PPC motherboard A1222 "Tabor" revealed

[eXeler0]

Originally Posted by OlafSch

Trying to compete with modern graphic cards in 3D area is a crazy idea. Perhaps offering some low level routines in FPGA so that 3D games could be optimized a little but more does not make sense. 68k needs as much as possible horsepower and RAM. The strength of Amiga was 2D gaming so there should be the focus, trying to compete with PCs today in 3D area makes no sense.

It's beyond crazy. It's actually impossible. AFAIK Modern GPUs are technologically beyond any FPGA known to man with billions of transistors, beyond GHz klockspeeds and recently 4096 bit memory architectures (HBM).

However, adding some 1998-era 3d functions could offer some new fun challanges for devs. (Thinking of the guy that rewrote part of the quake engine to use the DSP in the Atari Falcon for example).
And this of course, would sit on top of superAGA for 2D which would be fully compatible with AGA only without the obvious limits. ;-)

[Megol]

Perhaps the most realistic (still very unrealistic!) way to get to make an ASIC to improve 68k performance would be making a special-purpose FPGA? Making it open-source perhaps could build enough interest from people that it actually could be manufactured...

What I'm thinking of is a FPGA where:

. large memory blocks having dedicated interconnects to build caches.
. small memory blocks either have direct support for multiple writes or dedicated hardware to emulate multiple write ports (for register files).
. ALU blocks that supports add/sub/logical/shift operations, SIMD splitting etc.
. support for result bypassing.
. multiplier blocks suitable for SIMD execution, dedicated logic to build larger multipliers and direct support for signed/unsigned multiplications.
. CAM (Content Addressed Memory) support.
. Signal routing specialized for larger buses.

This would be less dense than a normal FPGA but for things that match the architecture the result could be impressive.

Alternatively it could be even more hardwired for processor applications, something like a processor with some programmable hardware to support decoding + architectural quirks.

[iggybeans]

OR, we could do an FPGA system with PCI-e support and offer RTG graphics as an upgrade option.

[eXeler0]

Originally Posted by iggybeans

OR, we could do an FPGA system with PCI-e support and offer RTG graphics as an upgrade option.

Deleted comment

[matthey]

Originally Posted by OlafSch

Trying to compete with modern graphic cards in 3D area is a crazy idea. Perhaps offering some low level routines in FPGA so that 3D games could be optimized a little but more does not make sense. 68k needs as much as possible horsepower and RAM. The strength of Amiga was 2D gaming so there should be the focus, trying to compete with PCs today in 3D area makes no sense.

A new Amiga would need to go retro which doesn't need state of the art technology. There are several options for adding 3D.

1) Reverse engineer and implement existing 3D from a Voodoo 4/5 or console compatible 3D.
2) Create a 3D core from scratch or Start with an open 3D core like GPL-GPU (https://github.com/asicguy/gplgpu) or Nyami.
3) License a synthesizeable 3D core like VideoCore IV, PowerVR, etc.

It would be nice to have shading units and 3D performance from this decade but older 3D compatibility has its advantages (cheaper).

Originally Posted by eXeler0

It's beyond crazy. It's actually impossible. AFAIK Modern GPUs are technologically beyond any FPGA known to man with billions of transistors, beyond GHz klockspeeds and recently 4096 bit memory architectures (HBM).

3D graphics need deep pipelines which may not suffer too much in an FPGA. The best graphics board manufacturers have finely tuned expertise in 3D performance and guard their secrets closely. Trying to compete with them would be crazy but isn't necessary.

Originally Posted by Megol

Perhaps the most realistic (still very unrealistic!) way to get to make an ASIC to improve 68k performance would be making a special-purpose FPGA?

Special purpose usually means a special price too. High end FPGAs are getting specialized hardware but they are expensive. IMO, creating an ASIC and selling cheap hardware makes more sense. A million dollars may sound expensive to individuals but designing and manufacturers boards like the Tabor probably costs hundreds of thousands already.

Originally Posted by iggybeans

OR, we could do an FPGA system with PCI-e support and offer RTG graphics as an upgrade option.

