UT2003

[trommaster]

yep, the new ati drivers can run ut2k3

[SkyNet]

Well that's good news. I sure hope ATI keeps following through with letting it's 3d cards be open-source drivers.

[JimH]

Well that's good news. I sure hope ATI keeps following through with letting it's 3d cards be open-source drivers.

ATI is an active supporter of the open source community. Not like some other companies. We all know who they are. ;D You will see features and support for cards not present in the current open source driver sometime in the future.

I will post a couple of comments Mike Harris made on the redhat-xfree86 mailing list.

Jim H

[JimH]

Comment #1


From: Mike A. Harris <mharris@redhat.com>
Reply-To: xfree86-list@redhat.com
To: xfree86-list@redhat.com
Subject: Re: ATI releases Linux drivers
Date: Sat, 23 Nov 2002 04:12:31 -0500 (EST)
On Fri, 22 Nov 2002, Dave Reed wrote:

>Yes, I made certain my Radeon 7500 was "built by ATI" when I
>bought it. Does anyone know if the open source developers plan
>to continue developing drivers for the new cards even those
>these binary drivers will be out there?

Do open source developers develop open source kernel drivers even
though binary only kernel drivers may be available for the given
hardware? Yes, of course. Why? Because of the benefits that
open source provides.

>Is ATI at least still releasing specs for their new cards so
>drivers can be written (that's the reason I switched from NVidia
>to ATI - I want to be able to upgrade my machine to the newer
>Red Hat versions w/o worrying about the driver not working with
>them).

I have not personally contacted ATI to get their latest specs for
the 9x00 boards yet - mostly because I haven't needed the specs
yet. I generally do not acquire docs for stuff until I need it,
so I can't answer that question. My assumption is that docs are
available as they always have been in the past.

>If open source drivers are being written, I can be assured that
>evenutally I can make that upgrade w/o losing the use of my
>graphics card.

It is impossible for anyone to answer that question. Until
something exists, anything is nothing more than speculation.

>If they're still releasing the specs and people are still
>writing the drivers, I can live with buying a new card from ATI,
>but if they've quit releasing their specs or there are no plans
>for open source drivers, then there's no "moral" reason to buy
>ATI and I'd look more closely at price, performance, etc. of ATI
>vs. NVidia.

ATI has not stopped supporting the open source community, and I
see no reason why they would stop doing so. As I mentioned in a
previous email, ATI contributed 2 large patches to XFree86.org
this week (I receive all patch submissions automatically)
implementing improved support, support for new hardware, and
numerous other things.

The biggest problem open source faces, is that 3D drivers are
large, complex beasts, and take a LOT of time to develop.
Writing 3D drivers requires a large range of skills, and
understanding of kernel device driver programming, 3D hardware,
OpenGL, Mesa, as well as XFree86 itself, and of 2D driver
authoring. It is a large complex task, and takes a lot of
dedicated time and effort to complete a driver. Anyone doing
this on their spare time as a volunteer, is likely to take quite
a long time to do it, and adding more developers only helps to a
certain degree. (See Brooks law). Most of the work is work that
needs to be completed serially, and so it doesn't lend well to
parallel development by many people. As such, it makes sense if
an open source project to support a given card is started 6-8
months *after* the card hits the market, and is done by unfunded
volunteers who have a real fulltime job doing something else, it
is going to take a long time until we see support.

And for the case of one of the vendors open sourcing their
drivers, they have different reasons for not doing so, some are
quite valid, and others are less valid.

Some reasons a given random hardware (of any hardware) or
software vendor might not open source their code:

1) It may contain code they've licensed from some other company,
and they may not have the right to open source it. Or it
might contain patented algorithms, etc. that they have
licenced from some other company and do not have the right to
redistribute the soource.

2) It may possibly contain patented intellectual property
that they own, or even unpatented techniques, trade
secrets, etc. and they do not want their competition to
learn how it works, and then improve their own drivers.

3) When a company open sources a driver, or any source code at
all for that matter, they now run the risk that they may have
infringed some patent UNKNOWINGLY in the code, and nobody ever
knew, neither them, nor the person who's patent they might
have infringed upon. An example being 2 completely different
companies both inventing the same idea independantly without
knowing it, and one of them patenting it. If the source code
is released, the owner of the patent might end up reading the
code and finding their patent used and suing the other
company - even though no patent was knowingly violated by the
other party. This is a very very SAD state of the patent
system, but it is a fact of life.

4) It is possible that a vendor has knowingly and purposefully
infringed upon some patent, and simply does not want anyone to
know abut it, as they could get sued. They simply wont ever
release their code period.

5) Their source code might be extremely hideous spagetti, and
releasing it could be a major embarassment to them, in
particular if paying customers see it. This could be a major
black eye to the company for trying to do something good.

6) Their code might do really really evil kernel hacks that
sacrifice stability for speed, or other nastiness, and they
do not want anyone to know about.

