Initial 3D support question
Hi, I have just read about the initial 3D support for RS690 at http://airlied.livejournal.com/56192.html so now I am a bit confused. I though the open source drivers were developed by Novell (you). Is Airlied's work a fork of your work or is this (your) work something different? I have just searched at Google for something about the 3D specs from AMD, but I couldn't find anything. Is the 3D support made without the 3D specs from AMD? Lots of love, Louise -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, Jan 28, 2008 at 04:15:17AM +0100, Louise Hoffman wrote:
Hi,
I have just read about the initial 3D support for RS690 at http://airlied.livejournal.com/56192.html
so now I am a bit confused. I though the open source drivers were developed by Novell (you).
Is Airlied's work a fork of your work or is this (your) work something different?
What do you expect me to say here? The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you. What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims. Luc Verhaegen. SUSE/Novell X Driver Developer. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, Jan 28, 2008 at 06:13:50AM +0100, Luc Verhaegen wrote:
On Mon, Jan 28, 2008 at 04:15:17AM +0100, Louise Hoffman wrote:
Hi,
I have just read about the initial 3D support for RS690 at http://airlied.livejournal.com/56192.html
so now I am a bit confused. I though the open source drivers were developed by Novell (you).
Is Airlied's work a fork of your work or is this (your) work something different?
What do you expect me to say here?
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation.
There has not been a single patch from either author to the radeonhd project, not before and not after -radeon work was started.
What label you want to stick to this is up to you.
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
Luc Verhaegen. SUSE/Novell X Driver Developer.
Luc Verhaegen. SUSE/Novell X Driver Developer. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, 2008-01-28 at 06:13 +0100, Luc Verhaegen wrote:
On Mon, Jan 28, 2008 at 04:15:17AM +0100, Louise Hoffman wrote:
Hi,
I have just read about the initial 3D support for RS690 at http://airlied.livejournal.com/56192.html
so now I am a bit confused. I though the open source drivers were developed by Novell (you).
Is Airlied's work a fork of your work or is this (your) work something different?
What do you expect me to say here?
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you.
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
So, if I understand correctly, nowaydays we have two competing drivers: - the shiny new radeonhd by AMD & Novell - the trusty old radeon by ATI & RedHat We're heading towards very interesting times ... Xav -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, Jan 28, 2008 at 11:21:32AM +0100, Xavier Bestel wrote:
On Mon, 2008-01-28 at 06:13 +0100, Luc Verhaegen wrote: So, if I understand correctly, nowaydays we have two competing drivers: - the shiny new radeonhd by AMD & Novell - the trusty old radeon by ATI & RedHat
Not exactly. I think a more appropriate wording would be that we have: - the trusty 'old' radeonhd by AMD & Novell - the 'shiny' new r5xx&r6xx code in the radeon driver that Dave Airlie and Alex Deucher are developing in their spare time. Paints a slightly different picture, don't you think? Sytse Wielinga -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, Jan 28, 2008 at 01:12:34PM +0100, Sytse Wielinga wrote:
On Mon, Jan 28, 2008 at 11:21:32AM +0100, Xavier Bestel wrote:
On Mon, 2008-01-28 at 06:13 +0100, Luc Verhaegen wrote: So, if I understand correctly, nowaydays we have two competing drivers: - the shiny new radeonhd by AMD & Novell - the trusty old radeon by ATI & RedHat
Not exactly. I think a more appropriate wording would be that we have:
- the trusty 'old' radeonhd by AMD & Novell - the 'shiny' new r5xx&r6xx code in the radeon driver that Dave Airlie and Alex Deucher are developing in their spare time.
Paints a slightly different picture, don't you think?
Sytse Wielinga
Shiny doesn't reflect the code or the tricks that need to be applied to keep on working against atombios, but whatever. Luc Verhaegen. SUSE/Novell X Driver Developer. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you.
So the R5xx and R6xx drivers are made without the specs from AMD?
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
As a end user, what does than mean? Will Red Hat users use "their" driver and Suse use their, or does the two projects focus on different things? -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, 28 Jan 2008 14:23:06 +0100 "Louise Hoffman" <louise.hoffman@gmail.com> wrote:
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you.
So the R5xx and R6xx drivers are made without the specs from AMD?
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
As a end user, what does than mean? Will Red Hat users use "their" driver and Suse use their, or does the two projects focus on different things?
If I'm not mistaken, the work from Redhat provides updates to the X driver package named "xserver-xorg-video-ati" (at least that's what it is called under Debian), while the work from Suse provides the "xserver-xorg-video-radeonhd" package. So, in an updated linux system, you should be able to choose any of the above, doesn't matter which distribution you use, they are both part of the X.org system. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, Jan 28, 2008 at 02:23:06PM +0100, Louise Hoffman wrote:
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you.
