[ltp] Getting nVidia to clone display on a projector
John Jason Jordan
linux-thinkpad@linux-thinkpad.org
Tue, 23 Nov 2010 09:50:50 -0800
On Tue, 23 Nov 2010 15:24:09 +0200
Micha Feigin <michf@post.tau.ac.il> dijo:
>On Mon, 22 Nov 2010 15:13:47 -0800
>John Jason Jordan <johnxj@comcast.net> wrote:
>
>> On Mon, 22 Nov 2010 23:23:52 +0200
>> Micha Feigin <michf@post.tau.ac.il> dijo:
>>
>> >> The problem that remains, however, is getting my laptop screen to
>> >> display anything other than 1680x1050. I am currently using the
>> >> latest nVidia driver with Fedora 13, x86_64, and it allows me to
>> >> select:
>> That is what the xorg.conf file created by the nVidia installer says
>> on my computer too. So you're saying the reason I get the option of
>> 1680x1050, and nothing lower until I go all the way down to 1280x1024
>> is because the nVidia driver decided that the intervening resolutions
>> wouldn't work at the above sync and refresh rates?
>Like I said. I used to have that problem. I don't know if it was a
>problem with the edid the screen sent or something with the xserver /
>nvidia driver, etc.
>
>I used to need the linked approach to create a custom edid file.
>Currently everything works fine on my laptop, so there was some
>software bug along the way that nouveau doesn't suffer from and nvidia
>does.
I have since discovered a utility called monitor-edid. This tool polls
the screen for its E-DID data and displays it on the command line. For
my T61 it says:
[root@Devil8 X11]# monitor-edid
EISA ID: IBM2887
EDID version: 1.3
EDID extension blocks: 0
Screen size: 33.1 cm x 20.7 cm (15.37 inches, aspect ratio 16/10 = 1.60)
Gamma: 2.2
Digital signal
# Monitor preferred modeline (60.0 Hz vsync, 63.9 kHz hsync,
ratio 16/10, 128 dpi) ModeLine "1680x1050" 120.6 1680 1712 1760
1888 1050 1051 1054 1065 -hsync -vsync
# Monitor supported modeline (50.0 Hz vsync, 53.2 kHz hsync,
ratio 16/10, 128 dpi) ModeLine "1680x1050" 100.53 1680 1712
1760 1888 1050 1051 1054 1065 -hsync -vsync
It is interesting that the only resolution the E-DID data from the
screen reports is 1680x1050. I also note from the Xorg.o.log file this
part (edited for readability):
[ 532.284] (--) NVIDIA(0): Connected display device(s) on Quadro NVS
140M at PCI:1:0:0
[ 532.284] (--) NVIDIA(0): IBM (DFP-0)
[ 532.285] (--) NVIDIA(0): IBM (DFP-0): 330.0 MHz maximum pixel clock
[ 532.285] (--) NVIDIA(0): IBM (DFP-0): Internal Dual Link LVDS
[ 532.290] (WW) NVIDIA(0): The EDID for IBM (DFP-0) contradicts
itself: mode "1680x1050" is specified in the EDID; however, the
EDID's valid HorizSync range (53.246-63.877 kHz) would exclude
this mode's HorizSync (42.6 kHz); ignoring HorizSync check for
mode "1680x1050".
[ 532.290] (WW) NVIDIA(0): The EDID for IBM (DFP-0) contradicts
itself: mode "1680x1050" is specified in the EDID; however, the
EDID's valid HorizSync range (53.246-63.877 kHz) would exclude
this mode's HorizSync (42.6 kHz); ignoring HorizSync check for
mode "1680x1050".
[ 532.341] (II) NVIDIA(0): Assigned Display Device: DFP-0
[ 532.341] (II) NVIDIA(0): Validated modes:
"nvidia-auto-select"
[ 532.341] (II) NVIDIA(0):
Virtual screen size determined to be 1680 x 1050
It looks to me like the E-DID data in the screen is messed up.
Furthermore, it looks like the E-DID data reports only one possible
resolution. That leads me to wonder where the nVidia driver came up
with the other resolutions that it presents to me: 1280x1024, etc. They
do work, but apparently the driver made them up. When I booted to Lucid
live CD with the nouveau driver it probably did the same thing, except
that intelligently offered me 1400x1050 as the next resolution after
1680x1050.
I also discovered a shareware utility called Powerstrip that apparently
can modify the E-DID data in a monitor. Unfortunately it runs only on
Windows, and the only Windows I have on this computer is running in
Virtualbox. Powerstrip is also scary. It's working OK at 1680x1050;
what if I mess it up and it won't work at all?
It would be interesting if there are any other people here who have a
15.4" T61 with the nVidia Quadro NVS 140M. Comparing notes might be
useful.