Apollo Team has new Cyclone 5 FPGA accellerator cards

[FrenchShark]

Originally Posted by pandy71

Are You sure? E clock is same for A1000 and for A4000 - seem that 8520 are the same speed...

I did not check the A4000 schematics but CBM/CSG has released a 2 MHz 8520 and the timer.device on A1200/A4000 Kickstart checks 4 different tick frequencies :
* 50 Hz with 700 KHz E clock,
* 60 Hz with 700 KHz E clock,
* 50 Hz with 1400 KHz E clock,
* 60 Hz with 1400 KHz E clock,

I would prefer to be wrong : it would give me less headache for converting my A500 or A2000 to AGA :-)

EDIT : I found the source code of timer.device v39.4, here is the power supply frequency detection routine:

SupplyFreqDetect

;We launch the timer B (frequency = 715,909Hz or 709,379Hz) and wait
;for 1 TOD tick. We guess the TOD's frequency with the number of
;decremented EClock ticks.

; +----- 65535 : Timer B start
; |
; |
; |
; +----- 53603 : 60 Hz tick (EClock = 715909 Hz)
; |
; +- 52476 : Threshold 50Hz/60Hz
; |
; +----- 51347 : 50 Hz tick (EClock = 709379 Hz)
; |
; +- 46509 : Threshold ECS/AGA
; |
; +----- 41671 : 60 Hz tick (EClock = 1431818 Hz) - AGA 8520
; |
; +- 39416 : Threshold 50Hz/60Hz
; |
; +----- 37160 : 50 Hz tick (EClock = 1418758 Hz) - AGA 8520

lea _ciaa,A0 ;Get base address of CIA A.

bclr #CIACRBB_ALARM,ciacrb(A0)
moveq #0,D0
move.b D0,todhi(A0)
move.b D0,todmid(A0)
move.b D0,todlow(A0) ;Clear the TOD clock

moveq #CIACRBF_RUNMODE,D0
move.b D0,ciacrb(A0) ;Activate one-shot mode for timer B

moveq #-1,D0
move.b D0,tblo(A0)
move.b D0,tbhi(A0) ;Load the timer B latch with $FFFF

move.b todlow(A0),D0
.WaitTick1
tst.b tbhi(A0) ;Does timer B work ?
beq.b .CiaAFail ;If not, Guru time !
cmp.b todlow(A0),D0
beq.b .WaitTick1 ;Wait for a TOD tick.

;The timer B starts decrementing from $FFFF.
move.b #CIACRAF_LOAD|CIACRAF_RUNMODE|CIACRAF_START,ciacrb(A0)

move.b todlow(A0),D0
.WaitTick2
tst.b tbhi(A0) ;Does timer B work ?
beq.b .CiaAFail ;If not, Guru time !
cmp.b todlow(A0),D0
beq.b .WaitTick2 ;Wait for another TOD tick.

moveq #CIACRBF_RUNMODE,D0
move.b D0,ciacrb(A0) ;Stop the timer B.

move.b tbhi(A0),D0
lsl.l #8,D0
move.b tblo(A0),D0 ;Read the timer B value.

cmpi.l #52476,D0 ;Timer B >= 52476 ?
bcs.b .Not60Hz ;No, skip the next two lines.
moveq #60,D0 ;Yes, we have a 60Hz tick.
bra.b .StoreTickInfo ;Store the information.
.Not60Hz

cmpi.l #46509,D0 ;Timer B >= 46509 ?
bcs.b .Not50Hz ;No, skip the next two lines.
moveq #50,D0 ;Yes, we have a 50Hz tick.
bra.b .StoreTickInfo ;Store the information.
.Not50Hz

cmpi.l #39416,D0 ;Timer B >= 39416 ?
bcs.b .Not30Hz ;No, skip the next two lines.
moveq #30,D0 ;Yes, we have a 30Hz tick.
bra.b .StoreTickInfo ;Store the information.
.Not30Hz

moveq #25,D0 ;Timer B < 39416 : 25Hz tick.
.StoreTickInfo
move.l #1000000,D1 ;1,000,000 µs
divu D0,D1 ;Divided by TOD freq = TOD period (in µs).
move.w D0,tdev_PSupplyFrqHz(A2) ;Store the frequency.
move.w D1,tdev_PSupplyPerMicroSec(A2) ;Store the period.
rts
.CiaAFail
move.l #$15000002,D0 ;Alert number (AN_TMBadSupply).
bra.w DisplayAlert ;Guru time.