So the R5xx and R6xx drivers are made without the specs from AMD?
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
As a end user, what does than mean? Will Red Hat users use "their" driver and Suse use their,
This apparently was the idea already in oktober, there's mails to that end on their mailinglists.
or does the two projects focus on different things?
They do focus on different things. Radeonhd vastly expands modesetting (like my unichrome code did before). And radeonhd focusses on decent driver development practices which do lead to better maintainability and easier debugging. We want our code to be dependable, and to last. Luc Verhaegen. SUSE/Novell X Driver Developer. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Hi Louise David Airlie was told not to work on ATI drivers due to an NDA conflict, now RedHat & AMD have agreed to let him continue to work on ATI Drivers, you can ask both David & Alex this yourself on the xorg mailing list. (also IIRC -radeon was merged with -ati driver in xorg) I didn't know Luc that this was the case, but what code could you possibly want to integrate from -ati to -radeonhd ? -JoJo On Jan 28, 2008 6:53 PM, Louise Hoffman <louise.hoffman@gmail.com> wrote:
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you.
So the R5xx and R6xx drivers are made without the specs from AMD?
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
As a end user, what does than mean? Will Red Hat users use "their" driver and Suse use their, or does the two projects focus on different things?
-- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
-- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Mon, Jan 28, 2008 at 07:58:59PM +0530, JoJo jojo wrote:
Hi Louise
David Airlie was told not to work on ATI drivers due to an NDA conflict, now RedHat & AMD have agreed to let him continue to work on ATI Drivers, you can ask both David & Alex this yourself on the xorg mailing list.
(also IIRC -radeon was merged with -ati driver in xorg)
I didn't know Luc that this was the case, but what code could you possibly want to integrate from -ati to -radeonhd ?
-JoJo
I am working on integrating r5xx 2D acceleration and drm support from -ati into radeonhd. Doing this in a clean and modular fashion is proving more work than originally expected, but the result is starting to look real nice and really hackable. Luc Verhaegen. SUSE/Novell X Driver Developer. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
I am working on integrating r5xx 2D acceleration and drm support from -ati into radeonhd. Doing this in a clean and modular fashion is proving more work than originally expected, but the result is starting to look real nice and really hackable.
I am impressed that you guys can figure out how to write hardware drivers... Even with the specs! Again, I read that the 2 modesettings specs were just taken off the shelf and put online, and that was pretty much all ATI had of written specs. All future specs would AMD have ATI engineers to write now, as it wasn't documented. If that's true, who would have though that's how hardware vendors work? =) -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
David Airlie was told not to work on ATI drivers due to an NDA conflict, now RedHat & AMD have agreed to let him continue to work on ATI Drivers, you can ask both David & Alex this yourself on the xorg mailing list.
I read (I think it was in one of the early kerneltrap articles) that AMD would not allow any developer that had worked on the closed source drivers to work on the open source drivers in order to avoid possible patent infringements clames. Is this the case here, or is that classified? -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Jan 28, 2008 8:23 AM, Louise Hoffman <louise.hoffman@gmail.com> wrote:
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you.
So the R5xx and R6xx drivers are made without the specs from AMD?
The code was written based on the ATOM parser from AMD and the wrapper from radeonhd.
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
As a end user, what does than mean? Will Red Hat users use "their" driver and Suse use their, or does the two projects focus on different things?
The two projects had different goals. Dave and I wanted to play with ATOMBIOS and to take advantage of the existing functionality that is already supported in the radeon driver (2D/3D accel, drm/command process support, etc.). ATOMBIOS is a set of card specific tables and scripts that can be used to program certain functions on the chip. This includes setting modes, routing crtcs, dpms, initializing the chip, etc. ATOMBIOS provides a common API to perform these functions. The x86 bios, the windows driver, fglrx, etc., use these scripts to program the card. This makes it much easier to bring up new asics and program the chips. ATOM allowed Dave and I to get the driver up quickly so we could start to play with things like 2D and 3D, using the command processor, etc. since much of that works unchanged on newer chips. Down the road I'd like to see one code base, but it will take some work and time of course. We'll get there eventually. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
As the license is the same, I think it's obvious that these two projects should merge at some point, if there isn't a reason not to do so. Isn't it?. 2008/1/28, Alex Deucher <alexdeucher@gmail.com>:
On Jan 28, 2008 8:23 AM, Louise Hoffman <louise.hoffman@gmail.com> wrote:
The R5xx and R6xx support in radeon is purely a private project of two people, redhat's Dave Airlie and ATIs Alex Deucher. This started in november, and copied a lot of radeonhd code. None of this happened in an amicable fashion, and there is no cooperation. What label you want to stick to this is up to you.
