| |
Iron Dignity |
| |
Hertzsprung (09:55 4/2/2003)
Phlamethrower (10:07 4/2/2003)
Hertzsprung (10:11 4/2/2003)
Phlamethrower (10:19 4/2/2003)
Hertzsprung (10:39 4/2/2003)
monkeyson2 (10:59 4/2/2003)
Hertzsprung (14:36 4/2/2003)
moss (15:12 4/2/2003)
Phlamethrower (17:02 4/2/2003)
Loris (17:26 4/2/2003)
ksattic (21:46 4/2/2003)
Phlamethrower (22:04 4/2/2003)
ksattic (22:37 4/2/2003)
Phlamethrower (23:02 4/2/2003)
johnstlr (10:18 5/2/2003)
Hertzsprung (09:38 6/2/2003)
andrew (13:17 6/2/2003)
moss (15:15 6/2/2003)
johnstlr (16:42 6/2/2003)
ksattic (17:05 6/2/2003)
Phlamethrower (17:15 6/2/2003)
ksattic (17:18 6/2/2003)
Phlamethrower (17:24 6/2/2003)
Phlamethrower (17:34 6/2/2003)
Phlamethrower (17:46 6/2/2003)
ksattic (18:51 6/2/2003)
|
| |
|
James Shaw |
Message #30870, posted by Hertzsprung at 09:55, 4/2/2003 |
 Ghost-like
Posts: 1746
|
Saw it for the first time just the other day.....
I'm speechless  |
| |
 [ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #30872, posted by Phlamethrower at 10:07, 4/2/2003, in reply to message #30870 |
 Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
What? The old demo (1998 I think) or a more recent version? |
| |
[ Log in to reply ] |
| |
|
James Shaw |
Message #30873, posted by Hertzsprung at 10:11, 4/2/2003, in reply to message #30872 |
Ghost-like
Posts: 1746
|
What? The old demo (1998 I think) or a more recent version? Yes, the only demo (afaik)  |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #30874, posted by Phlamethrower at 10:19, 4/2/2003, in reply to message #30873 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
Aye 
*wonders how well aemulor could run it...*  |
| |
[ Log in to reply ] |
| |
|
James Shaw |
Message #30875, posted by Hertzsprung at 10:39, 4/2/2003, in reply to message #30874 |
Ghost-like
Posts: 1746
|
Not that well, I fear.
You up for a bit of 32bitting,  ? |
| |
[ Log in to reply ] |
| |
|
Phil Mellor |
Message #30879, posted by monkeyson2 at 10:59, 4/2/2003, in reply to message #30875 |
 Please don't let them make me be a monkey butler
Posts: 12380
|
Make sure you finish the rest of the game off while you're at it.  |
| |
[ Log in to reply ] |
| |
|
James Shaw |
Message #30907, posted by Hertzsprung at 14:36, 4/2/2003, in reply to message #30879 |
Ghost-like
Posts: 1746
|
Make sure you finish the rest of the game off while you're at it. Yeahright. |
| |
[ Log in to reply ] |
| |
|
John Hoare |
Message #30913, posted by moss at 15:12, 4/2/2003, in reply to message #30907 |
Posts: 9348
|
Surely the Iyonix would be capable for a full version of ID?
Not that I can see it being made |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #30919, posted by Phlamethrower at 17:02, 4/2/2003, in reply to message #30913 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
Yes it would be capable, but it's questionable how good it will be - a game the same quality as the demo would be fantastic, but the PC version is likely to be a lot better and so they'd need access to the geforce's grunt to make it run well on an Iyonix... |
| |
[ Log in to reply ] |
| |
|
Tony Haines |
Message #30930, posted by Loris at 17:26, 4/2/2003, in reply to message #30873 |
 Ha ha, me mine, mwahahahaha
Posts: 1025
|
What? The old demo (1998 I think) or a more recent version? Yes, the only demo (afaik)
There are at least 2 (old) demos.
I know because I have 2. |
| |
[ Log in to reply ] |
| |
|
Simon Wilson |
Message #30948, posted by ksattic at 21:46, 4/2/2003, in reply to message #30930 |
 Finally, an avatar!
Posts: 1291
|
I'm hoping that Castle will produce an OpenGL driver for the Iyonix. Perhaps it would be better to go down the route MS took and create their own 3D api, tuning it for performance.
Either way, I hope we see hardware 3D graphics on RISC OS soon. |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #30949, posted by Phlamethrower at 22:04, 4/2/2003, in reply to message #30948 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
I don't think producing an API is a problem; the problem is that only nVidia know how to drive the 3D hardware on their cards
If 3dfx were still around everything would be alright, because the source code to their drivers has been available for years - in fact, if someone knows of a few thousand Voodoo cards we can have then our problems will be solved |
| |
[ Log in to reply ] |
| |
|
Simon Wilson |
Message #30950, posted by ksattic at 22:37, 4/2/2003, in reply to message #30949 |
Finally, an avatar!
