Natami

[amigadave]

@AJCopeland

If you get a working OpenGL, then you'd probably want to release a version of MiniGL for that. Not that many apps use Warp3D directly since it is a low level API, they tend to be GL based apps linked against minigl. They'd still require recompiling to use it though.

I don't think that backward compatibility with old Warp3D games should be considered at this time, if ever.  The Natami team has never claimed that it would be 100% compatible with all old 68K games and applications, as very few true Classic Amigas are 100% compatible with 100% of those old apps and games either.

Natami is shooting for as much compatibility with older 68k apps and games as possible, but the focus is and should be on future apps and games that have better performance due the the increased speed and resolution available on the Natami and ONLY on the Natami.

I would expect and hope for Natami specific, or "native" games and applications to be written by a few programmers that still appreciate what it is like to use an Amiga.  I am assuming that Natami will run AmigaOS3.9 from the start and that over time we will see more enhancements to the OS itself, like a Boing Bag 3, or an unofficial AmigaOS3.91, 3.92, 3.93, etc.

Natami native games and applications are no different that writing something only for the CSPPC/BPPC, or for a certain 680x0 CPU, or the difference between OCS, ECS and AGA.  Now (when Natami is finished) we will have a fourth option to write to that is SuperAGA and can be more demanding than what was possible on an 68060/50MHz.

Of course, programmers that want their work to be appreciated by a wider audience will make two versions of graphics, AGA and SuperAGA and will write the games or apps to be able to run sufficiently well on an 68020/14MHz or better, or if it is a more demanding app or game, maybe the minimum spec Amiga could be a 68030/40MHz, which would cover a lot of the accelerators out there.

My point is that it will be nice to have this faster than 68060/50MHz, SuperAGA Amiga compatible computer that some programmers can take advantage of with new code.

I guess someone could redo the graphics on some of the old Amiga 68k games to use SuperAGA resolutions, but I would hope for some original works, instead of just prettying up some of the old games and applications.

[klx300r]

great news! I hope it becomes reality :-)

[Karlos]

@amigadave

If you look at what you have just written, if that is the case, then why make a SuperAGA at all? The only reason to have AGA is for compatibility and much is made of SuperAGA's backwards compatibility. Compatibility is therefore an important consideration. I think it would be foolish not to consider compatibility with the broadest spectrum of software for the sake of not considering existing APIs.

One of the biggest criticisms levelled at the "next generation" machines initially was the fact there was little native software and no compatibility with legacy software that hit the metal.

I would suggest that without exposing the NatAmi SuperAGA features via existing APIs, you'd run into part of the same problem - new hardware with nothing that really takes advantage of it.

[amigadave]

@amigadave

If you look at what you have just written, if that is the case, then why make a SuperAGA at all? The only reason to have AGA is for compatibility and much is made of SuperAGA's backwards compatibility. Compatibility is therefore an important consideration. I think it would be foolish not to consider compatibility with the broadest spectrum of software for the sake of not considering existing APIs.

One of the biggest criticisms levelled at the "next generation" machines initially was the fact there was little native software and no compatibility with legacy software that hit the metal.

I would suggest that without exposing the NatAmi SuperAGA features via existing APIs, you'd run into part of the same problem - new hardware with nothing that really takes advantage of it.

I guess I will have to read what I wrote again, but I certainly did not mean to imply that the greatest amount of compatibility should be implemented.  But I think it was you, or someone else in this very thread who pointed out how little existing software was written to take advantage of Warp3D, or any kind of 3D on the Amiga in the past.

My point was that if that is truly the case, then creating a new type of 3D for this "SuperAGA" PLUS retaining 80% to 90% compatibility with the thousands of existing Amiga software titles would not be a bad thing.  If they can somehow create a better 3D system AND still keep compatibility with Warp3D, then so much the better.  Again, my only point is that if some percentage of older Amiga software is no longer compatible with the Natami, it will be the price we may need to pay for progressing to a better, faster, more colorful and higher resolution Amiga compatible system.

It is my understanding that the Natami team is doing everything they can to retain as much compatibility with all of the existing Amiga models and software as possible.  

