@amigakit

Quote:

amigakit wrote: @Karlos As I have mentioned in this thread several times before, there is no need to replace the main warp3d.library as has been promised in this bounty. A conventional Warp3D driver can be written. A-EON can work through the technical details and offer support if contact is made.

Respectfully I must disagree. There is every reason *not* to create a conventional driver. There's literally no part of the existing source that's of any value to a new driver. There's no need for the level of abstraction the legacy architecture comes with. Nobody needs the "support multiple GPUs on P96 and CGX concurrently" that probably never worked properly, nor the multiple trampoline from one library to another to do basic things. Developers don't need the headache of tracing difficult bugs through multiple libraries that have zero functional tests. Users don't need the hassle of multiple driver components. It's an absolute anachronism.

And frankly, the replacement doesn't need to be part of a closed source solution either.

_________________
Doing stupid things for fun...

@Hans

We've both worked on the W3D code. You had to actively bypass parts of it, I had to change part of the actual P96 driver for the Permedia. *

On paper the modular design is fine but honestly name any part of the Warp3D library stack that needs to be a separate library, that couldn't be achieved by compile time abstraction to maintain the separation of concern.

The solution as originally proposed is ideal. It doesn't need a 10+ year ball and chain.

Just so nobody here is in any doubt about my thinking on this: The *only* reason to make this a "proper" driver is so that someone can assert their ownership of it. There are no technical reasons for it and several against it.

And if anyone is still thinking "it's cool either way as long as I get a driver for my videocore", just reflect on the fact that without open source, there'd be no Emu68 or PiStorm to use it on in the first place.

If AEON want to use any of the new code in their own closed drivers for the 600GS and 1200, well, hypothetically there's no reason they couldn't provided the right license is used. MIT is the way to go here.

Everyone wins.

-edit-

@Hans

You also mentioned how easy it was to create two new drivers from scratch, but all that does is underscore the point that if it's so simple and nobody has done it for over a decade for 3.x/68K when there were people crying out for radeon R200 drivers (I had people ask me directly if I could do it) then it's clear there was no desire to do it at all.

And now the moment someone else decides to follow an already well-trodden path of complete greenfield replacement, they are suddenly expected to negotiate a "proper" solution from the IP holders? If they want to support the userbase, why not just open source legacy Warp3D?

Last edited by Karlos on 20-Jan-2025 at 11:07 AM.
Last edited by Karlos on 20-Jan-2025 at 10:30 AM.
Last edited by Karlos on 20-Jan-2025 at 10:13 AM.

_________________
Doing stupid things for fun...

@Karlos

Quote:

We've both worked on the W3D code. You had to actively bypass parts of it, I had to change part of the actual P96 driver for the Permedia.

I didn't have to bypass it. I just chose not to use it because I had a better solution.The Warp3D GFX driver subsystem was a hack to work around having multiple RTG systems, none of which had any concept of sharing the GPU with multiple drivers.

All I did was not use it. The lock and unlock operations became no-ops, and my drivers handled the rest via its own pipeline and the render manager.

We did add LockBitMapToBoard()/UnlockBitMapFromBoard() functions to Picasso96, so that bitmaps could be locked in place while the GPU was using them.

Quote:

On paper the modular design is fine but honestly name any part of the Warp3D library stack that needs to be a separate library, that couldn't be achieved by compile time abstraction to maintain the separation of concern.

Providing a common API to find and load the appropriate driver(s) is about the only role that the main library needs to do. The older v4 Warp3D might also do some draw op queuing for "indirect mode" (which was removed in v5), but that's about it.

Quote:

The solution as originally proposed is ideal. It doesn't need a 10+ year ball and chain. Just so nobody here is in any doubt about my thinking on this: The *only* reason to make this a "proper" driver is so that someone can assert their ownership of it. There are no technical reasons for it and several against it.

I think the main reason to make a proper driver, is because it's a nicer way to do things instead of repeatedly replacing the main library. Reimplementing libraries is a symptom of being unable to work together, which is driven in part by Driver Development Kits (DDKs) being closed, and requiring special permission to access.

You could also do it for the 1 (or 2) people who will stick a second graphics card in their machine and want to use Warp3D with that, and the PiStorm GPU.

Unlike you, I see no technical reasons against it.

Quote:

And if anyone is still thinking "it's cool either way as long as I get a driver for my videocore", just reflect on the fact that without open source, there'd be no Emu68 or PiStorm to use it on in the first place.

AFAIK, there's no reason why an actual Warp3D driver couldn't be open-source. Of course, it would suck if it could only be compiled by those with access to the DDK. If it were up to me, all AmigaOS Driver Development Kits (DDKs) would be opened up, and made publicly available.

Hans

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Hans

Quote:

I didn't have to bypass it. I just chose not to use it because I had a better solution.

Precisely! And that is what a greenfield total replacement is. A better solution. Thank you for making my point for me.

Again, the responsibilities of the different libraries can be handled by compile time abstraction. I don't by the benefit of the many libraries versus one monolith. Massive projects like the Linux kernel get away just fine with thousands of compile time abstractions. This is miniscule in comparison.