Posts: 1291
|
Shame that nVidia are keeping their hardware details secret. I realise the toughest part would be writing a 3D driver - on top of that you could probably have whatever API you wanted.
Are the Linux people in the same boat here? I know nVidia provide Linux drivers for some of their cards, but do the drivers support 3D and can they be reverse-engineered?
I feel some research coming on tonight... |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #30951, posted by Phlamethrower at 23:02, 4/2/2003, in reply to message #30950 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
Yes, the linux people are in the same boat; all nVidia provide is the precompiled core of their drivers (with full 3D support) and the source to the interfacing code to make it fit in with whatever version of Linux they're using. They can be reverse engineered, and apparently there are people who've done it, but nVidia make it quite clear that decompiling/reverse engineering/etc isn't allowed so if we were to do it then Castle certainly wouldn't support us.
As usual I've been collecting lots of insane ideas to solve the problem - so far I've got:
1. Decompile the code and translate it, but as I've said this is illegal
2. Port a PC emulator and have the nVidia drivers running under emulation (With interfacing code to let RISC OS access them of course). For this to work at a decent speed JIT would be required.
3. Leading on from the above, have a direct conversion of the Linux code without any decompilation inbetween - i.e. we put in the Linux drivers and get out RISC OS ones without us or anyone else finding out how they work. Technically this should be legal, but I've had a look at Intel's docs on the IA32 (aka x86) instruction set and it's massive and convoluted compared to the ARM set.
4. Also related to 2., if a PC card becomes available then people could install Linux on it and have the drivers accessible through RISC OS. This could also be extended to provide hardware floating point as well, much like a program did for the Risc PC's PC card.
If enough effort was put into it then 2-4 should be possible, 3. being best since it would provide the highest speed. Well, 4 may be faster, but people would have to spend £££
Plus, if the above don't tickle you're fancy, then there are also the following:
5. Brute force approach of sending data to the card until something interesting happens
6. Write a logging program/build some hardware to log what gets sent down the bus when the real drivers interact with the card
As you can see I've spent quite some time thinking about this
grr frustrating grr grr etc.
Also on a side note there is a project to produce an open hardware 3D card - http://www.icculus.org/manticore/ - but it's only in a very early stage
[Edited by Phlamethrower at 23:07, 4/2/2003] |
| |
[ Log in to reply ] |
| |
|
Lee Johnston |
Message #30957, posted by johnstlr at 10:18, 5/2/2003, in reply to message #30948 |
Member
Posts: 193
|
I'm hoping that Castle will produce an OpenGL driver for the Iyonix. Perhaps it would be better to go down the route MS took and create their own 3D api, tuning it for performance. MS didn't go down the DirectX route purely for performance reasons. They wanted to own the API. Furthermore they wanted to progress the API at a faster rate than the OpenGL ARB can achieve. Ok the first few versions of D3D were horrendeous but there's no denying that the standard API (ie no vendor specific extensions) is now a more capable API than OpenGL.
Mind you I'd prefer the OpenGL route simply because it isn't owned by MS and is available on more platforms. |
| |
[ Log in to reply ] |
| |
|
James Shaw |
Message #31058, posted by Hertzsprung at 09:38, 6/2/2003, in reply to message #30930 |
Ghost-like
Posts: 1746
|
What? The old demo (1998 I think) or a more recent version? Yes, the only demo (afaik)
There are at least 2 (old) demos.
I know because I have 2.
What's the difference between them? |
| |
[ Log in to reply ] |
| |
|
Andrew |
Message #31088, posted by andrew at 13:17, 6/2/2003, in reply to message #31058 |
 Handbag Boi
Posts: 3439
|
Take a look at 32-bit gaming : www.acorn-gaming.org.uk |
| |
[ Log in to reply ] |
| |
|
John Hoare |
Message #31110, posted by moss at 15:15, 6/2/2003, in reply to message #31088 |
Posts: 9348
|
*slaps nVidia about a bit for closed-sourceness* 
Are there *any* cards about that have open source drivers? You'd think it would *benefit* them; their cards would be used on as many machines as possible, and they'd get patches to fix bugs from the hacker community so their product would be far better, with a lot less effort from them.
Yes, I have read The Cathedral and the Bazzarr, what's your point? |
| |
[ Log in to reply ] |
| |
|
Lee Johnston |
Message #31120, posted by johnstlr at 16:42, 6/2/2003, in reply to message #31110 |
Member
Posts: 193
|
*slaps nVidia about a bit for closed-sourceness*
Are there *any* cards about that have open source drivers? You'd think it would *benefit* them; their cards would be used on as many machines as possible, and they'd get patches to fix bugs from the hacker community so their product would be far better, with a lot less effort from them.