Regards,

Frederic

[Gunnar]

Originally Posted by Lord Aga

But... why ? It isn't an Amiga-only board. We could have been having fun with simpler Speccy or C64 cores all this time.

You have to mind the situation they are in.
Mike re-uses an Open Source Amiga Chipset.
Mike also re-uses an Open Source 68K CPU.
This means every addition or enhancement he makes is by law also open source.

This means the day he does sell the first board
he has by law to also make all the enhancements open source.
If he already sold a single board then he is in theory already under pressure to provide all sources on request.

This is nice for the users of Minimig, Mist and all others as they get a free copy of his stuff.

Commercially this is not an easy situation for him as there are
very good boards on the market which have more main memory
and much better FPGAs for a lot less money.

Its really a tricky market situation for Mike.
As soon as he ships the first board he has to free up all his VHDL code.
Then its only matter of time and cheaper and technically maybe
more advanced boards can provide the same AMIGA image
as his has for less money with stronger FPGA and more memory..

[JimDrew]

Originally Posted by FrenchShark

IIRC, the DPLL code I posted handles timings outside 2us/4us cell size as long as the counter is within CTR_MIN_VAL and CTR_MAX_VAL.
GCR (4us bit cell) is handled with the clock enable, I even added HD floppy mode with a faster clock enable. I guess you have added the HD mode on the FPGA Arcade too (maybe even variable bit cell mode) ?

Regards,

Frederic

The DPLL logic you have is not correct for bitcell times that exceed the counter min and max, which is often used for copy protection. How Paula handles that is different depending on the number of valid bitcells up to when invalid bitcell is clocked. Paula can handle variable speed bitcells - this is common for Rob Northern protections. GCR is clocked with a different mechanism than MFM bitcells. In my copy and converter programs I turn the DMA on and constantly alter the DMA length register, address register, and sync match register. So, I can have a single DMA transfer that takes a month to complete if I wanted. That is something that no software Amiga emulation handles correctly.

[JimDrew]

Originally Posted by Gunnar

You have to mind the situation they are in.
Mike re-uses an Open Source Amiga Chipset.
Mike also re-uses an Open Source 68K CPU.
This means every addition or enhancement he makes is by law also open source.

That would be the case, if this were true - but it's not. In the beginning, the core was ported from minimig for testing the hardware. Now, every piece of the VHDL code is from scratch so that the code does not have to be given out, and can remain a commercial product. I do think that Mike will end up releasing the code to the public at some point, but that is not what is happening right now. Making all new from scratch, along with making the hardware abstraction layer (which handles a lot of the Amiga core functions) is what has been taking so long.

The biggest interest, at least in the U.S. is for the Amiga core.

[alexh]

Originally Posted by Gunnar

You have to mind the situation they are in.
Mike re-uses an Open Source Amiga Chipset. Mike also re-uses an Open Source 68K CPU. This means every addition or enhancement he makes is by law also open source.

Unless the original authors have granted him a different licence.

This means the day he does sell the first board he has by law to also make all the enhancements open source. If he already sold a single board then he is in theory already under pressure to provide all sources on request.

Only if his company ships the firmware.

[FrenchShark]

Originally Posted by pandy71

AFAIR Commodore mentioned such possibility in Budgie description - Budgie should be able perform conversion between 16 bit CPU bus and 32 bit CHIP AGA bus.
So this should be possible - A3000 schematics have something like Budgie but on TTL - bunch of 74646 buffers mostly.

If the AGA chipset is put on an FPGA based accelerator board, 95% of the original OCS/ECS chipset becomes useless (16-bit Chip RAM is too slow for AGA DMAs).
Only the I/O related chips stay a little bit useful on the A500/A2000 PCB :
- Paula (Serial port, Joystick/Mouse)
- 8520s (Floppy, Parallel port, Joystick/Mouse, Keyboard)

Regards,

Frederic

[Gunnar]

Originally Posted by JimDrew

That would be the case, if this were true

Does this really hold water?

He stated often enough himselfs to use the open source cores and to have added stuff to them.
He stated himself that he uses Minimig Core and added AGA features.
It was stated often enough that the TG68 core was used.
And the performance values that were posted were exactly the same values as TG68 scores in this type of FPGA.

And even if he now wrote now his own core : as he stated before that he used the open source core - legally he would now be in a position where he needs to prove that he did not copy even a single line of it.

For a commercial product you are NEVER EVER allowed to start with an open source code - even if you later write your own.
As soon as you start with the open source stuff you are legally in a tained position where a Free software laywer would make you to prove that you did not copy anything.

