virtualgl-users

[VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Stealthy <ste...@gm...>]

> Hi guys,  
>  
> Hope I can get some help on this one. I have an ec2 instance running using a GPU cluster. I am capturing the Xvfb virtual screen using  ffmpeg and would like to have my OPENGL enabled application use the GPU to render. It’s currently rendering everything using the CPU.  
>  
> I already installed virtualgl but I’m having some trouble intializing virtualgl with Xvfb and ensure the GPU is being used. Anybody could help on this?
>  
> Thanks in advance   
>  
> --  
> Desmond
>  

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[DRC <dco...@us...>]

"Trouble initializing VirtualGL with Xvfb" tells me nothing.  Please be 
specific as to the error messages you are encountering or what is not 
working as you expect.


On 11/6/13 2:16 PM, Stealthy wrote:
>
>> Hi guys,
>>
>> Hope I can get some help on this one. I have an ec2 instance running
>> using a GPU cluster. I am capturing the Xvfb virtual screen using
>>  ffmpeg and would like to have my OPENGL enabled application use the
>> GPU to render. It’s currently rendering everything using the CPU.
>>
>> I already installed virtualgl but I’m having some trouble intializing
>> virtualgl with Xvfb and ensure the GPU is being used. Anybody could
>> help on this?
>>
>> Thanks in advance

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Stealthy <ste...@gm...>]

I am not sure I going through the right steps to enable virtualgl. Post installation I started the Xvfb server with this command:  

Xvfb :0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &

I then started what I think is virtualgl with this command expecting any opengl application running after that would use the GPU:

/opt/VirtualGL/bin/vglclient

The application still only uses the CPU.

Am I missing something?

Thanks for the help btw.
--  
Desmond


On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:

> "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be  
> specific as to the error messages you are encountering or what is not  
> working as you expect.
>  
>  
> On 11/6/13 2:16 PM, Stealthy wrote:
> >  
> > > Hi guys,
> > >  
> > > Hope I can get some help on this one. I have an ec2 instance running
> > > using a GPU cluster. I am capturing the Xvfb virtual screen using
> > > ffmpeg and would like to have my OPENGL enabled application use the
> > > GPU to render. It’s currently rendering everything using the CPU.
> > >  
> > > I already installed virtualgl but I’m having some trouble intializing
> > > virtualgl with Xvfb and ensure the GPU is being used. Anybody could
> > > help on this?
> > >  
> > > Thanks in advance
>  
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models. Explore
> techniques for threading, error checking, porting, and tuning. Get the most  
> from the latest Intel processors and coprocessors. See abstracts and register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li... (mailto:Vir...@li...)
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>  
>  

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[DRC <dco...@us...>]

Yes, you're missing quite a lot.  Please read the manual.  VirtualGL 
needs two X servers-- a "3D X server" and a "2D X server."  The 3D X 
server has to be a real X server attached to the GPU.  It can be 
headless, but it can't be virtual.  The 2D X server can be a virtual X 
server, such as VNC or Xvfb.

Secondly, the way you launch a 3D application in VGL is by using vglrun, 
not vglclient.  The instructions do not ever even specify calling 
vglclient directly.  vglclient is generally called from vglconnect, but 
you don't need to ever use either of those unless your 2D X server and 
3D X server are on different machines (which they aren't.)

I'll summarize the steps you need to take (but please do read the manual.)
-- Set up the 3D X server, which should be a "real" X server connected 
to the GPU.  Make sure the GLX extension works by running glxinfo 
against the 3D X server and verifying that it is not using a software 
OpenGL renderer (Mesa.)
-- Run vglserver_config to give VirtualGL access to the 3D X server 
(this is only necessary if the 3D X server is running some sort of login 
manager under the root account, like gdm or xdm.  Otherwise, if you are 
just starting the 3D X server under your user account, then you should 
already have permission to access it.)
-- Start Xvfb (the 2D X server.)
-- Set DISPLAY to point to your Xvfb instance.
-- vglrun {application}


On 11/6/13 4:21 PM, Stealthy wrote:
> I am not sure I going through the right steps to enable virtualgl. Post
> installation I started the Xvfb server with this command:
>
> *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
>
> I then started what I think is virtualgl with this command expecting any
> opengl application running after that would use the GPU:
>
> /opt/VirtualGL/bin/vglclient
>
> The application still only uses the CPU.
>
> Am I missing something?
>
> Thanks for the help btw.
> --
> Desmond
>
> On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
>
>> "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be
>> specific as to the error messages you are encountering or what is not
>> working as you expect.
>>
>>
>> On 11/6/13 2:16 PM, Stealthy wrote:
>>>
>>>> Hi guys,
>>>>
>>>> Hope I can get some help on this one. I have an ec2 instance running
>>>> using a GPU cluster. I am capturing the Xvfb virtual screen using
>>>> ffmpeg and would like to have my OPENGL enabled application use the
>>>> GPU to render. It’s currently rendering everything using the CPU.
>>>>
>>>> I already installed virtualgl but I’m having some trouble intializing
>>>> virtualgl with Xvfb and ensure the GPU is being used. Anybody could
>>>> help on this?
>>>>
>>>> Thanks in advance
>>
>> ------------------------------------------------------------------------------
>> November Webinars for C, C++, Fortran Developers
>> Accelerate application performance with scalable programming models.
>> Explore
>> techniques for threading, error checking, porting, and tuning. Get the
>> most
>> from the latest Intel processors and coprocessors. See abstracts and
>> register
>> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>> _______________________________________________
>> VirtualGL-Users mailing list
>> Vir...@li...
>> <mailto:Vir...@li...>
>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models. Explore
> techniques for threading, error checking, porting, and tuning. Get the most
> from the latest Intel processors and coprocessors. See abstracts and register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>
>
>
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Stealthy <ste...@gm...>]

Thanks a lot for all the clarification, I’ll definitely spend more time in the manual. This is really helpful, I’ll take another stab at it.  

Thanks again!  

--  
Desmond


On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:

> Yes, you're missing quite a lot. Please read the manual. VirtualGL  
> needs two X servers-- a "3D X server" and a "2D X server." The 3D X  
> server has to be a real X server attached to the GPU. It can be  
> headless, but it can't be virtual. The 2D X server can be a virtual X  
> server, such as VNC or Xvfb.
>  
> Secondly, the way you launch a 3D application in VGL is by using vglrun,  
> not vglclient. The instructions do not ever even specify calling  
> vglclient directly. vglclient is generally called from vglconnect, but  
> you don't need to ever use either of those unless your 2D X server and  
> 3D X server are on different machines (which they aren't.)
>  
> I'll summarize the steps you need to take (but please do read the manual.)
> -- Set up the 3D X server, which should be a "real" X server connected  
> to the GPU. Make sure the GLX extension works by running glxinfo  
> against the 3D X server and verifying that it is not using a software  
> OpenGL renderer (Mesa.)
> -- Run vglserver_config to give VirtualGL access to the 3D X server  
> (this is only necessary if the 3D X server is running some sort of login  
> manager under the root account, like gdm or xdm. Otherwise, if you are  
> just starting the 3D X server under your user account, then you should  
> already have permission to access it.)
> -- Start Xvfb (the 2D X server.)
> -- Set DISPLAY to point to your Xvfb instance.
> -- vglrun {application}
>  
>  
> On 11/6/13 4:21 PM, Stealthy wrote:
> > I am not sure I going through the right steps to enable virtualgl. Post
> > installation I started the Xvfb server with this command:
> >  
> > *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
> >  
> > I then started what I think is virtualgl with this command expecting any
> > opengl application running after that would use the GPU:
> >  
> > /opt/VirtualGL/bin/vglclient
> >  
> > The application still only uses the CPU.
> >  
> > Am I missing something?
> >  
> > Thanks for the help btw.
> > --
> > Desmond
> >  
> > On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
> >  
> > > "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be
> > > specific as to the error messages you are encountering or what is not
> > > working as you expect.
> > >  
> > >  
> > > On 11/6/13 2:16 PM, Stealthy wrote:
> > > >  
> > > > > Hi guys,
> > > > >  
> > > > > Hope I can get some help on this one. I have an ec2 instance running
> > > > > using a GPU cluster. I am capturing the Xvfb virtual screen using
> > > > > ffmpeg and would like to have my OPENGL enabled application use the
> > > > > GPU to render. It’s currently rendering everything using the CPU.
> > > > >  
> > > > > I already installed virtualgl but I’m having some trouble intializing
> > > > > virtualgl with Xvfb and ensure the GPU is being used. Anybody could
> > > > > help on this?
> > > > >  
> > > > > Thanks in advance
> > >  
> > > ------------------------------------------------------------------------------
> > > November Webinars for C, C++, Fortran Developers
> > > Accelerate application performance with scalable programming models.
> > > Explore
> > > techniques for threading, error checking, porting, and tuning. Get the
> > > most
> > > from the latest Intel processors and coprocessors. See abstracts and
> > > register
> > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > _______________________________________________
> > > VirtualGL-Users mailing list
> > > Vir...@li... (mailto:Vir...@li...)
> > > <mailto:Vir...@li...>
> > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > >  
> >  
> >  
> >  
> >  
> > ------------------------------------------------------------------------------
> > November Webinars for C, C++, Fortran Developers
> > Accelerate application performance with scalable programming models. Explore
> > techniques for threading, error checking, porting, and tuning. Get the most
> > from the latest Intel processors and coprocessors. See abstracts and register
> > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> >  
> >  
> >  
> > _______________________________________________
> > VirtualGL-Users mailing list
> > Vir...@li... (mailto:Vir...@li...)
> > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >  
>  
>  
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models. Explore
> techniques for threading, error checking, porting, and tuning. Get the most  
> from the latest Intel processors and coprocessors. See abstracts and register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li... (mailto:Vir...@li...)
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>  
>  

