Apollo Maggie 3D Chip - Preview Demo

[Valken]

I LOVED a 3D accelerated gridder dungeon game so I paid attention to the demo video but modern 3D with shaders spoiled me on the PC.

It needs a killer app to make it wanted besides a glorified tech demo.

Port Dungeon Master or Eye of the Beholder or even Ultima Underworld to it.

[meynaf]

Originally Posted by Seiya

Gunnar said that he loves games, so vampire development is very much based on this aspect.
Amiga today is used only for: games.
The future of Amiga is: games

Amiga is retro gaming system and if there are so many beautiful applications, seriously things to do, for 99,9% it is used only for games.

For retro gaming we don't need more than what we already have.

[alexh]

Interesting development. Could be quite cool if it ever gets drivers for an existing Amiga 3D API. I applaud the developers. Must be quite satisfying to see your work on the screen.

As cool and interesting as this is. I imagine it's capabilities will be vastly inferior to the last generation of PCI GFX cards that are currently available for Amiga's with Mediator/Prometheus.

The big problem is the limited number of Amiga HDL developers. A new GPU design might not be the best idea. Perhaps adopting one of the multiple opensource GPU's and contributing to the AmigaOS drivers and general GPU architecture and bringing it to Apollo core might be the better way to go?

[eXeler0]

Originally Posted by alexh

Interesting development. Could be quite cool if it ever gets drivers for an existing Amiga 3D API. I applaud the developers. Must be quite satisfying to see your work on the screen.

As cool and interesting as this is. I imagine it's capabilities will be vastly inferior to the last generation of PCI GFX cards that are currently available for Amiga's with Mediator/Prometheus.

The big problem is the limited number of Amiga HDL developers. A new GPU design might not be the best idea. Perhaps adopting one of the multiple opensource GPU's and contributing to the AmigaOS drivers and general GPU architecture and bringing it to Apollo core might be the better way to go?

In theory this migh seem to make sense, but in reality, what gpu will fit into the Apollo V4 FPGA. The have a decent user base by now, it would make very little sense to start adding stuff that would render the entire platform obsolete. (by requireing a new bigger fpga)
I think we need to see it as an extension of the SAGA chipset tobgive it *some* sort of 3d capability, not as gluing a powerful GPU to an otherwise compareatevily weak retro platform. Probably more like the era when we had 3dNow, SSE3 instrad of full gpus.

[Valken]

I had another thought. Even it can accelerate games like 3D action such as Doom or Dukenukem (game engine level support), that would be pretty awesome on Amiga and even ST.

GZDoom (OpenGL and Vulkan 3D support) pretty much made Doom mods and new retro games on PC/Mac/Linux/Consoles a revival especially with cracy CPU/GPU prices these days.

[eXeler0]

Originally Posted by Valken

I had another thought. Even it can accelerate games like 3D action such as Doom or Dukenukem (game engine level support), that would be pretty awesome on Amiga and even ST.

GZDoom (OpenGL and Vulkan 3D support) pretty much made Doom mods and new retro games on PC/Mac/Linux/Consoles a revival especially with cracy CPU/GPU prices these days.

Original doom predates 3d hardware on PC so thats not gonna do much good, but a better Quake port is definitely a likely possibility.
The original Quake that was ported to Amiga might be optimized for fast CPUs but that version of the engine has so many limitations that its hard to make maps for it (Ive tried). A port of a newer engine would inevitabely require some sort of gfx accrleration and Maggie would have a purpose…

[Promilus]

@eXeler0 - you miss the point here. Most of ports uses either open source recreation of original engine (but still, in source form so available for modification) or already modified officially released source code. There's source code for Doom for quite a while (and so is for Quake), you can modify rendering engine whatever you wish. That's exactly why modifications like Dark Places for Quake has shaders while original game didn't. Also glQuake is modification itself. Originally quake didn't have support for hardware 3d acceleration and glQuake was official release (but not maintained by id) as a proving ground for Q2 hw accel embedded from a get go.

