Most bang for 600 USD?

[Mrs Beanbag]

http://www-wjp.cs.uni-saarland.de/publikationen/SWWPS04.pdf

GPUs, on the other hand, are made for massive parallelization, but aren't necessarily optimal for full on ray-tracing, for a variety of reasons, impressive though they are.

[danbeaver]

I was hoping the UltimatePPC would come in about that price range; I've got a lot of life left in my current Amy's

[freqmax]

I'd like to see a link for that. "Done in an FPGA" sounds like the crucial factor there, anyway - raytracing is made for massive parallelization, and in an FPGA that's easy to do. On a general-purpose CPU, not so much.

A box with standard CPU but multiple buses that let's one exploit multiple FPGAs and DSPs? That would at least open the software defined radio area and most definitely power raytracing.




As for raytracing: Realtime Ray Tracing of Dynamic Scenes on an FPGA Chip:

Using a single FPGA chip running at 90 MHz it offers realtime rendering performance of 20 to 60 frames per second

The SaarCOR prototype is build using a Xilinx Virtex-II 6000-4 FPGA [Xil03], that is hosted on the Alpha Data ADM-XRC-II PCI-board [Alp03]. The board contains six independent banks of 32-bit wide SRAM (each 4MB) running at the FPGA clock speed, a PCI-bridge, and a general purpose I/O-channel. This channel is connected to a simple digital to analog converter implementing a standard VGA output supporting resolutions of up to 1024 x 768 at 60 Hz.

I think the above specifies what is needed in terms of hardware to accomplish realtime raytracing. One catch is that a suitable FPGA seems to cost like 100 USD at digikey.

[Mrs Beanbag]

Indeed, and in fact if you look at it one way, the Blitter is a sort of FPGA - the minterms register is essentially a LUT!

[Digiman]

You can probably build brand new Amiga 1000s  for $600 in material costs in China easy BUT who is going to pay the millions in R&D to recreate a 100% compatible Amiga 1000 motherboard and do all the testing?

Manufacturing costs per unit are not the issue, it's finding a genius to create something fantastic and revolutionary from the ground up that is difficult, R&D and upfront investment is why it doesn't happen on a regular basis. Even Macs are just regular PCs in different cases.

[wawrzon]

I am considering a project of my own, making an accelerator for A1200 with an ARM CPU running software emulation.  What is the point of that?  Well lots of people are doing it with FPGAs but it would seem cheaper, faster and easier to do it this way as there are software emu's already available, ARM chips come in the GHz range already and are widely available (and the logical next step for desktop PCs in general if you ask me) and affordable.  The software would be in ROM on the card itself so you wouldn't know it was there.  Why not just emulate the whole Amiga on a PC?  Well because then you have to host the emulator in some other OS you have to boot up, ugh.  And it doesn't feel the same.

i proposed something of the like all along. if you do that, be my heroine. if not join aros68k project to massively upgrade amiga experience!

[freqmax]

An ARM accelerator can be accomplished by making a direct link from a mobile phone that usualy feature an ARM-CPU clocked to GHz to the CPU socket in the Amiga or other suitable motherboard connection.

[amiman99]

You can probably build brand new Amiga 1000s  for $600 in material costs in China easy BUT who is going to pay the millions in R&D to recreate a 100% compatible Amiga 1000 motherboard and do all the testing?

Manufacturing costs per unit are not the issue, it's finding a genius to create something fantastic and revolutionary from the ground up that is difficult, R&D and upfront investment is why it doesn't happen on a regular basis. Even Macs are just regular PCs in different cases.

The improved A1000 MB is already done, just needs production.
http://www.illuwatar.se/project_pages/gba1000/gba1000.htm

[commodorejohn]

Interesting; some of those scenes are pretty non-trivial, moreso than I'd have expected.

[Digiman]

Interesting project. No updates for 3 years on gba1000?

For mass production you'd need to make cases, replica keyboard+interface, an easy to produce equivalent of all the chips inside an Amiga 1000, re-manufacture 3.5" Amiga compatible FDDs etc so even if that was a PCB layout ready to print out on a suitable machine there is still over a £million in initial investment needed whether you made a souped up faster model or an exact replica of original performance (but using today's production techniques and technologies like FPGA etc).

[Iggy]

Personally, I'd just buy a 4000 on Ebay.

[freqmax]

I suspect even the TI:s "Stellaris LM4F120 LaunchPad - ARM Cortex-M4" for 5 USD would run circles round the A4000. So the thing is to get the most out of those 600 USD, and I suspect a combo of DSP-FPGA-RAM will do just that.

[Iggy]

I suspect even the TI:s "Stellaris LM4F120 LaunchPad - ARM Cortex-M4" for 5 USD would run circles round the A4000. So the thing is to get the most out of those 600 USD, and I suspect a combo of DSP-FPGA-RAM will do just that.

I don't know. We'll see when TI gets around to shipping me one. They're only 80 MHz so this isn't going to be A9 level performance.
 
Besides, I was comment on the idea of bothering with the gba1000.

[persia]

So there is no answer to the original question?

[psxphill]

Oh and for ARM emulation of 68k, it's a nice idea but will wreck cycle accurency and lock-step.

When you add any accelerator to an Amiga then cycle accuracy is no longer possible. The point is to make it run quicker than any 68k Amiga ever made. Most software won't care & if it does then it's most likely broken. It's not like the c64 where the processor is in complete synchronisation with the vic and sid.
 
Using one of intels new low power processors would be interesting too (like the atom). The choice should be made on whichever gets the most performance for the price (within reasonable price limits of course). AFAIK they are all little endian, so some performance would be lost in the conversion either way. Although ARM is 3ghz, that doesn't really mean anything. Instruction throughput is what you need to consider, it's entirely possible that a 1.5ghz processor can beat a 3ghz processor depending on their design (the p3 vs p4 came close).
 
I have had a similar idea for a while and would be prepared to put some time into the software side.

[danbeaver]

Many of these ideas are "neat!"
But let's look at the numbers:  The latest & greatest (by some opinion) Amiga is the X1000; price? About 2 grand USD. Number sold? I don't know, but I guess maybe a few hundred. Number of active Amiga users worldwide? Maybe 5,000. The number who just want to play retro games? Let's guess 1/2 to 2/3rds. The number of new, non-Amiga users, "Techies," who would buy a new Amiga? Probably the same number who bought an X1000 (mind you these are folks who may have heard of the Amiga name, could afford it, and decided they could put to use such a machine)?  Maybe a few hundred.

So, how do recoup your R&D?  You all answer that question, because 600 USD doesonly go very far.

[The main question, by the way has a punch-line if you add an "s" to the word "bang"]