Sure, but the Amiga custom chip graphics output would be separate from graphics board output. Many of us classic RTG graphics board users have put up with dual monitors and monitor switches but it is less than optimal. PPC boards like the Tabor aren't compatible with the custom chips so they only need graphics board output but less compatibility is less fun .

[eXeler0]

Originally Posted by matthey

A new Amiga would need to go retro which doesn't need state of the art technology. There are several options for adding 3D.

1) Reverse engineer and implement existing 3D from a Voodoo 4/5 or console compatible 3D.
2) Create a 3D core from scratch or Start with an open 3D core like GPL-GPU (https://github.com/asicguy/gplgpu) or Nyami.
3) License a synthesizeable 3D core like VideoCore IV, PowerVR, etc.

1. Would probably be a bad idea. Spending time and effort on something nVidia owns all patents for?

2. Probably the wisest choice here..

3. More Licensing fees in Amigaland..? probably not desirable..

[pandy71]

3D for FPGA http://en.wikipedia.org/wiki/M-Labs

But still this all have no sense nowadays... go for some cheap PC than FPGA - even AMD A4 will be 20 - 50 times more powerful in 3D than FPGA at fraction of cost...

[eXeler0]

" Ideally suited for the printing and imaging market, the P1022 processor includes advanced deep sleep power and energy management features that enable developers to design next-generation energy efficiency, environmental and governmental energy regulatory power requirements."

This is from the official data sheet for the p1022 CPU.

At least there are versions of Debian running on ppc e500 v2 core (the core which the p1022 is based on).

[Samurai_Crow]

For a second generation FPGA GPU, I would base it on the 3do. Its patents are nearly expired and was designed by former Amiga developer R. J. Mical.

[iggybeans]

Originally Posted by eXeler0

" Ideally suited for the printing and imaging market, the P1022 processor includes advanced deep sleep power and energy management features that enable developers to design next-generation energy efficiency, environmental and governmental energy regulatory power requirements."....

Yes, its energy efficient.
All recent PPCs are.
And it runs a variant of Linux that supports the spe fpu.

We have had this info for years.

Its still not compatible with current NG OS' and using it will still incur a performance penalty.

[iggybeans]

Originally Posted by Samurai_Crow

For a second generation FPGA GPU, I would base it on the 3do. Its patents are nearly expired and was designed by former Amiga developer R. J. Mical.

Interesting, ARM based with a planned successor that used dual PowerPC 602 processor. Sounds a little too NG for you Crow.

Although, now that I think about it there are FPGAs that feature dual PPCs (one that can be had from some sources for about $50).

Maybe that isn't such a bad idea.
Make it better than (but hopefully compatible with) the Amiga core, and give it directly embedded PPCs for WarpOS and NG OS support.

That would freak out Hyperion.

A board that could run OS3.1-OS4.1.

Real legacy support without a tether to Ben Hermans.

[Samurai_Crow]

Originally Posted by iggybeans

Interesting, ARM based with a planned successor that used dual PowerPC 602 processor. Sounds a little too NG for you Crow.

Although, now that I think about it there are FPGAs that feature dual PPCs (one that can be had from some sources for about $50).

Maybe that isn't such a bad idea.
Make it better than (but hopefully compatible with) the Amiga core, and give it directly embedded PPCs for WarpOS and NG OS support.

That would freak out Hyperion.

A board that could run OS3.1-OS4.1.

Real legacy support without a tether to Ben Hermans.

I was only referring to the GPU, not the CPU. The Sprytes in the 3do are more like a hardware accelerated Layers.library thus merging sprites, playfields, and BOBs as one entity.

[iggybeans]

Originally Posted by Samurai_Crow

I was only referring to the GPU, not the CPU. The Sprytes in the 3do are more like a hardware accelerated Layers.library thus merging sprites, playfields, and BOBs as one entity.

Understandable, if you are going this route you would still want 68K compatibility, but an Amiga core with improved graphics and two 500MHz PPC cores couldn't hurt.