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Stealthy <ste...@gm...>]

Hi guys,  

So I am trying to run a 3d application using OPENGL through an EC2 server. The ultimate goal being to capture the screen using Xvfb.

I installed virtualgl and turbovnc. when I run
/opt/VirtualGL/bin/glxinfo -display :0 -c

I get this

name of display: :0  
display: :0  screen: 0
direct rendering: Yes
server glx version string: 1.4 Mesa 9.2.4
server glx extensions:
    GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,  
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,  
    GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,  
    GLX_SGIX_pbuffer

client glx version string: 1.4 Mesa 9.2.4  
client glx extensions:
    GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,  
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,  
    GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,  
    GLX_SGIX_pbuffer


Obviously I don’t want to use Mesa. Do I have to start the X server in any specific manner to avoid using the MESA software renderer. Did I miss a step.

Thanks for your help guys!
--  
Desmond


On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:

> Thanks a lot for all the clarification, I’ll definitely spend more time in the manual. This is really helpful, I’ll take another stab at it.  
>  
> Thanks again!  
>  
> --  
> Desmond
>  
>  
> On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
>  
> > Yes, you're missing quite a lot. Please read the manual. VirtualGL  
> > needs two X servers-- a "3D X server" and a "2D X server." The 3D X  
> > server has to be a real X server attached to the GPU. It can be  
> > headless, but it can't be virtual. The 2D X server can be a virtual X  
> > server, such as VNC or Xvfb.
> >  
> > Secondly, the way you launch a 3D application in VGL is by using vglrun,  
> > not vglclient. The instructions do not ever even specify calling  
> > vglclient directly. vglclient is generally called from vglconnect, but  
> > you don't need to ever use either of those unless your 2D X server and  
> > 3D X server are on different machines (which they aren't.)
> >  
> > I'll summarize the steps you need to take (but please do read the manual.)
> > -- Set up the 3D X server, which should be a "real" X server connected  
> > to the GPU. Make sure the GLX extension works by running glxinfo  
> > against the 3D X server and verifying that it is not using a software  
> > OpenGL renderer (Mesa.)
> > -- Run vglserver_config to give VirtualGL access to the 3D X server  
> > (this is only necessary if the 3D X server is running some sort of login  
> > manager under the root account, like gdm or xdm. Otherwise, if you are  
> > just starting the 3D X server under your user account, then you should  
> > already have permission to access it.)
> > -- Start Xvfb (the 2D X server.)
> > -- Set DISPLAY to point to your Xvfb instance.
> > -- vglrun {application}
> >  
> >  
> > On 11/6/13 4:21 PM, Stealthy wrote:
> > > I am not sure I going through the right steps to enable virtualgl. Post
> > > installation I started the Xvfb server with this command:
> > >  
> > > *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
> > >  
> > > I then started what I think is virtualgl with this command expecting any
> > > opengl application running after that would use the GPU:
> > >  
> > > /opt/VirtualGL/bin/vglclient
> > >  
> > > The application still only uses the CPU.
> > >  
> > > Am I missing something?
> > >  
> > > Thanks for the help btw.
> > > --
> > > Desmond
> > >  
> > > On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
> > >  
> > > > "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be
> > > > specific as to the error messages you are encountering or what is not
> > > > working as you expect.
> > > >  
> > > >  
> > > > On 11/6/13 2:16 PM, Stealthy wrote:
> > > > >  
> > > > > > Hi guys,
> > > > > >  
> > > > > > Hope I can get some help on this one. I have an ec2 instance running
> > > > > > using a GPU cluster. I am capturing the Xvfb virtual screen using
> > > > > > ffmpeg and would like to have my OPENGL enabled application use the
> > > > > > GPU to render. It’s currently rendering everything using the CPU.
> > > > > >  
> > > > > > I already installed virtualgl but I’m having some trouble intializing
> > > > > > virtualgl with Xvfb and ensure the GPU is being used. Anybody could
> > > > > > help on this?
> > > > > >  
> > > > > > Thanks in advance
> > > >  
> > > > ------------------------------------------------------------------------------
> > > > November Webinars for C, C++, Fortran Developers
> > > > Accelerate application performance with scalable programming models.
> > > > Explore
> > > > techniques for threading, error checking, porting, and tuning. Get the
> > > > most
> > > > from the latest Intel processors and coprocessors. See abstracts and
> > > > register
> > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > > _______________________________________________
> > > > VirtualGL-Users mailing list
> > > > Vir...@li... (mailto:Vir...@li...)
> > > > <mailto:Vir...@li...>
> > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > >  
> > >  
> > >  
> > >  
> > >  
> > > ------------------------------------------------------------------------------
> > > November Webinars for C, C++, Fortran Developers
> > > Accelerate application performance with scalable programming models. Explore
> > > techniques for threading, error checking, porting, and tuning. Get the most
> > > from the latest Intel processors and coprocessors. See abstracts and register
> > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > >  
> > >  
> > >  
> > > _______________________________________________
> > > VirtualGL-Users mailing list
> > > Vir...@li... (mailto:Vir...@li...)
> > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > >  
> >  
> >  
> > ------------------------------------------------------------------------------
> > November Webinars for C, C++, Fortran Developers
> > Accelerate application performance with scalable programming models. Explore
> > techniques for threading, error checking, porting, and tuning. Get the most  
> > from the latest Intel processors and coprocessors. See abstracts and register
> > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > _______________________________________________
> > VirtualGL-Users mailing list
> > Vir...@li... (mailto:Vir...@li...)
> > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >  
> >  
> >  
>  
>  

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Kevin V. W. <va...@sa...>]

It looks like your 3D X server (on :0) is not using the driver for your
card. Make sure the "Device" section of xorg.conf has the correct driver
and BusID specified. Secondly, I don't understand why you are using
turbovnc. If you're trying to do captures with Xvfb, what's turobvnc for?