So while not groundbreaking technology by itself maggie just MIGHT be used by some ports should their sourcecode be modified to use those features (instead or beside software rendering routines). But - as always - it requires additional work by port developers. So if port uses Warp3D and Apollo Team writes a W3D driver or wrapper for Maggie it's just great. If port doesn't use Warp3D and someone writes subroutines using Maggie directly... that's also great - but for ApolloTeam card users only.

[Paulee_Alex_Bow]

Originally Posted by Seiya

Gunnar said that he loves games, so vampire development is very much based on this aspect.
Amiga today is used only for: games.
The future of Amiga is: games

Amiga is retro gaming system and if there are so many beautiful applications, seriously things to do, for 99,9% it is used only for games.

Not 99.99%, maybe 80%. There’s big groups of 2D/3D Amiga artists, musicians etc.

[saimon69]

Originally Posted by Paulee_Alex_Bow

Not 99.99%, maybe 80%. There’s big groups of 2D/3D Amiga artists, musicians etc.

That, however, (at least me) use their talent for, guess what?
Developing games ^^

[Bruce Abbott]

Originally Posted by alexh

As cool and interesting as this is. I imagine it's capabilities will be vastly inferior to the last generation of PCI GFX cards that are currently available for Amiga's with Mediator/Prometheus.

Since the vast majority of Amiga owners will never have such a setup anyway, this is not a big deal. But many people might buy one of the latest Apollo accelerators, and having faster 3D will be a nice bonus for them.

I am more concerned for those of us who can't justify the hassle and expense or don't want to further compromise the 'retro' status of our machines. Fancy 3D is all very well if that's all you want (but then why use an Amiga when a modern PC does it so much better without even trying?). But many of us spent a lot of money and some effort to make a 'high-end' Amiga with an 060 or Vampire, only to now find that new games produced for the Amiga need more. It shouldn't be necessary to keep paying and paying for ever more 'upgrades' - that's not the Amiga way!

Personally I am far more interested in what is being done at the lower end, with optimized coding that shows what our classic machines could have done 'back in the day' - stuff that you wouldn't believe was possible on a stock A500 or A1200, and would have blown our socks off if released in 1992. This is happening because some talented coders have revisited the classic Amiga and realized that it was capable of much more. But it needs a lot of work to get the best results. I worry that enormous processing power and fancy 3D may lure developers into only producing games for much higher spec 'Amigas' - for no real reason apart from that it's easier.

Someone mentioned 'Open Lara'. I would love to see the original Tomb Raider games ported to the Amiga, both to show that it could be done and because it would provide an excuse for me to play them again. The original DOS game ran fine on a Pentium 60 with software rendering. I got it going on a Vampire 600 V2 running under PC Task, and it worked perfectly apart from having a frame rate of about 3fps. I even tried it on my A1200 with 50MHz 030 and it worked fine on that too, but at ~0.5fps (probably would have gone faster with JIT, but my A1200 doesn't have enough memory for that). Emulation is probably making it at least 10 times slower than native code. Don't need fancy 3D hardware to port this game, just good coding!

The source code to Tomb Raider III was leaked a while back. Most of it is written in generic C, with a bit of x86 machine code for the low level rendering. Amazingly it also has some 68k asm in it! (commented out where it was translated to x86). But Open Lara doesn't stick to the original code, which will probably make it much harder to port and have worse performance on machines without 3D hardware (if even possible). It would be ironic if Tomb Raider was ported to 'the Amiga' but still required a machine that was out of the reach of most Amiga owners.

[Seiya]

i think exactly the same.
So many new games are created/converted in the latest year showing that good codes can be difference and that they run on a standard A500 or standard A1200.
GPU on Amiga is always a fascinating idea ok, but for old application that already exist or to make some benchmarks. Stop.

I think that today it is impossible to think to play a game like Open Lara on Amiga also with 68080 and a 3D Chip. Better to developing/converting games that could give satisfaction for all community and not for a little sub niche.

[nonarkitten]

Originally Posted by Promilus

Last pentium (p54c) was 233MHz which should give around 400MIPS when AC68080 has ~340.

LOL, the Apollo Core does NOT get 340 MIPS. Even by their own ridiculous graph, they only get a bit better than 141 MIPS.


The x13 and higher multipliers have never seen the light of day to the public and even at the virtually unattainable x16, it barely topped 200.