You don't see people doing this with other libraries, do you? Replacing the IEEE*.library with a stub that then has a software, 68881, 68882, 68040, 68060 sub library driver. Why? Because it's pointless indirection. You just have a version of the library directly compiled for your CPU. Well, you could say it's too performance critical and I would agree. But I see no reason why hardware rasterization should be any more indirect than strictly necessary, either. Rendering is also performance critical.

If you are a 68K classic user, you swap your GPU about as often as you swap your CPU, maybe even less, so why have all that extra clutter?

Last edited by Karlos on 20-Jan-2025 at 11:58 AM.
Last edited by Karlos on 20-Jan-2025 at 11:21 AM.
Last edited by Karlos on 20-Jan-2025 at 11:14 AM.

_________________
Doing stupid things for fun...

@Karlos

Quote:

You also mentioned how easy it was to create two new drivers from scratch, but all that does is underscore the point that if it's so simple and nobody has done it for over a decade for 3.x/68K when there were people crying out for radeon R200 drivers (I had people ask me directly if I could do it) then it's clear there was no desire to do it at all.

Not easy, because graphics drivers are a pain in the butt to write. I just see zero technical reasons against writing a proper driver.

Quote:

And now the moment someone else decides to follow an already well-trodden path of complete greenfield replacement, they are suddenly expected to negotiate a "proper" solution from the IP holders? If they want to support the userbase, why not just open source legacy Warp3D?

AFAICT, A-EON would appreciate the courtesy of anyone using their IP (including copyrighted names) talking with them, instead of going straight to the reimplementation route (yet again).

That, and any reimplementation should be given a different name, because the name Warp3D shouldn't be used without the copyright owner's permission.

Hans

Last edited by Hans on 20-Jan-2025 at 11:23 AM.

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Karlos

Quote:

Precisely! And that is what a greenfield total replacement is. A better solution. Thank you for making my point for me.

There's no harm in talking to A-EON.

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Hans

Quote:

AFAICT, as IP owner A-EON would appreciate the courtesy of anyone using their IP (including copyrighted names) talking with them, instead of going straight to the reimplementation route (yet again).

This is a straw man point. Who is suggesting using their IP *at all*, let alone without approaching them first? I don't think the campaign could be any clearer about it. We've already had confirmation in this thread that there's zero reason why a greenfield reimplementation is an infringement and nobody with past exposure to the driver code will be working on it.

Every mention of the IP here is just muddying the water for no good reason.

Perthaps one of the moderators can adjust the thread title and put Warp3D in quotes, or add a TM or something if it would help

Last edited by Karlos on 20-Jan-2025 at 12:03 PM.
Last edited by Karlos on 20-Jan-2025 at 11:29 AM.

_________________
Doing stupid things for fun...

@Hans

Quote:

Hans wrote: @Karlos Quote: Precisely! And that is what a greenfield total replacement is. A better solution. Thank you for making my point for me. There's no harm in talking to A-EON.

Actually I'd say you *must* talk to them if your intention is to create a Warp3D hardware driver. 100%, no arguments. They are the IP holders, they presuably have the driver NDK under NDA you need to sign, etc. Standard stuff, if a bit yesteryear.

Beyond the initial confusion the thread may have caused by referring the name, now that it's clear that there's no intention of using the name or any part of the IP, I don't really see the problem. Everyone seems to be in agreement that there's no problem with what is actually being proposed.

Anyway, none of this is up to me and I am sure that everyone involved wants to proceed best foot forwards.

_________________
Doing stupid things for fun...

@Karlos

Basically "Warp3D for PiStorm" was just a "catchphrase". Not a "Product name for release". With such an old and untended product (which also has been reimplemented several times, Wazp3D, GOA, the thing Apollo does,...) I had not expected this to be such a big issue ?

And yes, I adapted the text now, and it is now called "3D for PiStorm" (with some explanations of the possible solutions).

What reason to not do a driver ?

Well, the attitude Amigakit showed in the discussion here is a HUUUUUUUGE reason for example. We don't want to be dependent of that. Also makes it easier if something would need changing in the core libs... you never know what comes up... and let's say Amigakit has a certain reputation. So I thought from the start it is better to make it clear this is a reimplementation which does not affect their IP...

And yes, if they would play nicely actual cooperation could be done, especially if at one point also a version for Pi3 would be done (it is OpenSource and Amigakit could use it then as well to support their own hardware - maybe even get hints and stuff when running into problem with the adaption - of course this would require them to play nicely and stuff the approach they currently do).

I mean they want to bring out a new A1200 Board based on Pi. Are they NOT interested in getting further information which might get them a 3DHW Implementation ? And possibly further games to run on it ? Is this worthless to them ?