On Thu, Jan 9, 2014 at 7:36 PM, Stealthy <ste...@gm...> wrote:

>  Hi guys,
>
> So I am trying to run a 3d application using OPENGL through an EC2 server.
> The ultimate goal being to capture the screen using Xvfb.
>
> I installed virtualgl and turbovnc. when I run
> /opt/VirtualGL/bin/glxinfo -display :0 -c
>
> I get this
>
> name of display: :0
>
> display: :0  screen: 0
>
> direct rendering: Yes
>
> server glx version string: 1.4 Mesa 9.2.4
>
> server glx extensions:
>
>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
>
>     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,
>
>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,
>
>     GLX_SGIX_pbuffer
>
>
> client glx version string: 1.4 Mesa 9.2.4
>
> client glx extensions:
>
>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
>
>     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,
>
>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,
>
>     GLX_SGIX_pbuffer
>
> Obviously I don’t want to use Mesa. Do I have to start the X server in any
> specific manner to avoid using the MESA software renderer. Did I miss a
> step.
>
> Thanks for your help guys!
> --
> Desmond
>
> On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:
>
>  Thanks a lot for all the clarification, I’ll definitely spend more time
> in the manual. This is really helpful, I’ll take another stab at it.
>
> Thanks again!
>
> --
> Desmond
>
> On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
>
> Yes, you're missing quite a lot. Please read the manual. VirtualGL
> needs two X servers-- a "3D X server" and a "2D X server." The 3D X
> server has to be a real X server attached to the GPU. It can be
> headless, but it can't be virtual. The 2D X server can be a virtual X
> server, such as VNC or Xvfb.
>
> Secondly, the way you launch a 3D application in VGL is by using vglrun,
> not vglclient. The instructions do not ever even specify calling
> vglclient directly. vglclient is generally called from vglconnect, but
> you don't need to ever use either of those unless your 2D X server and
> 3D X server are on different machines (which they aren't.)
>
> I'll summarize the steps you need to take (but please do read the manual.)
> -- Set up the 3D X server, which should be a "real" X server connected
> to the GPU. Make sure the GLX extension works by running glxinfo
> against the 3D X server and verifying that it is not using a software
> OpenGL renderer (Mesa.)
> -- Run vglserver_config to give VirtualGL access to the 3D X server
> (this is only necessary if the 3D X server is running some sort of login
> manager under the root account, like gdm or xdm. Otherwise, if you are
> just starting the 3D X server under your user account, then you should
> already have permission to access it.)
> -- Start Xvfb (the 2D X server.)
> -- Set DISPLAY to point to your Xvfb instance.
> -- vglrun {application}
>
>
> On 11/6/13 4:21 PM, Stealthy wrote:
>
> I am not sure I going through the right steps to enable virtualgl. Post
> installation I started the Xvfb server with this command:
>
> *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
>
> I then started what I think is virtualgl with this command expecting any
> opengl application running after that would use the GPU:
>
> /opt/VirtualGL/bin/vglclient
>
> The application still only uses the CPU.
>
> Am I missing something?
>
> Thanks for the help btw.
> --
> Desmond
>
> On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
>
> "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be
> specific as to the error messages you are encountering or what is not
> working as you expect.
>
>
> On 11/6/13 2:16 PM, Stealthy wrote:
>
>
> Hi guys,
>
> Hope I can get some help on this one. I have an ec2 instance running
> using a GPU cluster. I am capturing the Xvfb virtual screen using
> ffmpeg and would like to have my OPENGL enabled application use the
> GPU to render. It’s currently rendering everything using the CPU.
>
> I already installed virtualgl but I’m having some trouble intializing
> virtualgl with Xvfb and ensure the GPU is being used. Anybody could
> help on this?
>
> Thanks in advance
>
>
>
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models.
> Explore
> techniques for threading, error checking, porting, and tuning. Get the
> most
> from the latest Intel processors and coprocessors. See abstracts and
> register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> <mailto:Vir...@li...<Vir...@li...>
> >
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
>
>
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models.
> Explore
> techniques for threading, error checking, porting, and tuning. Get the most
> from the latest Intel processors and coprocessors. See abstracts and
> register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>
>
>
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models.
> Explore
> techniques for threading, error checking, porting, and tuning. Get the
> most
> from the latest Intel processors and coprocessors. See abstracts and
> register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
>
>
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>


-- 
Kevin Van Workum, PhD
Sabalcore Computing Inc.
"Where Data Becomes Discovery"
http://www.sabalcore.com
877-492-8027 ext. 11

-- 

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Stealthy <ste...@gm...>]

Checked my xorg.conf file and the driver is nvidia, see below.
Section "Monitor"
    Identifier     "Monitor0"
    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"
    BusID          "PCI:0:3:0"
EndSection


Section "Screen"  
    Identifier     "Screen0"
    Device         "Device0"
    Monitor        "Monitor0"
    DefaultDepth    24
    Option         "UseDisplayDevice" "none"
    SubSection     "Display"
        Virtual     1024 768
        Depth       24
    EndSubSection
EndSection


 Not sure why it’s still using mesa.

 As to the use of turbovnc this is what I was told when I was setting this up:   
"VirtualGL needs two X servers-- a "3D X server" and a "2D X server." The 3D X
server has to be a real X server attached to the GPU. It can be
headless, but it can't be virtual. The 2D X server can be a virtual X
server, such as VNC or Xvfb.”

Isn’t this true? Are you saying I can just run everything through Xvfb and still have my application run through virtualgl and use the gpu? That would make things quite easier if that is possible.  

Thanks!
--  
Desmond


On Friday, January 10, 2014 at 3:53 AM, Kevin Van Workum wrote:

> It looks like your 3D X server (on :0) is not using the driver for your card. Make sure the "Device" section of xorg.conf has the correct driver and BusID specified. Secondly, I don't understand why you are using turbovnc. If you're trying to do captures with Xvfb, what's turobvnc for?
>  
>  
> On Thu, Jan 9, 2014 at 7:36 PM, Stealthy <ste...@gm... (mailto:ste...@gm...)> wrote:
> > Hi guys,  
> >  
> > So I am trying to run a 3d application using OPENGL through an EC2 server. The ultimate goal being to capture the screen using Xvfb.
> >  
> > I installed virtualgl and turbovnc. when I run  
> > /opt/VirtualGL/bin/glxinfo -display :0 -c
> >  
> > I get this
> >  
> >  
> > name of display: :0
> >  
> >  
> > display: :0  screen: 0
> >  
> >  
> > direct rendering: Yes
> >  
> >  
> > server glx version string: 1.4 Mesa 9.2.4
> >  
> >  
> > server glx extensions:
> >  
> >  
> >     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,  
> >  
> >  
> >     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,  
> >  
> >  
> >     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,  
> >  
> >  
> >     GLX_SGIX_pbuffer
> >  
> >  
> >  
> >  
> >  
> >  
> > client glx version string: 1.4 Mesa 9.2.4
> >  
> >  
> > client glx extensions:
> >  
> >  
> >     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,  
> >  
> >  
> >     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,  
> >  
> >  
> >     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,  
> >  
> >  
> >     GLX_SGIX_pbuffer
> >  
> >  
> >  
> > Obviously I don’t want to use Mesa. Do I have to start the X server in any specific manner to avoid using the MESA software renderer. Did I miss a step.
> >  
> > Thanks for your help guys!  
> > --  
> > Desmond
> >  
> >  
> > On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:
> >  
> > > Thanks a lot for all the clarification, I’ll definitely spend more time in the manual. This is really helpful, I’ll take another stab at it.  
> > >  
> > > Thanks again!  
> > >  
> > > --  
> > > Desmond
> > >  
> > >  
> > > On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
> > >  
> > > > Yes, you're missing quite a lot. Please read the manual. VirtualGL  
> > > > needs two X servers-- a "3D X server" and a "2D X server." The 3D X  
> > > > server has to be a real X server attached to the GPU. It can be  
> > > > headless, but it can't be virtual. The 2D X server can be a virtual X  
> > > > server, such as VNC or Xvfb.
> > > >  
> > > > Secondly, the way you launch a 3D application in VGL is by using vglrun,  
> > > > not vglclient. The instructions do not ever even specify calling  
> > > > vglclient directly. vglclient is generally called from vglconnect, but  
> > > > you don't need to ever use either of those unless your 2D X server and  
> > > > 3D X server are on different machines (which they aren't.)
> > > >  
> > > > I'll summarize the steps you need to take (but please do read the manual.)  
> > > > -- Set up the 3D X server, which should be a "real" X server connected  
> > > > to the GPU. Make sure the GLX extension works by running glxinfo  
> > > > against the 3D X server and verifying that it is not using a software  
> > > > OpenGL renderer (Mesa.)
> > > > -- Run vglserver_config to give VirtualGL access to the 3D X server  
> > > > (this is only necessary if the 3D X server is running some sort of login  
> > > > manager under the root account, like gdm or xdm. Otherwise, if you are  
> > > > just starting the 3D X server under your user account, then you should  
> > > > already have permission to access it.)
> > > > -- Start Xvfb (the 2D X server.)
> > > > -- Set DISPLAY to point to your Xvfb instance.
> > > > -- vglrun {application}
> > > >  
> > > >  
> > > > On 11/6/13 4:21 PM, Stealthy wrote:
> > > > > I am not sure I going through the right steps to enable virtualgl. Post
> > > > > installation I started the Xvfb server with this command:
> > > > >  
> > > > > *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
> > > > >  
> > > > > I then started what I think is virtualgl with this command expecting any  
> > > > > opengl application running after that would use the GPU:
> > > > >  
> > > > > /opt/VirtualGL/bin/vglclient
> > > > >  
> > > > > The application still only uses the CPU.
> > > > >  
> > > > > Am I missing something?  
> > > > >  
> > > > > Thanks for the help btw.
> > > > > --
> > > > > Desmond
> > > > >  
> > > > > On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
> > > > >  
> > > > > > "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be  
> > > > > > specific as to the error messages you are encountering or what is not
> > > > > > working as you expect.
> > > > > >  
> > > > > >  
> > > > > > On 11/6/13 2:16 PM, Stealthy wrote:
> > > > > > >  
> > > > > > > > Hi guys,
> > > > > > > >  
> > > > > > > > Hope I can get some help on this one. I have an ec2 instance running
> > > > > > > > using a GPU cluster. I am capturing the Xvfb virtual screen using
> > > > > > > > ffmpeg and would like to have my OPENGL enabled application use the
> > > > > > > > GPU to render. It’s currently rendering everything using the CPU.
> > > > > > > >  
> > > > > > > > I already installed virtualgl but I’m having some trouble intializing  
> > > > > > > > virtualgl with Xvfb and ensure the GPU is being used. Anybody could
> > > > > > > > help on this?
> > > > > > > >  
> > > > > > > > Thanks in advance
> > > > > >  
> > > > > > ------------------------------------------------------------------------------  
> > > > > > November Webinars for C, C++, Fortran Developers
> > > > > > Accelerate application performance with scalable programming models.
> > > > > > Explore
> > > > > > techniques for threading, error checking, porting, and tuning. Get the
> > > > > > most
> > > > > > from the latest Intel processors and coprocessors. See abstracts and
> > > > > > register
> > > > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > > > > _______________________________________________
> > > > > > VirtualGL-Users mailing list
> > > > > > Vir...@li... (mailto:Vir...@li...)
> > > > > > <mailto:Vir...@li...>
> > > > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > > > >  
> > > > >  
> > > > >  
> > > > >  
> > > > >  
> > > > > ------------------------------------------------------------------------------
> > > > > November Webinars for C, C++, Fortran Developers
> > > > > Accelerate application performance with scalable programming models. Explore
> > > > > techniques for threading, error checking, porting, and tuning. Get the most
> > > > > from the latest Intel processors and coprocessors. See abstracts and register
> > > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > > >  
> > > > >  
> > > > >  
> > > > > _______________________________________________
> > > > > VirtualGL-Users mailing list
> > > > > Vir...@li... (mailto:Vir...@li...)
> > > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > > >  
> > > >  
> > > >  
> > > > ------------------------------------------------------------------------------  
> > > > November Webinars for C, C++, Fortran Developers
> > > > Accelerate application performance with scalable programming models. Explore
> > > > techniques for threading, error checking, porting, and tuning. Get the most  
> > > > from the latest Intel processors and coprocessors. See abstracts and register
> > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > > _______________________________________________
> > > > VirtualGL-Users mailing list
> > > > Vir...@li... (mailto:Vir...@li...)
> > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > >  
> > > >  
> > > >  
> > >  
> > >  
> >  
> >  
> > ------------------------------------------------------------------------------
> > CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> > Learn Why More Businesses Are Choosing CenturyLink Cloud For
> > Critical Workloads, Development Environments & Everything In Between.
> > Get a Quote or Start a Free Trial Today.
> > http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> > _______________________________________________
> > VirtualGL-Users mailing list
> > Vir...@li... (mailto:Vir...@li...)
> > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >  
>  
>  
>  
> --  
> Kevin Van Workum, PhD
> Sabalcore Computing Inc.
> "Where Data Becomes Discovery"
> http://www.sabalcore.com
> 877-492-8027 ext. 11  
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.  
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>  
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li... (mailto:Vir...@li...)
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>  
>  

