New 68060 accelerator idea

[pandy71]

Originally Posted by modrobert

Interesting, I didn't know 68060 moved towards RISC compared to previous iterations, was assuming it was the peak of CISC. Yes, then it makes sense with 68020 as an open source reference design instead.

Good to know, because the logic analyzer probes (adapter boards with sockets) I've been looking for are a lot cheaper for 68020 compared to 68040/68060, and less CPU pins.

Motorola started slimtrim already in 68030 and each CPU after 68020 have less and less instructions (but usually faster execution thx to RISC, cache etc).

[rarefaction_]

Closer and closer to the design stage we come.

As it stands i'm going to start with a fast 060 board with fast mem then work in usb, ethernet and some form of RTG, then down the road a little further, high quality audio.

I'll save the PPC stuff for another version after I gauge the reaction and demand for the initial product.

I'm definitely not in this for the money - it's all about the passion for electronics, and of course the Amiga platform.

What about developing an FPGA / emulated core that is in the form factor of the PGA 68060 chip that people could exchange themselves?? Even if this was an AMD x86 CPU running a small kernel and UAE type emulation using JIT?

To reduce costs I think if I supply the board initially without the CPU, so the user can source one to their liking... This obviously would reduce the initial cost dramatically.

First things first is to get a working 060 board with DDR memory and a CF slot or 2 for storage.. there are a lot of open source projects around that can be adapted somewhat, at least giving me e starting point of sorts.

What do you guys think?

Cheers!

[Locutus]

Tip, keep a open development blog or atleast a github so people can follow your commits. With some luck you will get constructive support as well :-)

No Code/schematics shown -> total disbelief.

[trixster]

I applaude your ambition but I can't help but feel this project is just a little too ambitious. Good luck though.

[strim]

Originally Posted by Locutus

Tip, keep a open development blog or atleast a github so people can follow your commits. With some luck you will get constructive support as well :-)

Yes, agreed about that. Transparency and openness is the best thing that could happen to such project.

[alexh]

Originally Posted by rarefaction_

What about developing an FPGA / emulated core that is in the form factor of the PGA 68060 chip that people could exchange themselves??

Unlikely but not impossible. You not only need an FPGA but voltage regulators and voltage level shifters plus all the the FPGA related programming gubbins.

Originally Posted by rarefaction_

Even if this was an AMD x86 CPU running a small kernel and UAE type emulation using JIT?

How do you get the data in/out? You'd need an FPGA to convert from a bus on the AMD (dunno what they would be) to 68000 bus. You may be able to use an ARM SoC as they tend to have more user programmable pins but they are not usually designed to operate at MHz speeds.

Originally Posted by rarefaction_

First things first is to get a working 060 board with DDR memory and a CF slot or 2 for storage.. there are a lot of open source projects around that can be adapted somewhat, at least giving me e starting point of sorts.

Good luck.

Don't be put off that this is not a simple PCB design project. You've already stated it won't be "find the parts, bang them down on a PCB, job done".

It is going to be an interesting time for you. Learning VHDL/Verilog, FPGA synthesis & placement tools. PCB schematic capture and layout. Power regulation and voltage shifting plus 680x0 assembler for the BootROM etc.

Hobbyist level I'd say it is a good 3-4 man years work if you start from scratch with no skills.

Majesta, MikeJ, tnt23, strim, Jaqube, and maybe Schoenfeld are going to be your best contacts over these years.

[rarefaction_]

Originally Posted by alexh

How do you get the data in/out? You'd need an FPGA to convert from a bus on the AMD (dunno what they would be) to 68000 bus. You may be able to use an ARM SoC as they tend to have more user programmable pins but they are not usually designed to operate at MHz speeds.

I was thinking along the lines of a SoC type module in that form factor, not just a CPU with an adapter, So the SoC module has the x86 CPU, RAM, ROM, Flash, IO etc that is just in the form factor of a 68060 chip, so it's powered from the 3.3V bus and gives out the right signal level pin-out as the real chip.

I am not at hobbyist level, I do have a degree in EE, but learning 68k ASM, and the nitty gritty of the Amiga architecture are going to be my biggest learning curves form what I can ascertain. I do have experience with suites like Cadence and NI MultiSim, but i'm not 100% sure what software package would be the best to start something like this...

Thanks a lot from the feedback once again.

[eXeler0]

@rarefaction what is your take on RTG? Countless cheap chips there. Problem is usually on the driver side.

[beezle]

For some bizarre reason it appears he seems to think that you can tweak an existing driver for a completely different GPU and have it work. (ie. He suggested using a "radeon PCI" (I can only assume he means 68k radeon drivers)).

Thread was funny at one point, but reading it in full changed it from funny to "oh boy, here we go again" status pretty quickly.

[trixster]

Over a year later. Did this go anywhere?

[Signman]

Originally Posted by trixster

Over a year later. Did this go anywhere?

I'm going out on the limb and guess no.

[eXeler0]

Originally Posted by Signman

I'm going out on the limb and guess no.

He could be working in the silent as God knows, this crowd is tough on vaporware.. ;-)
Or he gave up once he truly realized all the hard stuff that needed to be done. In any case, it would be nice to get a word or two about the project. Even so..even if he walked away, it would be interesting to hear which part he underestimated and why. Someone could learn something from it.

Skickat från min LG-H850 via Tapatalk