At the risk of making myself an even bigger bumhole than I already am, it seems you can't copyright a name or phrase under UK law at least and there doesn't seem to be any registered trademarks in the UK, EU or US with "Warp3D" or "Warp 3D" in them:

https://www.gov.uk/search-for-trademark - sorry you'll have to type it in yourself. British engineering FTW

https://tmsearch.uspto.gov/search/search-information - also for the US


https://www.tmdn.org/tmview/#/tmview/results?page=1&pageSize=30&criteria=C&basicSearch=Warp3D

This doesn't change anything since the name isn't being used but it does also make me wonder what the fuss is about.

_________________
Doing stupid things for fun...

@MagicSN

Quote:

Well, the attitude Amigakit showed in the discussion here is a HUUUUUUUGE reason for example

What "attitude"- are you reading any of the posts I have made? Are you pushing a false narrative to try an inflame posts ? Read my comments again. All my posts have been supportive of Pistorm project and the desire to have a free of charge Warp3D compatible driver for Pistorm users. A-EON and AmigaKit hold the Pistorm project in high esteem. We have no interest in monetisation or charging royalties.

The bottom line is that you advertised your intention to replace the main Warp3D Library without consulting A-EON Technology Ltd. This is not the best technical method as mentioned by Hans De Ruiter in this thread. He is the foremost technical expert when it comes to Warp3D development.

Last edited by amigakit on 20-Jan-2025 at 01:10 PM.

_________________
Amiga Kit Amiga Store
Links: www.amigakit.com | New Products | A600GS

@amigakit

Quote:

The point is that a Warp3D project was announced that proposes replacing the warp3d.library without consulting A-EON first.

Aren’t you yourself the ones replacing just about the entire OS without consulting “the owner� first?

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

I really need to disengage but...

@amigakit

Quote:

amigakit wrote: @MagicSN The bottom line is that you advertised you are intention to replace the main Warp3D Library without consulting A-EON Technology Ltd.

Which under the circumstances of greenfield you already agreed is not necessary. Though I don't dispute it would be a nice courtesy if nothing else. It's irrelevant now because the cat is out ot the bag anyway. It's an argument point for it's own sake.

Quote:

This is not the best technical method as mentioned by Hans De Ruiter in this thread. He is the foremost technical expert when it comes to Warp3D development.

This is an argument to authority but nevertheless he has stated his opinions here and as always they're very sensible. But the technical bottom line is:

Hans:
Quote:

I think the main reason to make a proper driver, is because it's a nicer way to do things instead of repeatedly replacing the main library.

Note - subjective opinion and nothing specifically to do with the legacy framework being better in any way, just that having many potential main libaries is a potential mess. I would counter this by making a few very basic observations:

1. There's only one such main library being discussed.
2. Should there be more in future, it's mostly a one time installation problem. Detect the hardware, choose the variant and copy it to libs. Libraries with different CPU variants have done this with installer scripts for decades.
3. Even without an installer, your average highly expanded 68K classic user isn't going to be confused by sensibly named files in an archive and the instruction to put the right one into libs: and rename it. I mean they've been using the bloody thing for decades by now. I think they can be trusted.

Last edited by Karlos on 20-Jan-2025 at 01:23 PM.
Last edited by Karlos on 20-Jan-2025 at 01:20 PM.

_________________
Doing stupid things for fun...

@amigakit

Clean Room implementations are perfectly legal as far as I know

AROS f.e., is based on that

@Hans

Quote:

Warp3D shouldn't be used without the copyright owner's permission

You mean trademark.

If the string “Warp3D� is copyright protected, we are all in deep shit on this tread.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@MagicSN

Just because it appears as if you are not aware… AmigaKit are (ironically) not using (British) Raspberry Pi in their A600GS and A1200NG, they are using (Chinese) Orange Pi 3, quite different hardware than Raspberry Pi 3.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

Quote:

kolla wrote: @Hans Quote: Warp3D shouldn't be used without the copyright owner's permission You mean trademark. If the string “Warp3D� is copyright protected, we are all in deep shit on this tread.

I refer you to my last post. There doesn't appear to be a registered trademark, at least in the UK/EU/US.

I'm no legal expert but guess this means there's no "name" to even defend. Not that I propose using it as it would only cause unnecessary confusion.

Besides, thinking up alternative names is too much fun. What about IllegalRave3D ?

_________________
Doing stupid things for fun...

@Karlos

on wazp3d it is solved by using wazp3d and advise the user to rename it to warp3d

@Hans

is the official warp3d library planned to be ported to 68k, be it PiStorm or V4? Standard 68k only can run it in software mode so I do not know if it makes any sense.

On 68k I only know of warp3d/stormmesa from H&P and later improved version rfom thellier (wazp3d)

@OlafS25

Quite. To be honest though, there doesn't appear to be anything enforceable about the name because it's not a registered trademark in any of the places you might expect to find a significant userbase.

However it would be unwise to use the name at all since the proposal is not part of Warp3D and there's no need to create any more friction/confusion than we've seen so far. AEON own the existing IP and that should be respected.

An open source (preferably free software) *replacement* is everyone's. We're getting too old for all the cloak and dagger. All our closed source stuff will die with us.

_________________
Doing stupid things for fun...