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[DRC <dco...@us...>]

Yes, the operative word there is "or". VNC or Xvfb or another X proxy. You don't have to use VNC as your 2D X Server. Xvfb does work, although Xvfb has no built-in image transport, so you would be on your own for delivering the pixels from server to client.

> On Jan 10, 2014, at 12:53 PM, Stealthy <ste...@gm...> wrote:
> 
> Checked my xorg.conf file and the driver is nvidia, see below.
> Section "Monitor"
>     Identifier     "Monitor0"
>     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"
>     BusID          "PCI:0:3:0"
> EndSection
> 
> 
> Section "Screen"
>     Identifier     "Screen0"
>     Device         "Device0"
>     Monitor        "Monitor0"
>     DefaultDepth    24
>     Option         "UseDisplayDevice" "none"
>     SubSection     "Display"
>         Virtual     1024 768
>         Depth       24
>     EndSubSection
> EndSection
> 
>  Not sure why it’s still using mesa.
> 
>  As to the use of turbovnc this is what I was told when I was setting this up: 
> "VirtualGL needs two X servers-- a "3D X server" and a "2D X server." The 3D X
> server has to be a real X server attached to the GPU. It can be
> headless, but it can't be virtual. The 2D X server can be a virtual X
> server, such as VNC or Xvfb.”
> 
> Isn’t this true? Are you saying I can just run everything through Xvfb and still have my application run through virtualgl and use the gpu? That would make things quite easier if that is possible.
> 
> Thanks!
> -- 
> Desmond
>> On Friday, January 10, 2014 at 3:53 AM, Kevin Van Workum wrote:
>> 
>> It looks like your 3D X server (on :0) is not using the driver for your card. Make sure the "Device" section of xorg.conf has the correct driver and BusID specified. Secondly, I don't understand why you are using turbovnc. If you're trying to do captures with Xvfb, what's turobvnc for?
>> 
>> 
>>> On Thu, Jan 9, 2014 at 7:36 PM, Stealthy <ste...@gm...> wrote:
>>> Hi guys,
>>> 
>>> So I am trying to run a 3d application using OPENGL through an EC2 server. The ultimate goal being to capture the screen using Xvfb.
>>> 
>>> I installed virtualgl and turbovnc. when I run
>>> /opt/VirtualGL/bin/glxinfo -display :0 -c
>>> 
>>> I get this
>>> 
>>> name of display: :0
>>> display: :0  screen: 0
>>> direct rendering: Yes
>>> server glx version string: 1.4 Mesa 9.2.4
>>> server glx extensions:
>>>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap, 
>>>     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer, 
>>>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig, 
>>>     GLX_SGIX_pbuffer
>>> 
>>> client glx version string: 1.4 Mesa 9.2.4
>>> client glx extensions:
>>>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap, 
>>>     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer, 
>>>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig, 
>>>     GLX_SGIX_pbuffer
>>> 
>>> Obviously I don’t want to use Mesa. Do I have to start the X server in any specific manner to avoid using the MESA software renderer. Did I miss a step.
>>> 
>>> Thanks for your help guys!
>>> -- 
>>> Desmond
>>> 
>>>> On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:
>>>> 
>>>> Thanks a lot for all the clarification, I’ll definitely spend more time in the manual. This is really helpful, I’ll take another stab at it.
>>>> 
>>>> Thanks again!
>>>> 
>>>> -- 
>>>> Desmond
>>>>> On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
>>>>> 
>>>>> Yes, you're missing quite a lot. Please read the manual. VirtualGL
>>>>> needs two X servers-- a "3D X server" and a "2D X server." The 3D X
>>>>> server has to be a real X server attached to the GPU. It can be
>>>>> headless, but it can't be virtual. The 2D X server can be a virtual X
>>>>> server, such as VNC or Xvfb.
>>>>> 
>>>>> Secondly, the way you launch a 3D application in VGL is by using vglrun,
>>>>> not vglclient. The instructions do not ever even specify calling 
>>>>> vglclient directly. vglclient is generally called from vglconnect, but
>>>>> you don't need to ever use either of those unless your 2D X server and
>>>>> 3D X server are on different machines (which they aren't.)
>>>>> 
>>>>> I'll summarize the steps you need to take (but please do read the manual.)
>>>>> -- Set up the 3D X server, which should be a "real" X server connected
>>>>> to the GPU. Make sure the GLX extension works by running glxinfo
>>>>> against the 3D X server and verifying that it is not using a software
>>>>> OpenGL renderer (Mesa.)
>>>>> -- Run vglserver_config to give VirtualGL access to the 3D X server
>>>>> (this is only necessary if the 3D X server is running some sort of login
>>>>> manager under the root account, like gdm or xdm. Otherwise, if you are
>>>>> just starting the 3D X server under your user account, then you should
>>>>> already have permission to access it.)
>>>>> -- Start Xvfb (the 2D X server.)
>>>>> -- Set DISPLAY to point to your Xvfb instance.
>>>>> -- vglrun {application}
>>>>> 
>>>>> 
>>>>>> On 11/6/13 4:21 PM, Stealthy wrote:
>>>>>> I am not sure I going through the right steps to enable virtualgl. Post
>>>>>> installation I started the Xvfb server with this command:
>>>>>> 
>>>>>> *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
>>>>>> 
>>>>>> I then started what I think is virtualgl with this command expecting any
>>>>>> opengl application running after that would use the GPU:
>>>>>> 
>>>>>> /opt/VirtualGL/bin/vglclient
>>>>>> 
>>>>>> The application still only uses the CPU.
>>>>>> 
>>>>>> Am I missing something?
>>>>>> 
>>>>>> Thanks for the help btw.
>>>>>> --
>>>>>> Desmond
>>>>>> 
>>>>>>> On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
>>>>>>> 
>>>>>>> "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be
>>>>>>> specific as to the error messages you are encountering or what is not
>>>>>>> working as you expect.
>>>>>>> 
>>>>>>> 
>>>>>>>> On 11/6/13 2:16 PM, Stealthy wrote:
>>>>>>>> 
>>>>>>>>> Hi guys,
>>>>>>>>> 
>>>>>>>>> Hope I can get some help on this one. I have an ec2 instance running
>>>>>>>>> using a GPU cluster. I am capturing the Xvfb virtual screen using
>>>>>>>>> ffmpeg and would like to have my OPENGL enabled application use the
>>>>>>>>> GPU to render. It’s currently rendering everything using the CPU.
>>>>>>>>> 
>>>>>>>>> I already installed virtualgl but I’m having some trouble intializing
>>>>>>>>> virtualgl with Xvfb and ensure the GPU is being used. Anybody could
>>>>>>>>> help on this?
>>>>>>>>> 
>>>>>>>>> Thanks in advance
>>>>>>> 
>>>>>>> ------------------------------------------------------------------------------
>>>>>>> November Webinars for C, C++, Fortran Developers
>>>>>>> Accelerate application performance with scalable programming models.
>>>>>>> Explore
>>>>>>> techniques for threading, error checking, porting, and tuning. Get the
>>>>>>> most
>>>>>>> from the latest Intel processors and coprocessors. See abstracts and
>>>>>>> register
>>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>>>>>>> _______________________________________________
>>>>>>> VirtualGL-Users mailing list
>>>>>>> Vir...@li...
>>>>>>> <mailto:Vir...@li...>
>>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> ------------------------------------------------------------------------------
>>>>>> November Webinars for C, C++, Fortran Developers
>>>>>> Accelerate application performance with scalable programming models. Explore
>>>>>> techniques for threading, error checking, porting, and tuning. Get the most
>>>>>> from the latest Intel processors and coprocessors. See abstracts and register
>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> VirtualGL-Users mailing list
>>>>>> Vir...@li...
>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>>> 
>>>>> ------------------------------------------------------------------------------
>>>>> November Webinars for C, C++, Fortran Developers
>>>>> Accelerate application performance with scalable programming models. Explore
>>>>> techniques for threading, error checking, porting, and tuning. Get the most
>>>>> from the latest Intel processors and coprocessors. See abstracts and register
>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>>>>> _______________________________________________
>>>>> VirtualGL-Users mailing list
>>>>> Vir...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>> 
>>> 
>>> ------------------------------------------------------------------------------
>>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
>>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
>>> Critical Workloads, Development Environments & Everything In Between.
>>> Get a Quote or Start a Free Trial Today.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> VirtualGL-Users mailing list
>>> Vir...@li...
>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>> 
>> 
>> 
>> -- 
>> Kevin Van Workum, PhD
>> Sabalcore Computing Inc.
>> "Where Data Becomes Discovery"
>> http://www.sabalcore.com
>> 877-492-8027 ext. 11
>> 
>> 
>> ------------------------------------------------------------------------------
>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
>> Critical Workloads, Development Environments & Everything In Between.
>> Get a Quote or Start a Free Trial Today.
>> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>> _______________________________________________
>> VirtualGL-Users mailing list
>> Vir...@li...
>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> 
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today. 
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Kevin V. W. <va...@sa...>]

