How easy is it to make an FPGA both Opcode and Pin compatibl

[ChuckT]

How easy is it to have an FPGA that is both Opcode and Pin compatible?

If you were to clone a modern day chip, are there large enough FPGA chips or space to emulate a larger chip?

For example, I searched the internet with Google looking for an FPGA clone of the Motorola 68000 series CPU and couldn't find one. That doesn't mean there isn't one.

At what point does a chip become too large to become feasible?

I thought I would ask the question under "Newbies" than to clutter the logic section with a question.

[BigEd]

Well, FPGAs do keep getting bigger. The largest ones are amongst the largest chips made, but they are also extraordinarily expensive.

Fortunately, the 68k isn't so complicated as a CPU. I found some open source HDL at
http://opencores.org/project,ao68000
which says

Quote:

Uses about 4750 LE on Altera Cyclone II and about 45600 bits of RAM for microcode,

Edit:more resource pointers at http://www.amigacoding.de/index.php?topic=296.0

Note that CPUs after the 68000 tend to have on-chip caches. We're told that FPGAs do not efficiently implement caches, but that doesn't mean it can't be done - FPGAs do generally have on-chip RAM.

For a more modern example, the 2006 SPARC from Sun (now Oracle) was released as the OpenSPARC, and will fit into an £1800 FPGA chip.
http://www.oracle.com/technetwork/syste ... index.html
http://www.oracle.com/technetwork/syste ... 530358.pdf

Hope this helps
Ed

[ChuckT]

The 68050 has 2.5 million transistors. Can an FPGA emulate the 2.5 million transistors?
Can an FPGA be opcode and pin compatible?

[BigEd]

I don't see the '050 in http://en.wikipedia.org/wiki/Motorola_6 ... ly_members, but anyway:

- opcode-accurate: yes of course, if someone takes care and does the work.
- pin-compatible: yes, I don't see why not. But existing projects might not implement everything exactly - they might only implement what was needed.
- pinout/package compatible: no, surely an FPGA will be a different size and shape and have different power and ground pins. You'd need an adapter board if you wanted a plugin replacement.
- cycle-compatible: probably not, because that's more work, more verification, and a much smaller number of people would be interested in it.

Note that later CPUs are often disproportionately larger because they dedicate transistors and complexity to performance enhancements. An HDL version doesn't necessarily need to do the same, depending on what level of performance and compatibility you want. For example, cache size will only affect performance, not software-level behaviour.

[ChuckT]

BigEd wrote:

I don't see the '050 in http://en.wikipedia.org/wiki/Motorola_6 ... ly_members, but anyway:

I took the specs off of Freescale's website for the 050. Someone was trying to make a 060 and 070.

http://www.freescale.com/webapp/sps/sit ... de=MC68060
http://cache.freescale.com/files/32bit/ ... df?pspll=1

My computer didn't block these sites at home but Emisoft blocked them at work and I haven't had any security issues with this site so I feel it is safe.

Someone is still making a 68020 but I'm not sure what the difference is between a real 68020 and this one that is implemented:

http://www.tekmos.com/products/68020-microprocessors

My real question is: I know that people have worked on 6502 in FPGA so I'm wondering what the level of performance has been and how difficult it would be to go from implementing the 6502 in FPGA to implementing something larger. I know that Parallax's Propeller chip has been implemented in FPGA so I guess it is not impossible.

Thank you for taking the time in answering my questions.

[Dr Jefyll]

A 6502 core running on FPGA can achieve speeds five times faster, or more, than an actual 6502/65c02. Is this why you're interested in FPGA cores? Best to check, but I'm pretty sure if you do the comparison for a more complex CPU (such as 68000 compared with a 68000 FPGA core) you'll find the speedup is less dramatic.