I would never touch the minimig core - because of this legal situation.
My advice for him is to buy a license of the used open source cores...
Then he has a chance to get out of this.

[JimDrew]

You can start with whatever you damn please to test your hardware. If you spend more than a year recreating code using FDIL and other methods to get the exact timing (which has been publicly discussed constantly), you can certainly use that code in a commercial product. There is no legal situation when you are not using any percentage. There is no reason to "buy" anything... that is ridiculous. There would be no issues in the U.S. and we have the toughest copyright and patent laws in the world.

As Frederic points out, just to move to a complete AGA emulation you have to throw away most everything from the OCS/ECS chipset support anyways. Alice, Lisa, Gayle, etc. are all new and handle the jobs of Denise, Agnus, etc. etc. You really just can't patch on to the existing minimig code and make AGA work. You have to start over.

[Gunnar]

I think what Mike did is good,
For his work I wish him the best of luck.

The facts are:
Re-using open source cores without open sourcing your changes is simpy not legal.
If you start with an open source core and then do your own its legally normally seen as derived work - this means its open source again.
It does not matter who provides the core - if he does this or someone else.
As soon as the first FPGA image with single line of open source in it is shipped - the one shipping has to provide the sources.

Legally he put himself in a very messy situation.
Providing the sources is one way to get out of it.
If he does not want this then maybe consulting a lawrer specialized on this would be a good move.

[JimDrew]

If you re-used code, I would agree that would not be legal. The fact is using any reference material to create an original (non-copied) work is legal in any country on this planet. Copying code obviously is not legal, so comparing code side-by-side (along with the bit file you wanted) is an easy way to determine that has not occurred. Simple as that. To say that if I merely look at an open source project and then write my own code that does the exact same thing is not legal is simply ridiculous and clearly shows your lack of understanding of copyright laws. I can look at whatever I like, and as long as I do not steal any code, I can certainly make the same thing if I want to. In the U.S., this is called fair use, and the only exception that applies is for a utility patent. Sure, you might have to prove that you made original code, but so what? As long as you can do that, who cares? In the U.S. this would mean that someone would have to file a suit against you, and then pay *you* damages when they lose (providing of course you can prove your code is original). Can you imagine a world where we couldn't look at something and re-create it exactly and/or improved? There would be only one type of car, one type of computer, one type of cell phone, etc. Given this logic, nobody should be reproducing the Amiga custom chipsets.

I have the source code to much of the core and I can tell you it is substantially different from minimig. So, I have absolutely zero concerns about importing and selling this product.

[Gunnar]

Originally Posted by JimDrew

You can start with whatever you damn please to test your hardware.

While this might your personal opinion - this is not how law in the real world works.

In Companies where FPGA code is coded for money, there its common practise that the VHDL developers have to sign a legal statement - in which they state that they have not used or copied any Open source code, nor even looked at any open souce code related to their topic work - while developing their VHDL code.

What do think why do US companies require this legal statement?

Open source code is like a virus - when you use it in your project your whole code will end up open source.

But lets face it at the end of the day this is good that it works like this.
Otherwise people would just steal an open source code - rebrand it and sell it without opening it again. And this is stealing right?

I think Mike can get very easy out if this by simply asking Dennis und Tobias to sell him a license.

[JimDrew]

I don't think you get it.. there is nothing to get out of. People steal open source code and re-brand it as a commercial product every single day. Sad but true. It's up to the copyright holder to go after those people. Trust me, I have been in plenty of lawsuits over copyright infringements, both here in and overseas. If you are legit and have original code, there are no issues regardless of what you may have seen, experimented with, been exposed to, etc. This is not a personal opinion, it is legally tested fact that I have personally been through. How many copyright infringement lawsuits have you been involved in? Since you have seen the TG68 code, does that also mean you can not create your own CPU core without buying a license? Of course not! As long as you are not using any of that code, you can create whatever you like. You are certainly not starting in a "clean room" environment on a CPU core. You have obtained some knowledge from looking at that code... that knowledge (from what I have seen with some of the code) is knowing what NOT to do.

As far as U.S. companies having legal disclaimers pertaining to open source - that is to reduce the number of dumb lawsuits that might arise from someone believing that their work has been stolen. In reality, that disclosure does not protect anyone, it's more of a scare tactic to prevent their employees from stealing other work instead of using reference material to create their own. A court will decide an outcome should legal action be required. Simple as that. I am really surprised that anyone would believe that open source projects would have this kind of voodoo as reference material. Open source code is no different from commercial source code. The crime is the same (IP theft), regardless of where the stolen code comes from.