Edit:  The reason for having SuperAGA is to provide a step forward that will allow more colors and resolution for future programs.  A step forward that many Amiga users would have liked Commodore to make back in 1990 or some time close to that.

[Karlos]

Edit: The reason for having SuperAGA is to provide a step forward that will allow more colors and resolution for future programs. A step forward that many Amiga users would have liked Commodore to make back in 1990 or some time close to that.

This is exactly the problem that RTG already solved in 1993.

I'm not saying don't expose the hardware to metal bashers who want to do so for fun, but for OS clean stuff, using RTG seems logical. Pick either P96 or Cybergraphics and make a clone API that runs on SAGA.

Then you get to run all the RTG compatible software. The same applies to AHI and SAGA's improved Paula.

[bloodline]

In essence, Natami is a solution looking for a problem...

[Flashlab]

In essence, Natami is a solution looking for a problem...

Now that you found it, it succeeded?

[kolla]

This is exactly the problem that RTG already solved in 1993.

But RTG never solved the problem, it just worked its way around it. RTG never _replaced_ native displays, it just added more on the gfx board, never would the gfx board run the native displays. With SuperVGA however, you do replace native displays, and the new ones you add also run on the very same chipset. RTG is pointless since there is just one chipset to worry about here, unlike gfx cards of which there are so many.

I'm not saying don't expose the hardware to metal bashers who want to do so for fun, but for OS clean stuff, using RTG seems logical. Pick either P96 or Cybergraphics and make a clone API that runs on SAGA.

But why bother with CGfx and P96? Here you at last have capable native Amiga hardware, why would you want to toss a layer of drivers systems with dubious licensing issues in between?

Then you get to run all the RTG compatible software. The same applies to AHI and SAGA's improved Paula.

Geh - no! This is exactly the opposite of what this project is about :laughing:

There is no 3D engine for AmigaOS that runs on native chipset AFAIK, Warp3D at al all require RTG in form of CGFx or P96. On the other hand I dont know of a single game or piece of software that uses Warp3D and that is really worth the extra effort it would require to be compatible. IMO, Natami does the right thing when ignoring whatever happened after 1993

[Hammer]

Dude, maybe you did not notice.                  This is an AMIGA forum.

AMIGA is not bleeding edge since about 15 years anymore.
Nothing which came out since then for AMIGA was bleeding edge.
SAM is not bleeding edge, A1 was not bleeding edge and NATAMI is also not bleeding edge.

If you want bleeding edge then you need to go to forum about Cloud computing on CELL Chip,

AROS X86 (and ATI/NVIDIA GpGPU(1)) runs on "bleeding edge" hardware. STI CELL is not the only stream processor with a large GFLOP numbers. SAM can be link with current gen ATI Radeon HD 4350 i.e. go with Xbox360 route(2) for heavy GFLOPs.

Without factoring VMX128, CPU clock speed and L2 cache, PPE is almost the same level as PPC440. The problem with AOS is middleware support. The cost of SAMFlex doesn't benefit bang-for-buck scenarios.

1. Basic support only.
2. Majority of streaming GFLOPs provided by GpGPU.

[Fats]

No, it has an FPGA which is emulating the Amiga chipset. As for "Amiga Spirit"... Sorry but that is something entirely intangible and means different things to different people.

For it to be considered "real hardware" it would have to be an ASIC.

Remember then that a big part of the telecom infrastructure, including internet is in your definition not running on "real hardware" but just FPGAs.

greets,
Staf.

[the_leander]

But RTG never solved the problem, it just worked its way around it. RTG never _replaced_ native displays, it just added more on the gfx board, never would the gfx board run the native displays. With SuperVGA however, you do replace native displays, and the new ones you add also run on the very same chipset.

And how much current software is going to be able to take advantage of the new screenmodes?

 RTG is pointless since there is just one chipset to worry about here, unlike gfx cards of which there are so many.

Now, I may be wrong on this, but I thought the whole point of using something like cgfx or p96 was to avoid this very issue - you write to allow the use of APIs without having to worry about the underlying chipset. Thus, when you upgrade you get more features at a much lower cost in terms of hardware development.