7) Their code might violate the DMCA if open sourced (think of
disabling copy protection on your TVout).

There are many other possible reasons. I'm not trying to play
devils advocate or anything, or justify why a company might not
want to release their code as open source. I'm just trying to
illustrate some of the reasons why a given company might decide
not to open source.

In the end, before a company open sources something, they are
going to want to see what THEY get out of it. How does it
benefit THEM, and their STOCKHOLDERS(TM). One could argue that
by open sourcing one's drivers, the competition might be able to
improve their own closed drivers, and that could possibly affect
sales, and thus their stock price.

So, it is not always a simple decision for a company to make on
open sourcing their stuff. Do I personally want to see open
source drivers? You bet! And I like many, want to see open
source drivers, software, etc. regardless of all the other
things, however - I do understand that a company may have some
very good reasons for not doing so, and ultimately - they don't
_owe_ us anything. It is up to us, the open source community, to
both convince hardware vendors that open sourcing their drivers
is good, and to show them how THEY will benefit. We can show
them how WE benefit, but should they care? If we show them how
THEY benefit, and allay any fears they may have, then it might be
possible. However, the above points on patents ALWAYS will get
in the way, as long as the United States, and some other
countries have the stupid patent system and allow software
patents to exist.

--
Mike A. Harris ftp://people.redhat.com/mharris
OS Systems Engineer
XFree86 maintainer
Red Hat Inc.

[JimH]

Comment #2



On Fri, 22 Nov 2002, Dave Reed wrote:

>http://mirror.ati.com/companyinfo/press/2002/4574.html
>
>http://www.theinquirer.net/?article=6345
>
>I couldn't find source code - does anyone know if they plan to release
>the source code for their drivers?

The above drivers are ATI's proprietary drivers for XFree86.
Originally, ATI produced proprietary drivers for their FireGL
hardware. When the FireGL 8800 came out, they produced
proprietary drivers for that as well. These are very high end
boards that are very expensive, and not mainstream. At some
point ATI decided that they would also tweak the drivers to work
with the Radeon 8500 as well, since it is for the most part just
a slower version of the FireGL 8800.

While ATI was in the testing phase of testing the 8500 out with
the FireGL drivers, someone else found out that the drivers
worked with the 8500 too, and people posted to various lists that
the FireGL drivers magically worked with the 8500 (like they were
pulling one over on ATI, when in reality, ATI purposefully was
working on supporting the 8500, but just hadn't announced it
officially yet).

Recently, ATI made an official announcement of Radeon 8500
support in their drivers, and this latest announcement is just a
newer driver version which also supports their newer Radeon 9x00
hardware as well.

These drivers are closed source, and most likely will remain that
way, as they are merely provided to people mostly due to lots of
people requesting it.

ATI does however support open source as well, and contributes new
open source 2D driver support for all of their new hardware to
the XFree86 project very frequently. They also help developers
such as myself, the DRI project, XFree86, GATOS and others by
providing documentation to certain individuals, and in some cases
providing sample hardware for open source development.

ATI just sent patches to XFree86.org this week for example to add
support for the Radeon 9500 chips, and fix many bugs, improve
monitor detection and many other things.

The DRI project is currently in the midst of completing open
source Radeon 8500 3D support as well. This open source
development would not have been possible without the help of ATI
providing documentation, etc. to the DRI project.

There is currently to my knowledge no project underway to
implement open source support for Radeon 9x00 hardware, however
it could be that it is being started quietly (as the 8500 support
was), or it might be something that occurs sometime down the
road. In general, open source support for new hardware like this
is kept quiet until it reaches the state where it is remotely
ready to be seen by the unwashed masses. Most developers I
believe by now know if they acknowledge they are working on
something, that they will be endlessly bombed with email and
other communication daily asking if they can test things, asking
if the code is ready, asking why it is not public, etc...

In short, if you want to keep up to date with what hardware is
supported by DRI with open source drivers, you might want to join
the dri-devel mailing list. Judging by how long it will have
taken to get Radeon 8500 support once it is released in an
official XFree86 release, I believe it is fairly good assumption
that it will take that long or longer to see open source 3D
drivers for Radeon 9x00 as well.

>This may motivate me to replace my Radeon 7500 with a 9700 (now
>I can tell my wife what I want for Christmas :-)

If you want open source 3D support, or want working 3D out of the
box, right now the best card for 3D is the Radeon 7500. When
XFree86 4.3.0 is released, if all goes well, then the Radeon 8500
will fill that place.

Radeon 9x00 hardware will have only closed source drivers for the
immediate future. The difference now though, is that ATI is
providing their customers with their drivers they've
traditionally kept for their high end workstation customers, at
least until open source drivers may appear.

Hope this helps answer your questions.

Take care,
TTYL

--
Mike A. Harris ftp://people.redhat.com/mharris
OS Systems Engineer
XFree86 maintainer
Red Hat Inc.