Are you asking simply out of curiosity, or do you have a goal -- such as building a 6502 or 68000 project?

cheers,
Jeff

[ChuckT]

We were watching the progress of others try to make 68060 emulation in FPGA which they did but since they didn't get the results they wanted, they stopped.
Because there has been no progress for three years, we were trying to figure out the practicality of the project and I've been trying to explain to another user that it isn't so easy so I tried to make sense of the datasheet for them to show them the difficulty level because they are waiting and waiting for something a little more than vaporware and they are not willing to walk away after the project abruptly stopped with a lack of communication.

[BigEd]

Hi Chuck
When you say the project didn't get the results it wanted, is that to say that the CPU was finished, but not fast enough? Or too big to fit on the target FPGA? Or, perhaps, it was never completed or never debugged?

It is actually very difficult to finish a CPU project - for each successful effort there are several which don't make it. I think it's crucial to keep the thing as simple as you can - even though there are always temptations to make it more featureful or faster. In a hobby context, you also have the issue that people's motivation and interest level will ebb and flow, and real-life issues like work, family and health may intervene. And if there are even two people involved, there will be some scope for disagreement which may or may not resolve. Yet, with only one person, it will inevitably take longer. Even more reason to start simple, finish that, and only then move on to add features.

All that said, a CPU project is interesting, and anyone making progress on one will learn from it.

Cheers
Ed

[ChuckT]

BigEd wrote:

Hi Chuck
When you say the project didn't get the results it wanted, is that to say that the CPU was finished, but not fast enough? Or too big to fit on the target FPGA? Or, perhaps, it was never completed or never debugged?

Commodore went bankrupt and only Eyetech did something with the Amiga I.P. Everyone else knows the story that it has been 20 years filled with a lot of disappointments. Users found that you could recreate the Amiga using FPGA and their product was called the Minimig which works. There are other FPGA products in the work like the FPGA Arcade. There is another Amiga clone that was in the works which I will not name because I don't want to hurt feelings but everyone knows who they are. They spent eight years, only a few team members got prototypes and there were hardware bugs and technical issues that couldn't be easily overcome and because it could do everything but AGA mode, they wouldn't release the computer to the public. We were invited to someone's private hobby project which was a mistake because we thought for reason of being invited that we might be able to buy a computer which didn't happen because we were reminded that it was the owner's personal hobby project and he financed it and it belonged to him. He basically stopped talking and we haven't heard from him in three years. Some of us just stuck around because others didn't quit and we were hoping for reconciliation and some are still under the illusion that the project will go forward because they don't want to quit so some of them have kept the website up for the technical information and some are working on their own project. I basically told them they don't have a product and two other Amiga based companies will upgrade their projects and their "hobby" project will eventually become obsolete because the software company and the two established companies will rewrite what Amiga is and re-write history because having something tangible in your hands is more powerful than waiting eight years. In some ways waiting for Commodore's intellectual property to be revived and hanging onto a computer we all loved is nostalgia or a big concern for those already invested in it.

My only interest was surveying the available hardware for a single board computer.

[BigEd]

I think it's true that all the completed CPU projects are pretty much one person getting started - others might join in later - and not those which start with a wish-list conversation. There are always going to be more people with creative ideas than there are with the skills and motivation to finish a project. It's not a shortage of ideas which leads to a lack of completed projects, it's a shortage of finishing power. Group projects do have an extra failure mode, which is "musical differences." Projects with a commercial ambition have another failure mode, which involves keeping everything secret.

What do I recommend? Doing it yourself. Start simple. Learn by doing. Work in the open, and open source your work. By all means ask for help, but also be sure to do your own searches and study. If you don't have the wherewithal to pursue your own project, try to assist on an existing one. Bug reports, testing, documentation, bug fixes are all useful contributions.

For reference, I found a conversation about a 68060 project at
http://www.amiga.org/forums/archive/ind ... 63590.html
and it might serve as a useful illustration of how things often end up.

Cheers
Ed

Edit: typo

[BigDumbDinosaur]

BigEd wrote:

For reference, I found a conversation about a 68060 project at
http://www.amiga.org/forums/archive/ind ... 63590.html
and it might serve as a useful illustration of how things often end up.