[JimDrew]

If you wanted a A4000/Cyberstorm060 type of setup, this where Replay is headed:

http://www.cbmstuff.com/forum/showthread.php?tid=171

[OlafSch]

Originally Posted by JimDrew

I don't think you get it.. there is nothing to get out of. People steal open source code and re-brand it as a commercial product every single day. Sad but true. It's up to the copyright holder to go after those people. Trust me, I have been in plenty of lawsuits over copyright infringements, both here in and overseas. If you are legit and have original code, there are no issues regardless of what you may have seen, experimented with, been exposed to, etc. This is not a personal opinion, it is legally tested fact that I have personally been through. How many copyright infringement lawsuits have you been involved in? Since you have seen the TG68 code, does that also mean you can not create your own CPU core without buying a license? Of course not! As long as you are not using any of that code, you can create whatever you like. You are certainly not starting in a "clean room" environment on a CPU core. You have obtained some knowledge from looking at that code... that knowledge (from what I have seen with some of the code) is knowing what NOT to do.

As far as U.S. companies having legal disclaimers pertaining to open source - that is to reduce the number of dumb lawsuits that might arise from someone believing that their work has been stolen. In reality, that disclosure does not protect anyone, it's more of a scare tactic to prevent their employees from stealing other work instead of using reference material to create their own. A court will decide an outcome should legal action be required. Simple as that. I am really surprised that anyone would believe that open source projects would have this kind of voodoo as reference material. Open source code is no different from commercial source code. The crime is the same (IP theft), regardless of where the stolen code comes from.

Years ago I was active in natami forum and at that time a lot of announcements from FPGA Arcade (including the 060 daughter board). We thought it would "soon" be available. Then the Natami project faded away and FPGA Arcade became silent too for years. Then a serious competition is announced and suddenly there are news from FPGA Arcade again. What should I think about that?

[jbenam]

Originally Posted by JimDrew

If you wanted a A4000/Cyberstorm060 type of setup, this where Replay is headed:

http://www.cbmstuff.com/forum/showthread.php?tid=171

That daughter board made a lot of us drool in awe something like 3 years ago - and pretty much nobody has it except for yaqube.

Any real news about it, like a production run?

[OlafSch]

Originally Posted by JimDrew

If you wanted a A4000/Cyberstorm060 type of setup, this where Replay is headed:

http://www.cbmstuff.com/forum/showthread.php?tid=171

latest when the new accellerator hardware is available the moment of truth is there. Then "announcements" and "demoing" will not be enough anymore...

[Mrs Beanbag]

Originally Posted by Gunnar

Open source code is like a virus - when you use it in your project your whole code will end up open source.

Not necessarily, it depends on the license. LGPL, for instance, is designed to explicitly allow commercial use.

[Megol]

Originally Posted by Mrs Beanbag

Not necessarily, it depends on the license. LGPL, for instance, is designed to explicitly allow commercial use.

And it isn't true of GPL either. GPL isn't a good licence for hardware as it is hard to distinguish if other parts of a system is an extension of the GPL part or if it is separate and just communicating with the GPL part. Not strange as the GPL is intended for software.

Most licences aren't viral in any definition of the term - BSD, Apache etc.

[kamelito]

@Gunnar,

I don't see where you wanna go from here, except if you planned to ask for the source code to use it in your product.

Kamelito

Originally Posted by JimDrew

If you wanted a A4000/Cyberstorm060 type of setup, this where Replay is headed:

http://www.cbmstuff.com/forum/showthread.php?tid=171

Very interesting if the price is low enough. For now I'm just in a waiting mode for the final Core(s).
Kamelito

[JimDrew]

Originally Posted by jbenam

That daughter board made a lot of us drool in awe something like 3 years ago - and pretty much nobody has it except for yaqube.

Any real news about it, like a production run?

Mike purchased a slew of 060's and made a test fixture to make sure that they are not missing anything (common with knock offs from China). So, I would assume that Mike didn't do this to sit on them.

I have a developer daughter board myself, which is different from the 060 daughter board. Mike has a real job designing hardware and traveling all over the world sourcing components. With Wolfgang helping him with the ARM code and hardware layers, the progress has been speedy by comparison to Mike just tinkering over the years prior. Mike has hundreds of boards just sitting around doing nothing. That's a lot of money tied up that he would like to get back, and continue to profit from in the future with the production of new boards. The design is solid, and even though it is older, the FPGA is still large even by today's standards, so it's not outdated by any means.