNAL LINK   

Sorry for late reply. It's the version of SDL-1.2 that I use to compile Milkytracker with that has AMMX support. I am unsure if Bax has updated it to accommodate for AMMX2. For now AMMX1 will suffice. That version of SDL is available here: EXTERNAL LINK 
New binary releases of Milkytracker are put up here: EXTERNAL LINK


Eric Gus

Posts 477
13 May 2018 05:44


Bart Mikulski wrote:

 
  What is your games loading success ratio?
 
 

 
  Ive mostly just done demos and such .. no real games as of yet. And those demos have run ok.. no problems.

posts 32page  1 2