Yes, I have read The Cathedral and the Bazzarr, what's your point? I've already given some explanations for this on coderscauldron but for the sake of you guys.... (I'm such a martyr eh? )
Basically drivers are tied directly to the hardware. nVidia et al are "concerned" that by releasing their driver code their competitors may be able to learn about how their hardware works or gain some other competitive edge.
Also bear in mind that even the quality of the drivers can act in nVidia's favour. It's well known that John Carmack tends to support nVidia features in new software first because they tend to be more on the ball with driver releases. Imagine it... "Doom III, primarily developed with nVidia technology" or "Doom III works best on nVidia graphics cards".
I realise that making them open source may mean people could submit bug fixes but I think such benefits are outweighed by the potential risk to the intellectual property involved.
Finally nVidia drivers are available for Windows, linux and Mac OS. From nVidia's perspective, what other platforms are there?
Now I'm sure people could refute all my points but, given that nVidia drivers are closed source, then I'd guess I'm closer to the truth than "idealistic open source fanatics" (and I mean that in jest!!) |
| |
[ Log in to reply ] |
| |
|
Simon Wilson |
Message #31128, posted by ksattic at 17:05, 6/2/2003, in reply to message #31120 |
Finally, an avatar!
Posts: 1291
|
Picking up on what Jeffrey said, perhaps someone could build a PCI card to sniff what is sent across the bus to/from the NVidia PCI card. Of course, this would have to be done in a PC when a 3D app is running. Could software feasibly be written to do this?
Someone must have thought of a way to get around this problem. The Amiga people are in a similar situation with no knowledge of NVidia's hardware. |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #31131, posted by Phlamethrower at 17:15, 6/2/2003, in reply to message #31128 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
Could software feasibly be written to do this? You might be able to do it by writing a custom PCI library that sits between the Linux PCI code and the nVidia drivers; i.e. you would link the nVidia drivers against your PCI library (as opposed to the Linux C library), and then link that against the Linux library. Some symbol table hackery might be needed to stop name clashes, but if it works then your code should get called whenever the drivers try to access the hardware - allowing you to log what's going on, from the programmer's perspective.
Of course this will all fall apart if it works by writing directly into blocks of memory (Which I expect it does), unless of course you can find a way of having your code called soon after an update has been made.
I also think there's meant to be some multithreading at work, which could complicate matters a bit. |
| |
[ Log in to reply ] |
| |
|
Simon Wilson |
Message #31132, posted by ksattic at 17:18, 6/2/2003, in reply to message #31131 |
Finally, an avatar!
Posts: 1291
|
New idea: one of us gets a job at NVidia. |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #31133, posted by Phlamethrower at 17:24, 6/2/2003, in reply to message #31132 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
Or invent a microchip decompiler
*does a quick search*
Open source ATI drivers!
I'll look into this... |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #31135, posted by Phlamethrower at 17:34, 6/2/2003, in reply to message #31133 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
http://mirror.ati.com/support/faq/linux.html
3D Graphics Acceleration
ATI has made the necessary hardware and programming information available to Linux developers for the development of hardware 3D acceleration.
Family 3D Status
RADEON 9700 \ XFree86 drivers available on
RADEON 9000 +- the ATI website support hardware
RADEON 8500 / 3D acceleration
RADEON \- 3D drivers are available
RAGE 128 / from the DRI Project.
RAGE PRO -- 3D support is a work in progress
from the UTAH-GLX Project.
Accelerated-X products by XI Graphics, Inc. also offer hardware accelerated 3D using OpenGL. Accelerated-X is provided as a commercial alternative to the freeware XFree86.
*delves deeper*
[Edited by Phlamethrower at 00:38, 7/2/2003] |
| |
[ Log in to reply ] |
| |
|
Jeffrey Lee |
Message #31139, posted by Phlamethrower at 17:46, 6/2/2003, in reply to message #31135 |
Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff
Posts: 15100
|
There aren't any obvious card source code downloads available on the linked sites (And the ATI drivers only come in RPM form), but there is the ATI Developer site:
ATI Developer Relations offers you Software Development Kits.
Driver Development Kits, sample code, early driver releases and third party application testing. *wonders whether they'd consider supporting RISC OS*
and for good measure:
*uses a crowbar to beat Castle into a bloody pulp for using an nVidia card instead of an ATI one*
[Edited by Phlamethrower at 17:47, 6/2/2003] |
| |
[ Log in to reply ] |
| |
|
Simon Wilson |
Message #31150, posted by ksattic at 18:51, 6/2/2003, in reply to message #31139 |
Finally, an avatar!
Posts: 1291
|
I prefer ATI cards - faster 2D, and currently the fastest 3D performance if you disregard the monstrosity that NVidia just released (the FX?)
The NVidia cards are good for decent 3D on a budget, though - but what's the point if you can't have drivers? |
| |
[ Log in to reply ] |
| |
|