Basically a lot of back-and-forth, and nothing actually accomplished.

[jac_goudsmit]

I'm not sure who/what you guys are talking about when you mention locked-up Commodore IP, but there's a project called MIST which has an FPGA and an Arm CPU on board, and lets you emulate an Amiga (not sure which one, probably not 68060 based anyway) or an Atari ST. As far as I can tell, it's in production and you can order it for EUR 199.99.

http://lotharek.pl/product.php?pid=96

But I'm sure you already know that, and anyway, what was this thread about again?

===Jac

[ChuckT]

There has been a lawsuit between the hardware maker and the software maker for Amiga. I could find the pages on Amigaworld but the story goes on for pages and pages and pages.

No one knows who owns what in the Amiga I.P. because there is doubt in a lot of things.

http://amigaworld.net/modules/newbb/vie ... at&order=0

Tulip Computers changed their names to Nedfield and they can't use the brand name "Commodore".

http://amigaworld.net/modules/newbb/vie ... orum=17&96

Then I read that someone was selling the Amiga I.P.

http://www.osnews.com/story/23983/Amiga ... _Trademark

So it is very confusing indeed.

[mstram]

BigEd wrote:

For a more modern example, the 2006 SPARC from Sun (now Oracle) was released as the OpenSPARC, and will fit into an £1800 FPGA chip.
http://www.oracle.com/technetwork/syste ... index.html
http://www.oracle.com/technetwork/syste ... 530358.pdf

Thanks for posting that !

Here in my "cave", I hadn't heard about the OpenSparc !

Mike

[ChuckT]

jac_goudsmit wrote:

I'm not sure who/what you guys are talking about when you mention locked-up Commodore IP, but there's a project called MIST which has an FPGA and an Arm CPU on board, and lets you emulate an Amiga (not sure which one, probably not 68060 based anyway) or an Atari ST. As far as I can tell, it's in production and you can order it for EUR 199.99.

http://lotharek.pl/product.php?pid=96

But I'm sure you already know that, and anyway, what was this thread about again?

===Jac

The Amiga I.P. will always be a mess as to whom owns what. Some of the threads are 18 pages long.

Amiga Inc. started trial versus Hyperion:
http://www.amigaworld.net/modules/newbb ... 32&forum=2

Amiga Inc vs Hyperion: today the 10 days deadline for Hyperion to respond to the Court case:
http://www.amigaworld.net/modules/newbb ... 37&forum=2

First round goes to Hyperion"
http://www.amigaworld.net/modules/newbb ... 29&forum=2

Hyperion VOF v. Amino Development Corporation:
http://www.amigaworld.net/modules/newbb ... 76&forum=2

Amiga Inc v. Hyperion VOF (update 22 Feb 2008):
http://www.amigaworld.net/modules/newbb ... 97&forum=2

Amiga Inc v. Hyperion VOF part 2:
http://www.amigaworld.net/modules/newbb ... 25&forum=2

The lawsuit... a never ending story.:
http://www.amigaworld.net/modules/newbb ... 79&forum=2

T. Frieden, H. Frieden & A. Vallinotto v. Amiga Inc. & Hyperion VOF:
http://www.amigaworld.net/modules/newbb ... 26&forum=2

Itec steps forward in the Big Apple, part 2 DISMISSED:
http://www.amigaworld.net/modules/newbb ... 59&forum=2

New court documents #124 - #126:
http://www.amigaworld.net/modules/newbb ... 61&forum=2

Hyperion and Amiga Inc. reach settlement!:
http://www.amigaworld.net/modules/newbb ... 77&forum=2

Hyperion/ITEC/AMINO Agreement posted on Justia.com:
http://www.amigaworld.net/modules/newbb ... 48&forum=2

All of the things which came out in the lawsuit have made us doubt if certain parties ever had a right to the I.P.
Other people are selling the rights even though the court granted Hyperion the right to the operating system they made for Amiga.

This message goes with my previous message.

Chuck