The improved sprite functions sound interesting.

[Megol]

Originally Posted by matthey

Special purpose usually means a special price too. High end FPGAs are getting specialized hardware but they are expensive. IMO, creating an ASIC and selling cheap hardware makes more sense. A million dollars may sound expensive to individuals but designing and manufacturers boards like the Tabor probably costs hundreds of thousands already.

Well it would be an ASIC so there are extra costs. But it wouldn't be a replacement for a standard FPGA, it could be made much smaller and more efficient for CPU implementations.

The advantage for doing something like that would be increasing the potential market. Making a new microcontroller platform is simply useless - there are all kinds of variants out there from low to high cost and making something similar would take many years and/or a lot of licences for building blocks (memory, I/O etc.).

In comparison the CPU-FPGA would be useful for software->hardware translations, processor experimenters, university courses in computing science and even (if the costs can be made low enough) for retrocomputing enthusiasts. I'd assume such a chip would have a hardware memory controller, general purpose I/O (low speed) and PCIe connections. If one need more I/O standard support it could be done in external FPGA chip(s).

And there are many ways one could improve performance for making a processor in such a chip.

[trixster]

To quote Toni Wilen from 04 Dec 14:

".....no one is going to use it without support driver software. New stuff that can be only used by direct hardware access and can't be used (and developed) with any existing hardware: Why would anyone use it?

It seems every FPGA implementation with (planned) extensions will have same problem: who does the software? (and how? it is practically impossible to extend existing graphics.library stuff) No one is going to do it if it is not going to be standard. Single standard implementation + Amiga? Not going to happen!"

Ok, he wasn't talking directly about some of the stuff discussed in this thread, but I think it's a valid point. Who will use all these great ideas? There's no new software - no new games, no new productivity software in development to tempt people back to a platform which in everyone's eyes disappeared in the 90s. The Amiga name is not enough. All my mates who owned amigas in the past have no knowledge of anything post workbench 3.0. AmigaOS 4, ppc, fpga; it's all stuff that just exists in the head of a bunch of us retro enthusiasts. A bigger audience to justify development cost and time is not going to come back. No one is going to spend hundreds of thousands delevoping 3D cores for a machine which has nothing to run on it!

[eXeler0]

Ok, not exactly on topic, but still..relevant..
New computers / hardware projects can attract a big crowd if done right...

https://www.kickstarter.com/projects...er/description

Over $2million collected so far...

[wawa]

Originally Posted by eXeler0

Ok, not exactly on topic, but still..relevant..
New computers / hardware projects can attract a big crowd if done right...

https://www.kickstarter.com/projects...er/description

Over $2million collected so far...

if they are cheap enough...
just another example how terribly wrong the policy around os4 occures to be. wrong, as a vision of a way forward, maybe not that wrong for small scale business that needs a steady revenue and without much ambition.

[kolla]

Why all this talk of 3D when there is little/close to no software that makes use of 3D on Amiga?

[iggybeans]

Originally Posted by kolla

Why all this talk of 3D when there is little/close to no software that makes use of 3D on Amiga?

Because its evolutionary.
If you already have 2D, 3D would be the natural progression.

[ReadOnlyCat]

I am not too sure what is the intended end point of these speculative discussions but aren't you tackling the problem in reverse?

The Amiga was special because its features made it special.
The harmonious mix of hardware and software innovation was the key to both its ease of use and power.

It does not matter that it had a 68k, a blitter, and stuff. What mattered for users is what these kittens enabled. What mattered is that users got their jobs done, be it gaming, or else.

If you want to make an Amiga like modern computer, start with the features you want in, start with the job you want it to accomplish with the computer, not the components. And that's determined by software these days, not hardware.

Given how obsolete AmigaOS is (in terms of users-job to be done), frankly, the hardware should be the least of our concerns.

Also I should mention that people are moving away from home computers and using more and more their mobile devices for computing tasks. Failing to address that is guaranteed death if the goal is a mainstream computer (if the goal is a niche market, then it's fine).