Ok so,
I found how to do it, and I think you can be interested by the method.
First of all, I'm using only geforce cards.

The option indicated in the URL to use VGL on EC2, named "UseDisplayDevice" set
to "none" is only working with Quadro and Tesla cards.
So I needed to find another way to run headless, AND to use each GPU one buy
one, with 1 virtual display by GPU (eg: :0.0, :0.1 etc ...).

So the solution is very simple, when you know it :)
Note that the Linux Nvidia's driver is unified, then you can use at the same
time in the same computer some Tesla, Quadro and Geforce cards.

X has a very great option I discovered, who is "ConnectedMonitor" followed
by a string.
Then, to make my 2 GTX295 working in headless mode, and be able to use each
GPU independently, I used this command :
nvidia-xconfig -a --connected-monitor=DFP --virtual=1280x1024

It results this xorg.conf (don't pay attention to the number of GPU, who is
3 instead of 4, I got a dead GPU :( )  :

# nvidia-xconfig: X configuration file generated by nvidia-xconfig
# nvidia-xconfig:  version 256.53  (buildmeis...@builder97.nvidia.com)  Fri
Aug 27 20:55:22 PDT 2010

Section "ServerLayout"
    Identifier     "Layout0"
    Screen      0  "Screen0"
    Screen      1  "Screen1" RightOf "Screen0"
    Screen      2  "Screen2" RightOf "Screen1"
    InputDevice    "Keyboard0" "CoreKeyboard"
    InputDevice    "Mouse0" "CorePointer"
EndSection

Section "Files"
EndSection

Section "InputDevice"
    # generated from default
    Identifier     "Mouse0"
    Driver         "mouse"
    Option         "Protocol" "auto"
    Option         "Device" "/dev/psaux"
    Option         "Emulate3Buttons" "no"
    Option         "ZAxisMapping" "4 5"
EndSection

Section "InputDevice"
    # generated from default
    Identifier     "Keyboard0"
    Driver         "kbd"
EndSection

Section "Monitor"
    Identifier     "Monitor0"
    VendorName     "Unknown"
    ModelName      "Unknown"
    HorizSync       28.0 - 33.0
    VertRefresh     43.0 - 72.0
    Option         "DPMS"
EndSection

Section "Monitor"
    Identifier     "Monitor1"
    VendorName     "Unknown"
    ModelName      "Unknown"
    HorizSync       28.0 - 33.0
    VertRefresh     43.0 - 72.0
    Option         "DPMS"
EndSection

Section "Monitor"
    Identifier     "Monitor2"
    VendorName     "Unknown"
    ModelName      "Unknown"
    HorizSync       28.0 - 33.0
    VertRefresh     43.0 - 72.0
    Option         "DPMS"
EndSection
Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce GTX 295"
    BusID          "PCI:4:0:0"
EndSection

Section "Device"
    Identifier     "Device1"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce GTX 295"
    BusID          "PCI:5:0:0"
EndSection

Section "Device"
    Identifier     "Device2"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce GTX 295"
    BusID          "PCI:8:0:0"
EndSection

Section "Screen"
    Identifier     "Screen0"
    Device         "Device0"
    Monitor        "Monitor0"
    DefaultDepth    24
    Option         "ConnectedMonitor" "DFP"
    SubSection     "Display"
        Virtual     1280 1024
        Depth       24
    EndSubSection
EndSection

Section "Screen"
    Identifier     "Screen1"
    Device         "Device1"
    Monitor        "Monitor1"
    DefaultDepth    24
    Option         "ConnectedMonitor" "DFP"
    SubSection     "Display"
        Virtual     1280 1024
        Depth       24
    EndSubSection
EndSection

Section "Screen"
    Identifier     "Screen2"
    Device         "Device2"
    Monitor        "Monitor2"
    DefaultDepth    24
    Option         "ConnectedMonitor" "DFP"
    SubSection     "Display"
        Virtual     1280 1024
        Depth       24
    EndSubSection
EndSection


Then with this, I'm able to have the 3D on each virtual monitor, and let's X
believe it got the monitors connected.
Then, like I said, I'm running headless (with geforce cards), and I'm able
to launch vgl on choosen GPU with the -d :0.X argument.

Hope this helps other people :)

On Wed, May 4, 2011 at 8:56 PM, DRC <dcomman...@users.sourceforge.net>wrote:

> Example of a multi-card xorg.conf:
>
>
> http://www.mail-archive.com/virtualgl-users@lists.sourceforge.net/msg00335.html
>
> How to run headless:
>
> http://comments.gmane.org/gmane.comp.video.opengl.virtualgl.user/477
>
>
> On 5/4/11 6:09 AM, Philippe wrote:
> > Hello,
> > I've got two questions without answer, maybe because I don't found it in
> > the documentation, or maybe because I just don't understand it in the
> doc.
> >
> > First, I'm trying to use VirtualGL on a server without plugging a screen
> > on the graphic card.
> > I already configured vgl with vglserver_config, and with this, I'm able
> > to use VGL without the need to open a session before.
> > But if I try with no screen plugged on the graphic card, I just get the
> > classical "[VGL] ERROR: Could not open display :0." (same with xdpyinfo
> > -display :0 ).
> > So how can I be able to use the capabilities of VGL without screen ?
> >
> >
> > The other question is related to the use of multiple GPUs with Xinerama.
> > The VGL doc says :
> > VirtualGL can redirect the OpenGL commands from a 3D application to any
> > 3D graphics card in the server machine. In order for this to work,
> > however, all of the 3D graphics cards must be attached to different
> > screens on the same X server using Xinerama. They can then be
> > individually addressed by setting |VGL_DISPLAY| to (or
> > invoking |vglrun -d| with) |:0.0|, |:0.1|, |:0.2|, etc.
> >
> > So, I want to do this, obviously without any plugged screen, and then I
> > need xinerama to create/attach the displays I suppose.
> > How to do this then ?
> >
> > Thanks for answers guys, and don't hesitate to request more informations
> !
> >
> > Have a nice day.
> >
> > Philippe.
> >
> >
> >
> >
> ------------------------------------------------------------------------------
> > WhatsUp Gold - Download Free Network Management Software
> > The most intuitive, comprehensive, and cost-effective network
> > management toolset available today.  Delivers lowest initial
> > acquisition cost and overall TCO of any competing solution.
> > http://p.sf.net/sfu/whatsupgold-sd
> >
> >
> >
> > _______________________________________________
> > VirtualGL-Users mailing list
> > VirtualGL-Users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
> ------------------------------------------------------------------------------
> WhatsUp Gold - Download Free Network Management Software
> The most intuitive, comprehensive, and cost-effective network
> management toolset available today.  Delivers lowest initial
> acquisition cost and overall TCO of any competing solution.
> http://p.sf.net/sfu/whatsupgold-sd
> _______________________________________________
> VirtualGL-Users mailing list
> VirtualGL-Users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
------------------------------------------------------------------------------
WhatsUp Gold - Download Free Network Management Software
The most intuitive, comprehensive, and cost-effective network 
management toolset available today.  Delivers lowest initial 
acquisition cost and overall TCO of any competing solution.
http://p.sf.net/sfu/whatsupgold-sd
_______________________________________________
VirtualGL-Users mailing list
VirtualGL-Users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/virtualgl-users

Reply via email to