[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: multi-head with different resolutions & other issues

On 06/18/04 03:44:59, Flavio Stanchina wrote:

Kacper Wysocki wrote:

[...] came across the packages and this list when upgrading to 2.6.7
broke the ati installer.


Could you please post any error messages you got? My packages work
fine apparently (upgraded to 2.6.7 myself yesterday) but I'd like to
know exactly which patch makes a difference. I guess it's this:
http://www.stanchina.net/~flavio/debian/fglrx-archive/msg00050.html


I'm too lazy, but from memory it was in agpart_be.c while compiling the
kernel module, got a message to the effect of
"count not a member of struct `page'". In any case, the problem was
indeed the one described in the link above.


Is it possible to set up multi-head "xinerama" dri-enabled fglrx
with  different resolutions on different monitors? [...]


AFAIK, no it's not possible. If you want DRI on both monitors, you
*have* to use the same resolution; if you use Xinerama, you'd better
use the same resolution anyway, otherwise the window manager gets
confused (in my experience).


But xinerama with different resolutions is possible, and works quite
nicely with the opensource radeon driver. DRI with pseudo-xinerama and
different resolutions should work too. That's why it puzzles me that
ati didn't include this in their driver (if that's the case).
Admittedly, a xinerama desktop works best if both screens have the same
resolution - I guess the only fix for that is to get a new monitor.


AFAIAC the big issue with xinerama is that one cannot use dri.


I tested dual-head briefly with my brother's 9600XT and I had DRI
working on both monitors. I didn't test with Xinerama. Indeed, I
would guess that it's not easy to get a 3D application working when
it is split across two different displays, and that's where MergedFB
comes in.


3D apps work very nicely with nvidia's "pseudo-xinerama" mode (and
reputedly with MergeFB), I just wish the fglrx driver had this
functionality. Xinerama was never designed with this in mind, though,
and that's why nvidia basically bypasses it all with TwinView, and
Xdevs bluntly say dri will never work with xinerama. It's a software limitation, not a hardware one. 


On a related note, I'm running debian sid - what's the preferred way
of  obtaining & installing the latest radeon dri module and gl
libraries?


I'm not sure I understand your question here. I think you just want
to add my package repository to your /etc/apt/sources.list and let me
do the hard work of packaging new driver releases.


Nope, what I meant was in relation to the opensource radeon/dri stuff.
Sorry if that wasn't clear - after all, the question was OT on a fglrx
list anyways. I had no issues obtaining and installing the fglrx driver
before I upgraded to 2.6.7 - but I prefer open source when possible.


Flavio, I'm running an "Evil Master" (or Powered by ATI) Radeon
8500LE  128MB DDR with a dual-head setup [...]  The LCD screen is
not DVI, but I  am using a DVI->CRT converter, and everything is
working fine.



Yes, the problem is not the DVI itself, it's just the digital part of
it with the Linux driver, and apparently only with my 8500LE. Nobody
else reported DVI problems, but not many reported success either. If
you could test with a DVI screen, that would make a difference.


I'd test if I had a DVI screen. Does the dvi output work with the open
source drivers? If it does then obviously ati messed something up in
the driver code, but then again I'm betting you've already done the
"ATI Linux driver feedback" thing.

Cheers,
	Kacper