What do people think about this as next Gen AMIGA?

[Gunnar]

SOCKIT Board

* FPGA with over 100,000 LE
* In FPGA 512 KB build in 2nd Level cache useable by CPU
* As CPU option for 68K Softcore with 200 Mips
* 2 Gigabyte of fast DDR3 Memory
* Ethernet
* USB
* VGA out
* Audio out

[Samurai_Crow]

With a new graphics core that's AGA compatible it would be great! Otherwise why bother? There isn't much non-chipset banging software for Amiga but if you want to run AROS you would be better off with a PC.

[brett71]

Originally Posted by Gunnar

SOCKIT Board

* FPGA with over 100,000 LE
* In FPGA 512 KB build in 2nd Level cache useable by CPU
* As CPU option for 68K Softcore with 200 Mips
* 2 Gigabyte of fast DDR3 Memory
* Ethernet
* USB
* VGA out
* Audio out

As much as my curiosity is always piqued when discussions of a "next-gen Amiga" come out, I'd like to see more specifics on such a machine. And after the promise and total debacle of Natami, I'd much rather not hear anything until hardware is ready for sale and shipping. Call me jaded, if you like.

[robinsonb5]

Originally Posted by brett71

As much as my curiosity is always piqued when discussions of a "next-gen Amiga" come out, I'd like to see more specifics on such a machine. And after the promise and total debacle of Natami, I'd much rather not hear anything until hardware is ready for sale and shipping. Call me jaded, if you like.

The hardware's ready for sale and shipping now. That's the big attraction of this board - it's a general purpose development board made by Arrow. In fact it's probably the first general purpose dev board to be both (a) beefy enough to be interesting and (b) not priced in the stratosphere.

It "just" needs the core to be written.

Of course writing an AGA core for it is a massive undertaking, but getting the existing Minimig OCS core, Gunnar's CPU core and maybe RTG graphics running on it within a matter of months rather than years shouldn't be beyond the bounds of possibility.

(As a bonus the FPGA also contains a dual core ARM Cortex A9 - I'm sure we could come up with some interesting uses for that!)

[Gunnar]

Originally Posted by brett71

I'd much rather not hear anything until hardware is ready for sale and shipping. Call me jaded, if you like.

The hardware is well tested and sold hundred of times.
The above board were on promotoin for $99 per piece.
Awesome value for the money in my opinon.

The promotion is now over the normal price is $299

[Lord Aga]

Seems cool How many LEs could we fill up right now ? I suppose there is enough room for this to be (close)future proof.

If I have found the right thing... it's 300 bucks. That's dirt cheap for a complete powerhouse system.

[Retrofan]

I wonder what OS would be the best to use with those specs. I don't know if it can run OS4.1 either. If the board isn't too expensive it would be a marvellous board to run OS3.9 and AmiKit Real with my RTG plugin, but you know there you can not enter HTML5 web pages with actual browsers.

[Gunnar]

Originally Posted by Lord Aga

Seems cool How many LEs could we fill up right now ?

Lets give some ballpark numbers

* AMIGA Chipset OCS
~ 4.000 LE

* AMIGA Chipset AGA
~ 5.000 LE

* Fast 68K CPU with Caches
With weak Superscalar
~ 9.000 LE

* Very fast 68K CPU with strong super scalar and all bells and whistles
~ 16.000 LE

* 68060 compatible FPU with ~ 100 MFlops
~ 10.000 LE

* 24 Bit RTG GFX Card
~ 1.000 LE


The FPGA is fast and has enough room to add lots of features

[Lord Aga]

Comfortable enough for years to come

[Gunnar]

Originally Posted by robinsonb5

Of course writing an AGA core for it is a massive undertaking, but getting the existing Minimig OCS core, Gunnar's CPU core and maybe RTG graphics running on it within a matter of months rather than years shouldn't be beyond the bounds of possibility.

If you have a working base setup with DMA channels then
writing a truecolor RGT core is peanuts - really.
I did it before. :-D
It can be done on a weekend.