DRC was correct. You need a 3D X server, which is your Xorg X11 server. And
you need a 2D server; either VNC or Xvfb, not both. If you goal is to use
Xvfb to capture images, then use Xvfb as the 2D server.


On Fri, Jan 10, 2014 at 1:53 PM, Stealthy <ste...@gm...> wrote:

> Checked my xorg.conf file and the driver is nvidia, see below.
>
> Section "Monitor"
>
>     Identifier     "Monitor0"
>
>     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"
>
>     BusID          "PCI:0:3:0"
>
> EndSection
>
>
>
> Section "Screen"
>
>     Identifier     "Screen0"
>
>     Device         "Device0"
>
>     Monitor        "Monitor0"
>
>     DefaultDepth    24
>
>     Option         "UseDisplayDevice" "none"
>
>     SubSection     "Display"
>
>         Virtual     1024 768
>
>         Depth       24
>
>     EndSubSection
>
> EndSection
>
>  Not sure why it’s still using mesa.
>
>  As to the use of turbovnc this is what I was told when I was setting this
> up:
> "VirtualGL needs two X servers-- a "3D X server" and a "2D X server." The
> 3D X
>
> server has to be a real X server attached to the GPU. It can be
>
> headless, but it can't be virtual. The 2D X server can be a virtual X
>
> server, such as VNC or Xvfb.”
>
>
> Isn’t this true? Are you saying I can just run everything through Xvfb and
> still have my application run through virtualgl and use the gpu? That would
> make things quite easier if that is possible.
>
> Thanks!
> --
> Desmond
>
> On Friday, January 10, 2014 at 3:53 AM, Kevin Van Workum wrote:
>
> It looks like your 3D X server (on :0) is not using the driver for your
> card. Make sure the "Device" section of xorg.conf has the correct driver
> and BusID specified. Secondly, I don't understand why you are using
> turbovnc. If you're trying to do captures with Xvfb, what's turobvnc for?
>
>
> On Thu, Jan 9, 2014 at 7:36 PM, Stealthy <ste...@gm...>wrote:
>
>  Hi guys,
>
> So I am trying to run a 3d application using OPENGL through an EC2 server.
> The ultimate goal being to capture the screen using Xvfb.
>
> I installed virtualgl and turbovnc. when I run
> /opt/VirtualGL/bin/glxinfo -display :0 -c
>
> I get this
>
> name of display: :0
>
> display: :0  screen: 0
>
> direct rendering: Yes
>
> server glx version string: 1.4 Mesa 9.2.4
>
> server glx extensions:
>
>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
>
>     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,
>
>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,
>
>     GLX_SGIX_pbuffer
>
>
> client glx version string: 1.4 Mesa 9.2.4
>
> client glx extensions:
>
>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
>
>     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,
>
>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,
>
>     GLX_SGIX_pbuffer
>
> Obviously I don’t want to use Mesa. Do I have to start the X server in any
> specific manner to avoid using the MESA software renderer. Did I miss a
> step.
>
> Thanks for your help guys!
> --
> Desmond
>
> On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:
>
>  Thanks a lot for all the clarification, I’ll definitely spend more time
> in the manual. This is really helpful, I’ll take another stab at it.
>
> Thanks again!
>
> --
> Desmond
>
> On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
>
> Yes, you're missing quite a lot. Please read the manual. VirtualGL
> needs two X servers-- a "3D X server" and a "2D X server." The 3D X
> server has to be a real X server attached to the GPU. It can be
> headless, but it can't be virtual. The 2D X server can be a virtual X
> server, such as VNC or Xvfb.
>
> Secondly, the way you launch a 3D application in VGL is by using vglrun,
> not vglclient. The instructions do not ever even specify calling
> vglclient directly. vglclient is generally called from vglconnect, but
> you don't need to ever use either of those unless your 2D X server and
> 3D X server are on different machines (which they aren't.)
>
> I'll summarize the steps you need to take (but please do read the manual.)
> -- Set up the 3D X server, which should be a "real" X server connected
> to the GPU. Make sure the GLX extension works by running glxinfo
> against the 3D X server and verifying that it is not using a software
> OpenGL renderer (Mesa.)
> -- Run vglserver_config to give VirtualGL access to the 3D X server
> (this is only necessary if the 3D X server is running some sort of login
> manager under the root account, like gdm or xdm. Otherwise, if you are
> just starting the 3D X server under your user account, then you should
> already have permission to access it.)
> -- Start Xvfb (the 2D X server.)
> -- Set DISPLAY to point to your Xvfb instance.
> -- vglrun {application}
>
>
> On 11/6/13 4:21 PM, Stealthy wrote:
>
> I am not sure I going through the right steps to enable virtualgl. Post
> installation I started the Xvfb server with this command:
>
> *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
>
> I then started what I think is virtualgl with this command expecting any
> opengl application running after that would use the GPU:
>
> /opt/VirtualGL/bin/vglclient
>
> The application still only uses the CPU.
>
> Am I missing something?
>
> Thanks for the help btw.
> --
> Desmond
>
> On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
>
> "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be
> specific as to the error messages you are encountering or what is not
> working as you expect.
>
>
> On 11/6/13 2:16 PM, Stealthy wrote:
>
>
> Hi guys,
>
> Hope I can get some help on this one. I have an ec2 instance running
> using a GPU cluster. I am capturing the Xvfb virtual screen using
> ffmpeg and would like to have my OPENGL enabled application use the
> GPU to render. It’s currently rendering everything using the CPU.
>
> I already installed virtualgl but I’m having some trouble intializing
> virtualgl with Xvfb and ensure the GPU is being used. Anybody could
> help on this?
>
> Thanks in advance
>
>
>
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models.
> Explore
> techniques for threading, error checking, porting, and tuning. Get the
> most
> from the latest Intel processors and coprocessors. See abstracts and
> register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> <mailto:Vir...@li...<Vir...@li...>
> >
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
>
>
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models.
> Explore
> techniques for threading, error checking, porting, and tuning. Get the most
> from the latest Intel processors and coprocessors. See abstracts and
> register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>
>
>
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
> ------------------------------------------------------------------------------
> November Webinars for C, C++, Fortran Developers
> Accelerate application performance with scalable programming models.
> Explore
> techniques for threading, error checking, porting, and tuning. Get the
> most
> from the latest Intel processors and coprocessors. See abstracts and
> register
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
>
>
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
>
> --
> Kevin Van Workum, PhD
> Sabalcore Computing Inc.
> "Where Data Becomes Discovery"
> http://www.sabalcore.com
> 877-492-8027 ext. 11
>
>
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
>
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>


