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
The team will post updates and news about our project here

Software for Vampirepage  1 2 3 4 

Barana V
(Needs Verification)
Posts 59/ 1
02 Jun 2019 06:56


Wow. It looks like my observations were correct. When you pick someone up for their attitude imo, they can't help but answer you in the same spirit they have towards the first target. My goal here is to not enflame, just express elation that my observations proved true. I have no further insights appart from those unheeded insights I and others have given on this thread.and no further observations to add that are beneficial to the communitiy at this time.
Anyway back to the topic at hand.
  Peace.


Mike Kopack

Posts 268
02 Jun 2019 19:07


So a quick google search came up with this:
 
  EXTERNAL LINK 
  From what I can tell, you need to make some config files that describe the instructions of the new architecture you want to support, add in a compiler flag to specify that architecture as the target, and a few other things.
 
  Now obviously this is going to require somebody who has knowledge of the 080 core (so probably somebody from the Apollo core team), somebody with some ASM experience, and GCC knowledge. On the plus side, there's already the GCC toolchain for cross compiling Amiga stuff:
 
  EXTERNAL LINK 
  So it should just be additions on top of that work to add in '080 core and AMMX support (ie, we should be able to look at what's already in there for adding in '020-'060 support as guidance for how to add in the '080 stuff...).
 
  Not saying I'm expecting the Apollo core guys to do this work, but at the present time they probably have the most in-depth knowledge of the core to be able to do it.

Putting that support into the older amiga compilers (like SAS/C) is going to be MUCH tougher without the source code to those tools to add it in. To be honest, though, these days it makes a lot more sense to do your development on a modern x86 machine and cross-compile for the Amiga and then test on either UAE or moving it to the real hardware - less dealing with the OS crashing and wiping out your work, and much faster compile times, modern tools, etc.
 
  That all said, I'd LOVE to see a full rebuild of the OS to take advantage of the Vampire across the board. I'm betting we'd see some significant performance improvements. While just doing specific software would help, having the kernel built for the CPU will provide even more performance.

And by the way, I HAVE written a compiler (from scratch) in both my undergrad and Masters Degree CS programs, although we didn't do gcc, so it's a totally different beast. Heck, in my undergrad we didn't even get to use tools like Lexx and YACC/Bison - we had to write all that stuff by hand!


Steve Ferrell

Posts 424
03 Jun 2019 01:00


Mike Kopack wrote:

So a quick google search came up with this:
 
  EXTERNAL LINK   
  From what I can tell, you need to make some config files that describe the instructions of the new architecture you want to support, add in a compiler flag to specify that architecture as the target, and a few other things.
 
  Now obviously this is going to require somebody who has knowledge of the 080 core (so probably somebody from the Apollo core team), somebody with some ASM experience, and GCC knowledge. On the plus side, there's already the GCC toolchain for cross compiling Amiga stuff:
 
  EXTERNAL LINK   
  So it should just be additions on top of that work to add in '080 core and AMMX support (ie, we should be able to look at what's already in there for adding in '020-'060 support as guidance for how to add in the '080 stuff...).
 
  Not saying I'm expecting the Apollo core guys to do this work, but at the present time they probably have the most in-depth knowledge of the core to be able to do it.
 
  Putting that support into the older amiga compilers (like SAS/C) is going to be MUCH tougher without the source code to those tools to add it in. To be honest, though, these days it makes a lot more sense to do your development on a modern x86 machine and cross-compile for the Amiga and then test on either UAE or moving it to the real hardware - less dealing with the OS crashing and wiping out your work, and much faster compile times, modern tools, etc.
 
  That all said, I'd LOVE to see a full rebuild of the OS to take advantage of the Vampire across the board. I'm betting we'd see some significant performance improvements. While just doing specific software would help, having the kernel built for the CPU will provide even more performance.
 
  And by the way, I HAVE written a compiler (from scratch) in both my undergrad and Masters Degree CS programs, although we didn't do gcc, so it's a totally different beast. Heck, in my undergrad we didn't even get to use tools like Lexx and YACC/Bison - we had to write all that stuff by hand!

Yes, the Apollo Team will have to provide guidance at a minimum and they my have to do a lot of in-house development because much of the 68080 is proprietary and Gunnar will most likely want to protect his  intellectual property as much as possible.  I.E Gunnar has not released any info or documentation regarding the 68080's MMU, and that's his prerogative.

You're arriving at the same conclusions I have.  GCC seems to be the best candidate.  It's open source, well documented and already provides support up to the 68060 so the effort required to update it for the 080 and MMX should be minimal, at least in theory.

The two most popular compilers for the Amiga back during the heydays were Lattice/SAS C and Aztec C, but I'm betting it will be impossible to obtain the sources for those compilers.

Another good candidate is VBCC: EXTERNAL LINK 
Jens Schoenfeld became a sponsor of VBCC a couple years ago:  EXTERNAL LINK


Barana V
(Needs Verification)
Posts 59/ 1
03 Jun 2019 01:48


Now we're talking, iirc not that much can be done about it, but aos was originally written with the Greenhills compiler, and once a copy was located, Olaf Barthel and crew used it to produce 3.1.4.
Apparently the 3.1 source used headers and such was a little specific to the Greenhills compiler.
  I was waiting for someone to mention vbcc after all that talk of 'theres no Amiga C compiler to the job'.
Vbcc is interesting as it uses vasm  in its toolchain
Vasm has ammx instructions dded  into it of late.
So no it isn't OSS but freeware.....


Steve Ferrell

Posts 424
03 Jun 2019 21:25


Mike Kopack wrote:

 
  That all said, I'd LOVE to see a full rebuild of the OS to take advantage of the Vampire across the board. I'm betting we'd see some significant performance improvements. While just doing specific software would help, having the kernel built for the CPU will provide even more performance.
   
 
 

 
It's probably more realistic to hope for AROS 68K  to be recompiled for 68080.  I doubt any version of OS 3.x.x will ever be free of the legal entanglements. The Vampire now boots AROS and it's running at a very acceptable speed even without being able to take full advantage of the 080 enhancements.
 
Pedro Cotter even has some videos of AROS 68K in action on the Vampire here: EXTERNAL LINK   
and here: EXTERNAL LINK   
and Simo has a video here:
EXTERNAL LINK   
It's hard to believe that time has moved so quickly.  These videos were made 9 months ago!
 
 
 

posts 65page  1 2 3 4