So the R5xx and R6xx drivers are made without the specs from AMD?
The code was written based on the ATOM parser from AMD and the wrapper from radeonhd.
What i can tell you is that AMD and SUSE are continuing their close cooperating on the radeonhd driver, on providing providing support for new hardware and on further features and functionality. We are providing the best technical solution, for a driver that will be solid and long lasting, requiring a minimal maintainance burden in the long term. We do not cut corners to be able to make big claims.
As a end user, what does than mean? Will Red Hat users use "their" driver and Suse use their, or does the two projects focus on different things?
The two projects had different goals. Dave and I wanted to play with ATOMBIOS and to take advantage of the existing functionality that is already supported in the radeon driver (2D/3D accel, drm/command process support, etc.). ATOMBIOS is a set of card specific tables and scripts that can be used to program certain functions on the chip. This includes setting modes, routing crtcs, dpms, initializing the chip, etc. ATOMBIOS provides a common API to perform these functions. The x86 bios, the windows driver, fglrx, etc., use these scripts to program the card. This makes it much easier to bring up new asics and program the chips. ATOM allowed Dave and I to get the driver up quickly so we could start to play with things like 2D and 3D, using the command processor, etc. since much of that works unchanged on newer chips. Down the road I'd like to see one code base, but it will take some work and time of course. We'll get there eventually.
Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
-- Néstor Amigo Cairo +34 687 96 74 81 nestorac@gmail.com -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
The two projects had different goals. Dave and I wanted to play with ATOMBIOS and to take advantage of the existing functionality that is already supported in the radeon driver (2D/3D accel, drm/command process support, etc.). ATOMBIOS is a set of card specific tables and scripts that can be used to program certain functions on the chip. This includes setting modes, routing crtcs, dpms, initializing the chip, etc. ATOMBIOS provides a common API to perform these functions. The x86 bios, the windows driver, fglrx, etc., use these scripts to program the card. This makes it much easier to bring up new asics and program the chips. ATOM allowed Dave and I to get the driver up quickly so we could start to play with things like 2D and 3D, using the command processor, etc. since much of that works unchanged on newer chips. Down the road I'd like to see one code base, but it will take some work and time of course. We'll get there eventually.
Thanks for the explanation =) I didn't really understood what ATOMBIOS was, bit now I just think of it as an API to make driver developers work easier =) -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
The downside of the AtomParser approach (as far as my very limited knowledge lasts) is that the code in the AtomParser is i386 depending, so there is no easy way to run this card on e.g. PPC or any other architecture. (But there are graphic card bioses downloadable from ATI for some cards for the PPC architecture and I heard some other implementations try to run the atombios code with an i386 simulator). LieGrü, strub --- Louise Hoffman <louise.hoffman@gmail.com> schrieb:
The two projects had different goals. Dave and I wanted to play with ATOMBIOS and to take advantage of the existing functionality that is already supported in the radeon driver (2D/3D accel, drm/command process support, etc.). ATOMBIOS is a set of card specific tables and scripts that can be used to program certain functions on the chip. This includes setting modes, routing crtcs, dpms, initializing the chip, etc. ATOMBIOS provides a common API to perform these functions. The x86 bios, the windows driver, fglrx, etc., use these scripts to program the card. This makes it much easier to bring up new asics and program the chips. ATOM allowed Dave and I to get the driver up quickly so we could start to play with things like 2D and 3D, using the command processor, etc. since much of that works unchanged on newer chips. Down the road I'd like to see one code base, but it will take some work and time of course. We'll get there eventually.
Thanks for the explanation =) I didn't really understood what ATOMBIOS was, bit now I just think of it as an API to make driver developers work easier =) -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
__________________________________ Ihre erste Baustelle? Wissenswertes für Bastler und Hobby Handwerker. www.yahoo.de/clever -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Tue, 29 Jan 2008 15:00:30 +0100 (CET) Mark Struberg <struberg@yahoo.de> wrote:
The downside of the AtomParser approach (as far as my very limited knowledge lasts) is that the code in the AtomParser is i386 depending, so there is no easy way to run this card on e.g. PPC or any other architecture. (But there are graphic card bioses downloadable from ATI for some cards for the PPC architecture and I heard some other implementations try to run the atombios code with an i386 simulator).
LieGrü, strub
Well it's exactly the opposite. ATOM are script which don't have anythings to do with X86 opcode making execution of this kind of things really easy on others architecture like PPC. Cheers, Jerome Glisse -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
participants (10)
-
Alex Deucher -
Dimitris Lampridis -
Jerome Glisse -
JoJo jojo -
Louise Hoffman -
Luc Verhaegen -
Mark Struberg -
Néstor Amigo Cairo -
Sytse Wielinga -
Xavier Bestel