-- 
Kevin Van Workum, PhD
Sabalcore Computing Inc.
"Where Data Becomes Discovery"
http://www.sabalcore.com
877-492-8027 ext. 11

-- 

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Stealthy <ste...@gm...>]

Sounds good. So I’ll just start my X server like usual and point my Xvfb server to the same display?

About the Mesa issue below. You see anything abnormal in my xorg.conf file below? Don’t understand why it’s still using Mesa when it should be using hardware acceleration through the GPU.

Thanks for the help guys!

--  
Desmond




On Friday, January 10, 2014 at 11:53 AM, Kevin Van Workum wrote:

> DRC was correct. You need a 3D X server, which is your Xorg X11 server. And you need a 2D server; either VNC or Xvfb, not both. If you goal is to use Xvfb to capture images, then use Xvfb as the 2D server.
>  
>  
> On Fri, Jan 10, 2014 at 1:53 PM, Stealthy <ste...@gm... (mailto:ste...@gm...)> wrote:
> > Checked my xorg.conf file and the driver is nvidia, see below.
> >  
> > Section "Monitor"
> >  
> >  
> >     Identifier     "Monitor0"
> >  
> >  
> >     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"
> >  
> >  
> >     BusID          "PCI:0:3:0"
> >  
> >  
> > EndSection
> >  
> >  
> >  
> >  
> >  
> >  
> >  
> >  
> >  
> >  
> > Section "Screen"
> >  
> >  
> >     Identifier     "Screen0"
> >  
> >  
> >     Device         "Device0"
> >  
> >  
> >     Monitor        "Monitor0"
> >  
> >  
> >     DefaultDepth    24
> >  
> >  
> >     Option         "UseDisplayDevice" "none"
> >  
> >  
> >     SubSection     "Display"
> >  
> >  
> >         Virtual     1024 768
> >  
> >  
> >         Depth       24
> >  
> >  
> >     EndSubSection
> >  
> >  
> > EndSection
> >  
> >  
> >  
> >  Not sure why it’s still using mesa.
> >  
> >  As to the use of turbovnc this is what I was told when I was setting this up:   
> > "VirtualGL needs two X servers-- a "3D X server" and a "2D X server." The 3D X
> >  
> > server has to be a real X server attached to the GPU. It can be
> >  
> >  
> > headless, but it can't be virtual. The 2D X server can be a virtual X
> >  
> >  
> >  
> > server, such as VNC or Xvfb.”
> >  
> >  
> >  
> >  
> >  
> >  
> > Isn’t this true? Are you saying I can just run everything through Xvfb and still have my application run through virtualgl and use the gpu? That would make things quite easier if that is possible.
> >  
> >  
> > Thanks!
> > --  
> > Desmond
> >  
> >  
> > On Friday, January 10, 2014 at 3:53 AM, Kevin Van Workum wrote:
> >  
> >  
> >  
> > > It looks like your 3D X server (on :0) is not using the driver for your card. Make sure the "Device" section of xorg.conf has the correct driver and BusID specified. Secondly, I don't understand why you are using turbovnc. If you're trying to do captures with Xvfb, what's turobvnc for?
> > >  
> > >  
> > > On Thu, Jan 9, 2014 at 7:36 PM, Stealthy <ste...@gm... (mailto:ste...@gm...)> wrote:
> > > > Hi guys,  
> > > >  
> > > > So I am trying to run a 3d application using OPENGL through an EC2 server. The ultimate goal being to capture the screen using Xvfb.
> > > >  
> > > > I installed virtualgl and turbovnc. when I run  
> > > > /opt/VirtualGL/bin/glxinfo -display :0 -c
> > > >  
> > > > I get this
> > > >  
> > > >  
> > > > name of display: :0
> > > >  
> > > >  
> > > > display: :0  screen: 0
> > > >  
> > > >  
> > > > direct rendering: Yes
> > > >  
> > > >  
> > > > server glx version string: 1.4 Mesa 9.2.4
> > > >  
> > > >  
> > > > server glx extensions:
> > > >  
> > > >  
> > > >     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,  
> > > >  
> > > >  
> > > >     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,  
> > > >  
> > > >  
> > > >     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,  
> > > >  
> > > >  
> > > >     GLX_SGIX_pbuffer
> > > >  
> > > >  
> > > >  
> > > >  
> > > >  
> > > >  
> > > > client glx version string: 1.4 Mesa 9.2.4
> > > >  
> > > >  
> > > > client glx extensions:
> > > >  
> > > >  
> > > >     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,  
> > > >  
> > > >  
> > > >     GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_copy_sub_buffer,  
> > > >  
> > > >  
> > > >     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers, GLX_SGIX_fbconfig,  
> > > >  
> > > >  
> > > >     GLX_SGIX_pbuffer
> > > >  
> > > >  
> > > >  
> > > > Obviously I don’t want to use Mesa. Do I have to start the X server in any specific manner to avoid using the MESA software renderer. Did I miss a step.
> > > >  
> > > > Thanks for your help guys!  
> > > > --  
> > > > Desmond
> > > >  
> > > >  
> > > > On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:
> > > >  
> > > > > Thanks a lot for all the clarification, I’ll definitely spend more time in the manual. This is really helpful, I’ll take another stab at it.  
> > > > >  
> > > > > Thanks again!  
> > > > >  
> > > > > --  
> > > > > Desmond
> > > > >  
> > > > >  
> > > > > On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
> > > > >  
> > > > > > Yes, you're missing quite a lot. Please read the manual. VirtualGL  
> > > > > > needs two X servers-- a "3D X server" and a "2D X server." The 3D X  
> > > > > > server has to be a real X server attached to the GPU. It can be  
> > > > > > headless, but it can't be virtual. The 2D X server can be a virtual X  
> > > > > > server, such as VNC or Xvfb.
> > > > > >  
> > > > > > Secondly, the way you launch a 3D application in VGL is by using vglrun,  
> > > > > > not vglclient. The instructions do not ever even specify calling  
> > > > > > vglclient directly. vglclient is generally called from vglconnect, but  
> > > > > > you don't need to ever use either of those unless your 2D X server and  
> > > > > > 3D X server are on different machines (which they aren't.)
> > > > > >  
> > > > > > I'll summarize the steps you need to take (but please do read the manual.)  
> > > > > > -- Set up the 3D X server, which should be a "real" X server connected  
> > > > > > to the GPU. Make sure the GLX extension works by running glxinfo  
> > > > > > against the 3D X server and verifying that it is not using a software  
> > > > > > OpenGL renderer (Mesa.)
> > > > > > -- Run vglserver_config to give VirtualGL access to the 3D X server  
> > > > > > (this is only necessary if the 3D X server is running some sort of login  
> > > > > > manager under the root account, like gdm or xdm. Otherwise, if you are  
> > > > > > just starting the 3D X server under your user account, then you should  
> > > > > > already have permission to access it.)
> > > > > > -- Start Xvfb (the 2D X server.)
> > > > > > -- Set DISPLAY to point to your Xvfb instance.
> > > > > > -- vglrun {application}
> > > > > >  
> > > > > >  
> > > > > > On 11/6/13 4:21 PM, Stealthy wrote:
> > > > > > > I am not sure I going through the right steps to enable virtualgl. Post
> > > > > > > installation I started the Xvfb server with this command:
> > > > > > >  
> > > > > > > *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render -noreset &
