[clug] Nvidia based Laptops in the $1200-$2000 range

[Francis James Whittle]

On Sun, 2009-05-31 at 21:54 +1000, Andrew Janke wrote:
> Ah. From what I knew the story went something like this:
> 
> AMD buys ATI makes noise about everything in ATI drivers doing open
> source. The news finally comes out of embargo on the phonorix? forums

The idea that ATI were ever going to open up their drivers was pure
speculation by optimists.

> ATI/AMD then release lots of NDA documentation and wait for others to
> write better open source drivers while continuing to work on their
> own.

Sort of yes and no.  A lot of documentation was released under NDA with
particular developers that were/are working with AMD to create FOSS
drivers.  Mostly this concentrated on the r3xx-r5xx (previous
generation) GPUs however did also give pointers to how to work the
shader (?) engine and thus 2D acceleration on the r6xx/r7xx chips

> This was all circa late 2007 I thought? I am sure there is someone on
> this list who is more knowledgeable about this than I. In my case am I
> lowly FOSS app developer as opposed to a kernel/video card driver
> hacker. Again, if there is a good FOSS ATI driver for the current
> generation of laptop cards that will do what I want (radeonhd) then I
> will be all over it.

Since then (earlier this year in fact) almost all documentation useful
for creating a 3D driver for ATI's most recent chips has been released
to anyone who wants to look at it.  This is a huge step forward, however
has occured in the middle of an attempt to rejuvenate the DRI  and GLX
stacks to try and overtake certain... other... platforms again (You may
have heard of the Gallium stack over the last couple of years, well now
it's being worked on properly).  At the same time an amount of
re-engineering is happening to the existing radeon drivers, including
attempts to futurify it with DRI2 and Gallium, and the drivers for the
newer chips are being (slowly) made to work with this.  It's estimated
that the new drivers will be available some time around Q4 of this year,
but don't hold your breath.  In the mean time, for most purposes the
Catalyst drivers work well enough.

> As of today http://www.x.org/wiki/radeonhd still says "Future work is
> happening especially on more advanced features like 2D, 3D, and video
> acceleration". Or am I looking at the wrong driver?

radeonhd, yes.... There's oft talk of merging it and radeon, or putting
all the r500 work from it back into radeon and concentrating it on newer
chips (later pushing it into the "unified" ati stack).  I can tell you
its VC tree has a little more activity than radeon, anyway.

> Anyhow since then from my own tests of later model ATI and Nvidia
> cards when asked to do more than the normal (multiple monitors, etc)
> Nvidia has always seemed to be better and/or easier to make it happen.
> I do a fair amount of OpenGL/gtkgl/Coin/etc dev work so this sort of
> stuff is somewhat important.

I'd have to argue that ATI seems to do a better job of TV-Out, but apart
from that I've never really seen much [dis]advantage either way when
trying to do thinks like setting up multiple monitors.

> >> Now you're making me rethink my nvidia hardline stance! (rarrrgh!)
> >
> > Good. :)
> 
> Well it would be if there was a viable FOSS alternative that did what
> I want/need! So for now I have to choose between the lesser of two
> "evils". :)  And right now I am tending towards the one starting with
> N.

Well it depends on how long you want to wait.
ATi should have open drivers that work for 3D in 6-12 months time, and
they work well with 2D (even XVideo, or at least they well work for me)
right now, and support XRandR to at least 1.2 (covering multiple
monitors, rotation, mirroring, etc);  Actually I'd go so far to say that
my personal experience has been that the open drivers have actually been
better for the stuff that's implemented than the vendor supplied ones.
As for nVidia, any hope for open drivers that work hinges on the nouveau
reverse engineering project.  However, the 3D performance of their
closed drivers is better, and they release bug fixes more or less as
they happen - which means more widespread testing and the ability to fix
regressions before they become too much of a frustration, which has been
one of the major problems with ATi drivers.

On Topic though, I steered my girlfriend towards a Compaq CQ60-TX214
which cost about $1500 and has nVidia graphics.  It seems to work well
with Ubuntu, and the graphics is about on par with my desktop box which
has a 256MB Radeon 2600, which is not too shabby for a laptop (Although
my desktop box is probably hindered somewhat from being an AGP chipset
which was obsolete, what? 3 years ago?).  Haven't had any troubles
setting it up, although we haven't tried doing any multimonitor stuff at
all.