The State Of Open-Source Radeon Driver Features

[ChrisXY]

b.)AMD is not endorsing or supporting or writing or financing or [fill whatver you want] r600g in any way

So what do these 5 people AMD has employed for the open source driver do all day? And why doesn't AMD pay them for the code they commit?

[jrch2k8]

So what do these 5 people AMD has employed for the open source driver do all day? And why doesn't AMD pay them for the code they commit?

not sure is 5 is the right numbers but as far as i know this fill the entry i writed before -- provide skeleton code/support and ofc they can contribute to other parts of the code that they see fit since is an OSS project but i think bridgman can provide a more precise answer of their jobs exactly

the fact AMD pay some devs to improve things is not different than MyXcompany. inc doing it or yourXcompany. inc doing it since the code they provide follow the project guidelines and is not managed or owned by AMD

ofc AMD maybe interested in help to bring faster certain features to r600g for internal reasons but it doesnt make their driver or property they are just another contributor

[Panix]

You cannot compare Nvidia and AMD open source efforts:
Nvidia -> go to prison if you write anything about the specs!
AMD -> here are all the specs, write a driver for whatever you need we only provide software for gaming platforms like Windows (same with motherboard chipsets etc. see coreboot)

I didn't even post in the other thread because I don't care what they support in their binary drivers.

Where did I make that comparison? I just contrasted how people act or more accurately, 'react' to what Nvidia and AMD/ATI do.

AMD/ATI doesn't release all the specs as I understand it. There is no support with hardware acceleration and various licensing info or whatever restrictions there are. Also, on the feature set at the Radeon Feature web page, you can note all the 'features' that are listed as 'to do' yet that never changes. Nouveau developers have no help so yes, of course, that is a different situation. But, I didn't say that Nvidia and AMD/ATI are comparable regarding open source. I just said that AMD/ATI has overrated support - their support and investment in support is way lower than what many here try to assert. No one should be happy with these companies which bend over for Windows.

[Panix]

AMD driver is FGLRX
r600g is a community driver with open documentation provided by AMD

so maybe we should rename r600g to NonAMDGenericLinuxCommunityDriverThatSupportsAMDHa rdwarebecausetheygaveusdocs600g??

403 times already
a.)r600g is a linux native !!!!community!!!! driver
b.)AMD is not endorsing or supporting or writing or financing or [fill whatver you want] r600g in any way
c.)AMD just promised Documentation, skeleton code and hardware support for the docs[aka ask the hardware dudes why [fill your asm atom] is doing B instead of A ] and so far they have keep their side of deal[pm bits maybe in a grey area]
d.)the entire linux graphic stack have the massive amount of 10 developers[<--tops] and is already impressive they reach this state in just 2 or 3 years [the previous years were graphic stack design not drivers]
e.)wanna make the development faster ??
1.)contribute code[read todo in git]
2.)hire a developer or get a group of ppl and make a fund to pay some developers
3.)sweet talk university students into gallium for their summer time[<-- the very nerdy ones]
4.)provide detailed bug reports
5.)stop whining and GOTO 1

Or support Linux and not just Windows.

[bridgman]

...Also, on the feature set at the Radeon Feature web page, you can note all the 'features' that are listed as 'to do' yet that never changes. Nouveau developers have no help so yes, of course, that is a different situation.

This is simply not true. New features get added as older ones get implemented.

There will always be a TODO list, just like with a Windows driver -- you just don't get to see the TODO list for Windows

[disi]

Where did I make that comparison? I just contrasted how people act or more accurately, 'react' to what Nvidia and AMD/ATI do.

AMD/ATI doesn't release all the specs as I understand it. There is no support with hardware acceleration and various licensing info or whatever restrictions there are. Also, on the feature set at the Radeon Feature web page, you can note all the 'features' that are listed as 'to do' yet that never changes. Nouveau developers have no help so yes, of course, that is a different situation. But, I didn't say that Nvidia and AMD/ATI are comparable regarding open source. I just said that AMD/ATI has overrated support - their support and investment in support is way lower than what many here try to assert. No one should be happy with these companies which bend over for Windows.

I think it is a unique situation.

In MS Windows people are in fact not allowed to write their own drivers, in fact you cannot load any unsigned driver since Windows Vista 64bit (I think). People could write their own driver, pay MS to sign it and then load it into the OS. There are some hacks to use a dodgy certificate etc. but I never passed the barrier to e.g. load a driver to a RAM drive or the ext2 drivers in recent versions.

In Linux, people rather prefer to write their own drivers and AMD provides the information needed to do so. I am not following any discussions between the AMD people and Radeon developers and how much information is released. Often you hear bridgeman saying something like there could be improvements in the open driver compared to the binary. Then people ask to release the code of the binary driver, which is the big argument going on especially for power management.
/{not sure part} The graphics card comes with power tables, which are set in the chip by the manufacturer like MSI, XFX etc. and those are accessable by the open driver. Many cheap cards lack enough steps in those tables... it seems the binary driver can work around this problem and scale the frequencies independently {not so sure part}/

Additionally there is still the firmware, which needs to be loaded into the kernel and is provided by AMD (the firmware for the Nouveau drivers are kind of legally provided, because people would need to download the driver from Nvidia and extract the part themselves to be legal). This is how much AMD does so far and I think it is a lot.

[ChrisXY]

not sure is 5 is the right numbers but as far as i know this fill the entry i writed before -- provide skeleton code/support and ofc they can contribute to other parts of the code that they see fit since is an OSS project but i think bridgman can provide a more precise answer of their jobs exactly

the fact AMD pay some devs to improve things is not different than MyXcompany. inc doing it or yourXcompany. inc doing it since the code they provide follow the project guidelines and is not managed or owned by AMD

ofc AMD maybe interested in help to bring faster certain features to r600g for internal reasons but it doesnt make their driver or property they are just another contributor

http://en.wikipedia.org/wiki/Graphic...FOSS#ATI.2FAMD

[disi]

http://en.wikipedia.org/wiki/Graphic...FOSS#ATI.2FAMD

Everything is cited with Michael Larabel I guess its true but just because its written on wikipedia...

[GreatEmerald]

Yeah well, lots of people read something on a forum somewhere and then repeat it. Unfortunately none of them actually know what they're talking about, and it's complete BS.

Yea, from what reliable information I have seen, NVIDIA hardware is better supported just because FGLRX was heavily bug-ridden for a long time.

[artivision]

Yea, from what reliable information I have seen, NVIDIA hardware is better supported just because FGLRX was heavily bug-ridden for a long time.

Example of Nvidia quality: It was 2010-2011 when Nvidia gave static-compiling support for Linux. When the GLSL-compiler compiles and optimizes graphics(shaders), Nvidia driver saves this code (probably in home folder, MBytes to GBytes). Then with some hacks, the driver doesn't compile again the same thing. So the game is slower only the first time you run it.