> > > > > > >  
> > > > > > > I then started what I think is virtualgl with this command expecting any  
> > > > > > > opengl application running after that would use the GPU:
> > > > > > >  
> > > > > > > /opt/VirtualGL/bin/vglclient
> > > > > > >  
> > > > > > > The application still only uses the CPU.
> > > > > > >  
> > > > > > > Am I missing something?  
> > > > > > >  
> > > > > > > Thanks for the help btw.
> > > > > > > --
> > > > > > > Desmond
> > > > > > >  
> > > > > > > On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
> > > > > > >  
> > > > > > > > "Trouble initializing VirtualGL with Xvfb" tells me nothing. Please be  
> > > > > > > > specific as to the error messages you are encountering or what is not
> > > > > > > > working as you expect.
> > > > > > > >  
> > > > > > > >  
> > > > > > > > On 11/6/13 2:16 PM, Stealthy wrote:
> > > > > > > > >  
> > > > > > > > > > Hi guys,
> > > > > > > > > >  
> > > > > > > > > > Hope I can get some help on this one. I have an ec2 instance running
> > > > > > > > > > using a GPU cluster. I am capturing the Xvfb virtual screen using
> > > > > > > > > > ffmpeg and would like to have my OPENGL enabled application use the
> > > > > > > > > > GPU to render. It’s currently rendering everything using the CPU.
> > > > > > > > > >  
> > > > > > > > > > I already installed virtualgl but I’m having some trouble intializing  
> > > > > > > > > > virtualgl with Xvfb and ensure the GPU is being used. Anybody could
> > > > > > > > > > help on this?
> > > > > > > > > >  
> > > > > > > > > > Thanks in advance
> > > > > > > >  
> > > > > > > > ------------------------------------------------------------------------------  
> > > > > > > > November Webinars for C, C++, Fortran Developers
> > > > > > > > Accelerate application performance with scalable programming models.
> > > > > > > > Explore
> > > > > > > > techniques for threading, error checking, porting, and tuning. Get the
> > > > > > > > most
> > > > > > > > from the latest Intel processors and coprocessors. See abstracts and
> > > > > > > > register
> > > > > > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > > > > > > _______________________________________________
> > > > > > > > VirtualGL-Users mailing list
> > > > > > > > Vir...@li... (mailto:Vir...@li...)
> > > > > > > > <mailto:Vir...@li...>
> > > > > > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > > > > > >  
> > > > > > >  
> > > > > > >  
> > > > > > >  
> > > > > > >  
> > > > > > > ------------------------------------------------------------------------------
> > > > > > > November Webinars for C, C++, Fortran Developers
> > > > > > > Accelerate application performance with scalable programming models. Explore
> > > > > > > techniques for threading, error checking, porting, and tuning. Get the most
> > > > > > > from the latest Intel processors and coprocessors. See abstracts and register
> > > > > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > > > > >  
> > > > > > >  
> > > > > > >  
> > > > > > > _______________________________________________
> > > > > > > VirtualGL-Users mailing list
> > > > > > > Vir...@li... (mailto:Vir...@li...)
> > > > > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > > > > >  
> > > > > >  
> > > > > >  
> > > > > > ------------------------------------------------------------------------------  
> > > > > > November Webinars for C, C++, Fortran Developers
> > > > > > Accelerate application performance with scalable programming models. Explore
> > > > > > techniques for threading, error checking, porting, and tuning. Get the most  
> > > > > > from the latest Intel processors and coprocessors. See abstracts and register
> > > > > > http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> > > > > > _______________________________________________
> > > > > > VirtualGL-Users mailing list
> > > > > > Vir...@li... (mailto:Vir...@li...)
> > > > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > > > >  
> > > > > >  
> > > > > >  
> > > > >  
> > > > >  
> > > >  
> > > >  
> > > > ------------------------------------------------------------------------------
> > > > CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> > > > Learn Why More Businesses Are Choosing CenturyLink Cloud For
> > > > Critical Workloads, Development Environments & Everything In Between.
> > > > Get a Quote or Start a Free Trial Today.
> > > > http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> > > > _______________________________________________
> > > > VirtualGL-Users mailing list
> > > > Vir...@li... (mailto:Vir...@li...)
> > > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > > >  
> > >  
> > >  
> > >  
> > > --  
> > > Kevin Van Workum, PhD
> > > Sabalcore Computing Inc.
> > > "Where Data Becomes Discovery"
> > > http://www.sabalcore.com
> > > 877-492-8027 ext. 11  
> > > ------------------------------------------------------------------------------
> > > CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> > > Learn Why More Businesses Are Choosing CenturyLink Cloud For
> > > Critical Workloads, Development Environments & Everything In Between.
> > > Get a Quote or Start a Free Trial Today.  
> > > http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> > >  
> > > _______________________________________________
> > > VirtualGL-Users mailing list
> > > Vir...@li... (mailto:Vir...@li...)
> > > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> > >  
> > >  
> > >  
> > >  
> >  
> >  
> >  
> > ------------------------------------------------------------------------------
> > CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> > Learn Why More Businesses Are Choosing CenturyLink Cloud For
> > Critical Workloads, Development Environments & Everything In Between.
> > Get a Quote or Start a Free Trial Today.
> > http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> > _______________________________________________
> > VirtualGL-Users mailing list
> > Vir...@li... (mailto:Vir...@li...)
> > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >  
>  
>  
>  
> --  
> Kevin Van Workum, PhD
> Sabalcore Computing Inc.
> "Where Data Becomes Discovery"
> http://www.sabalcore.com
> 877-492-8027 ext. 11  
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.  
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>  
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li... (mailto:Vir...@li...)
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>  
>  

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[DRC <dco...@us...>]

No, not the same display.  The 3D X server and 2D X server need to be on 
different displays.  Usually the 3D X server is on :0 if you have just 
one of them, and the 2D X server is on some other display number.

I would suggest looking at the X server log to see if perhaps the nVidia 
driver is having difficulty starting up.  I personally have no 
first-hand experience with EC2 instances.  I know through user reports 
that VGL works with them, but I haven't ever actually tested this myself.