Starting with an ECS chipset is not bad.
AGA is actually not that different from ECS.
AGA is technically more an enhancemnt of ECS.
If the ECS chipset is desinged halve way sensible than you can keep
like 80% of the ECS chipset and only need to change 20% to get AGA out of it.

Our CPU core is based around the standard ALTERA Bus protocols.
This means instantiating it in an ALTERA chip like this SOCKIT and using the socket DMA/Memory controler "click and play".

[voyager]

but then again, it's nice that the hardware is here now, and the price is somewhat ok, it takes years? to write the software for it. We are not getting any younger you know :-)

[Gunnar]

Originally Posted by voyager

but then again, it's nice that the hardware is here now, and the price is somewhat ok, it takes years? to write the software for it. We are not getting any younger you know :-)

Why would you want to write new software?
AMIGA OS is there already - just run it.
Hundreds of AMIGA games are there - just play them.
If you want new software install AROS 68k in it and surfe the internet with a modern browser...

[robinsonb5]

Originally Posted by Gunnar

If you have a working base setup with DMA channels then
writing a truecolor RGT core is peanuts - really.

Yes, indeed.

I did it before. :-D
It can be done on a weekend.

Likewise - the tricky part is CGFX or P96 drivers.

Starting with an ECS chipset is not bad.
AGA is actually not that different from ECS.
AGA is technically more an enhancemnt of ECS.

Agreed - two extra bitplanes, double and quad speed fetch modes and some banking registers - oh, and high-res sprites.
I did actually start poking around with extending the Minimig core's colour registers a long time ago. Didn't get that far, since I was still learning (still am!) and didn't have much time to play.

[Amiga1992]

Originally Posted by Gunnar

Why would you want to write new software?

Why would you want to make a computer like this? Might as well help AROS be as good as it can be on actually modern hardware.

[Gunnar]

Originally Posted by Akira

Why would you want to make a computer like this?

Lets say you put into the board the following VHDL logic.

* AGA
* 200 Mips 68K CPU
* Truecolor GFX Card
* Network
* 2 Gigabytes of fast memory <- Giga not Mega!

What would an A4000 system with the above specs cost you today?

You would get a compatible System to the best AMIGA 4000
with Cyberstrom, Cybervision, Network Card, USB Card,
for a fraction of the price.
And the system would be brandnew, with waranty.

[robinsonb5]

Originally Posted by Akira

Why would you want to make a computer like this?

Because it's a fun and interesting project, of course!

[Mrs Beanbag]

Originally Posted by Akira

Why would you want to make a computer like this? Might as well help AROS be as good as it can be on actually modern hardware.

Why would you not want to?

This IS actually modern hardware!

[brett71]

I do stand corrected....I didn't realize the "SOCKIT Board" at the top of Gunnar's first post was actually a product name that was already in existence...

The specs look impressive and I agree, compared to the price of a used 4000, this thing would be far superior even if it basically emulated the same hardware specs as a 4000.

[voyager]

Originally Posted by Gunnar

Why would you want to write new software?
AMIGA OS is there already - just run it.
Hundreds of AMIGA games are there - just play them.
If you want new software install AROS 68k in it and surfe the internet with a modern browser...

No no no.. ;-) I mean the software for the FPGA core it self

[Gunnar]

Originally Posted by brett71

The specs look impressive and I agree, compared to the price of a used 4000, this thing would be far superior even if it basically emulated the same hardware specs as a 4000.

With the fast FPGA and the huge caches you can give it in this FPGA - your 68K would be real fast - it would run cycles around PPC power up systems.


And actually the Sockit board is not the only board on the market.
There are also another good boards on the market.
One of them cost $170 but it has only 512 MB memory.


I wonder if creating special hardware like the Minimig is the future or if choosing existing FPGA systems that give great value for little money is the future.

Would a model be possible that developers create new CPU and chipset and offer this for this stock cards for a little bonus..

What do you think?