Indeed many of the problems with the Amiga and not being able to shed the "classic" chipset is down to people writing direct to the hardware.

But why bother with CGfx and P96? Here you at last have capable native Amiga hardware, why would you want to toss a layer of drivers systems with dubious licensing issues in between?

In a word? Compatibility.

There is no 3D engine for AmigaOS that runs on native chipset AFAIK

Quake runs on AGA, poorly, but it runs.

, Warp3D at al all require RTG in form of CGFx or P96. On the other hand I dont know of a single game or piece of software that uses Warp3D and that is really worth the extra effort it would require to be compatible.

Question is, is it worth the effort of learning to bang the hardware (again) when there are APIs that are current on the Amiga and already have software to use on them.

 IMO, Natami does the right thing when ignoring whatever happened after 1993

I have to be honest, the way you're describing it... Certainly from a user's perspective you just made the whole deal that much less appealing. The idea of purposefully disallowing a huge amount of software (or at the very least, hobbling it) does seem very counter-productive. It's like taking all the shortcomings of Amithlon but with much less of the benefit.

[the_leander]

Remember then that a big part of the telecom infrastructure, including internet is in your definition not running on "real hardware" but just FPGAs.

Citation?

[Hammer]

Remember then that a big part of the telecom infrastructure, including internet is in your definition not running on "real hardware" but just FPGAs.

greets,
Staf.

Modern X86 CPUs doesn't natively execute X86 ISA i.e. converts/translate/decode from X86 CISC to a custom RISC like ISA. Modern X86 decoders has can be reprogrammed e.g. micro-code updates (hardware bug fixes).

[the_leander]

Modern X86 CPUs doesn't natively execute X86 ISA i.e. converts/translate/decode from X86 CISC to a custom RISC like ISA. Modern X86 decoders has can be reprogrammed e.g. micro-code updates (hardware bug fixes).

Was it you that said something like 20% of the logic on a modern x86 is dedicated to this task?

[Hammer]

Was it you that said something like 20% of the logic on a modern x86 is dedicated to this task?

In the early RISCy X86 era, it takes a large chunk of silicon. These days, X86 CISC tax is insignificant i.e. X86 ISA didn't change much.

[kolla]

And how much current software is going to be able to take advantage of the new screenmodes?

Depends how it's written. Some software will happily gobble up the extra chipram and faster graphics, for example DPaint might be able to happily animate fast in 1080x720 HAM8 modes. How does that work on UAE?

Now, I may be wrong on this, but I thought the whole point of using something like cgfx or p96 was to avoid this very issue - you write to allow the use of APIs without having to worry about the underlying chipset. Thus, when you upgrade you get more features at a much lower cost in terms of hardware development.

Correct. But you seem to forget what Natami is about. Doesnt it strike you somewhat strange to argue about chipsets when the Natami will use a 68k processor? It's very much a souped up Super MiniMig, not a SAM or Pegasos replacement, for crying out loud.

Indeed many of the problems with the Amiga and not being able to shed the "classic" chipset is down to people writing direct to the hardware.

And that is exactly what Natami is about, letting people have some hardware to write directly to again

Quake runs on AGA, poorly, but it runs.

Is that really the best you can think of? :laughing:

Question is, is it worth the effort of learning to bang the hardware (again) when there are APIs that are current on the Amiga and already have software to use on them.

Again, you seem to have forgotten (or never grasped) what Natami is about. It's for those who already know how to bang hardware, and would like to keep banging hardware, and for those of us who can enjoy what comes out of it all, and can afford it.

The rest of you should rather look for better hardware for OS4.x and MorphOS instead of wasting time on this useless debate

I have to be honest, the way you're describing it... Certainly from a user's perspective you just made the whole deal that much less appealing. The idea of purposefully disallowing a huge amount of software (or at the very least, hobbling it) does seem very counter-productive. It's like taking all the shortcomings of Amithlon but with much less of the benefit.

Correct - you're learning now