And these are SysInfo MIPS running on a version of SysInfo specifically tweaked to gain advantage on the Vampire while not leveraging any advantage of the 68060's slightly different pipeline behaviour. SysInfo MIPS are about 50% higher than real MIPS on high-end CPUs.

If we ballpark and say that's ~100 Dhrystone MIPS then the AC68080 is around a Pentium 75.

[pandy71]

Curious how Apollo Maggie 3D core compare to RPi BCM VideoCore and perhaps PowerVR SGX544 present in AM5729.
Of course i share same impression as you guys that it is slightly too late for modern 3D on Amiga but still if 3D will be popularized trough some simple standard (perhaps unique for Amiga due memory and CPU constraints) API it could be beneficial in overall.

[alexh]

Originally Posted by eXeler0

In theory this migh seem to make sense, but in reality, what gpu will fit into the Apollo V4 FPGA. I think we need to see it as an extension of the SAGA chipset to give it *some* sort of 3d capability, not as gluing a powerful GPU to an otherwise comparatively weak retro platform.

Open-source GPUs are in their infancy. DirectX 5/6 level. They are fun projects being run by enthusiasts. They appear to be today at the level of the earliest GPUs albeit with the hindsight of 20 years of GPU pipeline development to refer to. The developers will be using FPGAs approx the same complexity as Apollo V4. It depends how much of the V4 is left after the Apollo core is in.

Originally Posted by eXeler0

Probably more like the era when we had 3dNow, SSE3 instead of full gpus.

Maybe but I don't think so. You are going to want a rasterising HW pipeline that can become compatible with the 3D engine pipelines such as OpenGL, DirectX, Vulcan etc.

[NovaCoder]

Even if the vampire FPGA computer runs at a hundreds of MIPS and they somehow manage to support an old OpenGL standard using this new graphics chip it would only allow them to run old PC games like quake 2.

A raspberry pi (or a modern PC) will be able to give you a better OpenGL retro gaming experience at much (much) less cost.

And don't say ' but with the vampire FPGA 3D computer it will be running on an Amiga' because that statement is obviously ridiculous.

[Promilus]

@NovaCoder - yes, but we do own classic amiga and buy expensive additions to it not because our motivation is rational one. It's emotional one

[jbenam]

Originally Posted by NovaCoder

And don't say ' but with the vampire FPGA 3D computer it will be running on an Amiga' because that statement is obviously ridiculous.

Once upon a time, people asked the philosophical question of the ship of Theseus - nowadays they will ask "if replacing every custom chip in an Amiga, is it still an Amiga?"

[grond]

Originally Posted by nonarkitten

If we ballpark and say that's ~100 Dhrystone MIPS then the AC68080 is around a Pentium 75.

You are writing nonsense because of your personal disagreements with Gunnar. As you know the 68080 has a much, much faster memory interface than a Pentium 75 (which has a theoretical maximum of 8 bytes * 25 MHz = 200 MB/s). The 080 also does more instructions per clock cycle than a Pentium and each 68k instruction does more work than an x86 instruction. In addition to this the 080 is clocked higher than 75 MHz.

I do agree with you, however, that the 080 as found in the V2 and V4 certainly doesn't do 400 MIPS in any realistic usecase.

[alexh]

Originally Posted by NovaCoder

A raspberry pi (or a modern PC) will be able to give you a better OpenGL retro gaming experience at much (much) less cost.

PiSTorm (the ~800 MIPs raspberry pi based 680x0 accelerator for 16-bit Amiga's) running Emu68 exposes the 3D hardware registers to the 680x0 as if it were an Amiga gfx card. There is a good chance that Warp3D (or some alternative) could be written for PiSTorm in the future.

[skan]

Originally Posted by alexh

PiSTorm (the ~800 MIPs raspberry pi based 680x0 accelerator for 16-bit Amiga's) running Emu68 exposes the 3D hardware registers to the 680x0 as if it were an Amiga gfx card. There is a good chance that Warp3D (or some alternative) could be written for PiSTorm in the future.

This is cool. If we find a way to wrap back WarpUP to 68K, we could run f.i. WipeoutXL on emu68+3D. Crazy.