On 1/10/14 1:55 PM, Stealthy wrote:
>
> Sounds good. So I’ll just start my X server like usual and point my Xvfb
> server to the same display?
>
> About the Mesa issue below. You see anything abnormal in my xorg.conf
> file below? Don’t understand why it’s still using Mesa when it should be
> using hardware acceleration through the GPU.
>
> Thanks for the help guys!
>
> --
> Desmond
>
> On Friday, January 10, 2014 at 11:53 AM, Kevin Van Workum wrote:
>
>> DRC was correct. You need a 3D X server, which is your Xorg X11
>> server. And you need a 2D server; either VNC or Xvfb, not both. If you
>> goal is to use Xvfb to capture images, then use Xvfb as the 2D server.
>>
>>
>> On Fri, Jan 10, 2014 at 1:53 PM, Stealthy <ste...@gm...
>> <mailto:ste...@gm...>> wrote:
>>> Checked my xorg.conf file and the driver is nvidia, see below.
>>>
>>> Section "Monitor"
>>>
>>> Identifier"Monitor0"
>>>
>>> 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"
>>>
>>> BusID          "PCI:0:3:0"
>>>
>>> EndSection
>>>
>>>
>>>
>>> Section "Screen"
>>>
>>> Identifier"Screen0"
>>>
>>> Device "Device0"
>>>
>>> Monitor "Monitor0"
>>>
>>> DefaultDepth    24
>>>
>>> Option"UseDisplayDevice" "none"
>>>
>>>     SubSection     "Display"
>>>
>>> Virtual     1024 768
>>>
>>> Depth       24
>>>
>>>     EndSubSection
>>>
>>> EndSection
>>>
>>>
>>>  Not sure why it’s still using mesa.
>>>
>>>  As to the use of turbovnc this is what I was told when I was setting
>>> this up:
>>> "VirtualGL needs two X servers-- a "3D X server" and a "2D X server."
>>> The 3D X
>>>
>>> server has to be a real X server attached to the GPU. It can be
>>>
>>> headless, but it can't be virtual. The 2D X server can be a virtual X
>>>
>>> server, such as VNC or Xvfb.”
>>>
>>>
>>> Isn’t this true? Are you saying I can just run everything through
>>> Xvfb and still have my application run through virtualgl and use the
>>> gpu? That would make things quite easier if that is possible.
>>>
>>>
>>> Thanks!
>>> --
>>> Desmond
>>>
>>> On Friday, January 10, 2014 at 3:53 AM, Kevin Van Workum wrote:
>>>
>>>> It looks like your 3D X server (on :0) is not using the driver for
>>>> your card. Make sure the "Device" section of xorg.conf has the
>>>> correct driver and BusID specified. Secondly, I don't understand why
>>>> you are using turbovnc. If you're trying to do captures with Xvfb,
>>>> what's turobvnc for?
>>>>
>>>>
>>>> On Thu, Jan 9, 2014 at 7:36 PM, Stealthy <ste...@gm...
>>>> <mailto:ste...@gm...>> wrote:
>>>>> Hi guys,
>>>>>
>>>>> So I am trying to run a 3d application using OPENGL through an EC2
>>>>> server. The ultimate goal being to capture the screen using Xvfb.
>>>>>
>>>>> I installed virtualgl and turbovnc. when I run
>>>>> /opt/VirtualGL/bin/glxinfo -display :0 -c
>>>>>
>>>>> I get this
>>>>>
>>>>> name of display: :0
>>>>>
>>>>> display: :0  screen: 0
>>>>>
>>>>> direct rendering: Yes
>>>>>
>>>>> server glx version string: 1.4 Mesa 9.2.4
>>>>>
>>>>> server glx extensions:
>>>>>
>>>>>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
>>>>>
>>>>>     GLX_EXT_visual_info, GLX_EXT_visual_rating,
>>>>> GLX_MESA_copy_sub_buffer,
>>>>>
>>>>>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers,
>>>>> GLX_SGIX_fbconfig,
>>>>>
>>>>>     GLX_SGIX_pbuffer
>>>>>
>>>>>
>>>>> client glx version string: 1.4 Mesa 9.2.4
>>>>>
>>>>> client glx extensions:
>>>>>
>>>>>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
>>>>>
>>>>>     GLX_EXT_visual_info, GLX_EXT_visual_rating,
>>>>> GLX_MESA_copy_sub_buffer,
>>>>>
>>>>>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers,
>>>>> GLX_SGIX_fbconfig,
>>>>>
>>>>>     GLX_SGIX_pbuffer
>>>>>
>>>>>
>>>>> Obviously I don’t want to use Mesa. Do I have to start the X server
>>>>> in any specific manner to avoid using the MESA software renderer.
>>>>> Did I miss a step.
>>>>>
>>>>> Thanks for your help guys!
>>>>> --
>>>>> Desmond
>>>>>
>>>>> On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:
>>>>>
>>>>>> Thanks a lot for all the clarification, I’ll definitely spend more
>>>>>> time in the manual. This is really helpful, I’ll take another stab
>>>>>> at it.
>>>>>>
>>>>>> Thanks again!
>>>>>>
>>>>>> --
>>>>>> Desmond
>>>>>>
>>>>>> On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
>>>>>>
>>>>>>> Yes, you're missing quite a lot. Please read the manual. VirtualGL
>>>>>>> needs two X servers-- a "3D X server" and a "2D X server." The 3D X
>>>>>>> server has to be a real X server attached to the GPU. It can be
>>>>>>> headless, but it can't be virtual. The 2D X server can be a
>>>>>>> virtual X
>>>>>>> server, such as VNC or Xvfb.
>>>>>>>
>>>>>>> Secondly, the way you launch a 3D application in VGL is by using
>>>>>>> vglrun,
>>>>>>> not vglclient. The instructions do not ever even specify calling
>>>>>>> vglclient directly. vglclient is generally called from
>>>>>>> vglconnect, but
>>>>>>> you don't need to ever use either of those unless your 2D X
>>>>>>> server and
>>>>>>> 3D X server are on different machines (which they aren't.)
>>>>>>>
>>>>>>> I'll summarize the steps you need to take (but please do read the
>>>>>>> manual.)
>>>>>>> -- Set up the 3D X server, which should be a "real" X server
>>>>>>> connected
>>>>>>> to the GPU. Make sure the GLX extension works by running glxinfo
>>>>>>> against the 3D X server and verifying that it is not using a
>>>>>>> software
>>>>>>> OpenGL renderer (Mesa.)
>>>>>>> -- Run vglserver_config to give VirtualGL access to the 3D X server
>>>>>>> (this is only necessary if the 3D X server is running some sort
>>>>>>> of login
>>>>>>> manager under the root account, like gdm or xdm. Otherwise, if
>>>>>>> you are
>>>>>>> just starting the 3D X server under your user account, then you
>>>>>>> should
>>>>>>> already have permission to access it.)
>>>>>>> -- Start Xvfb (the 2D X server.)
>>>>>>> -- Set DISPLAY to point to your Xvfb instance.
>>>>>>> -- vglrun {application}
>>>>>>>
>>>>>>>
>>>>>>> On 11/6/13 4:21 PM, Stealthy wrote:
>>>>>>>> I am not sure I going through the right steps to enable
>>>>>>>> virtualgl. Post
>>>>>>>> installation I started the Xvfb server with this command:
>>>>>>>>
>>>>>>>> *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render
>>>>>>>> -noreset &
>>>>>>>>
>>>>>>>> I then started what I think is virtualgl with this command
>>>>>>>> expecting any
>>>>>>>> opengl application running after that would use the GPU:
>>>>>>>>
>>>>>>>> /opt/VirtualGL/bin/vglclient
>>>>>>>>
>>>>>>>> The application still only uses the CPU.
>>>>>>>>
>>>>>>>> Am I missing something?
>>>>>>>>
>>>>>>>> Thanks for the help btw.
>>>>>>>> --
>>>>>>>> Desmond
>>>>>>>>
>>>>>>>> On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
>>>>>>>>
>>>>>>>>> "Trouble initializing VirtualGL with Xvfb" tells me nothing.
>>>>>>>>> Please be
>>>>>>>>> specific as to the error messages you are encountering or what
>>>>>>>>> is not
>>>>>>>>> working as you expect.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On 11/6/13 2:16 PM, Stealthy wrote:
>>>>>>>>>>
>>>>>>>>>>> Hi guys,
>>>>>>>>>>>
>>>>>>>>>>> Hope I can get some help on this one. I have an ec2 instance
>>>>>>>>>>> running
>>>>>>>>>>> using a GPU cluster. I am capturing the Xvfb virtual screen using
>>>>>>>>>>> ffmpeg and would like to have my OPENGL enabled application
>>>>>>>>>>> use the
>>>>>>>>>>> GPU to render. It’s currently rendering everything using the CPU.
>>>>>>>>>>>
>>>>>>>>>>> I already installed virtualgl but I’m having some trouble
>>>>>>>>>>> intializing
>>>>>>>>>>> virtualgl with Xvfb and ensure the GPU is being used. Anybody
>>>>>>>>>>> could
>>>>>>>>>>> help on this?
>>>>>>>>>>>
>>>>>>>>>>> Thanks in advance
>>>>>>>>>
>>>>>>>>> ------------------------------------------------------------------------------
>>>>>>>>> November Webinars for C, C++, Fortran Developers
>>>>>>>>> Accelerate application performance with scalable programming
>>>>>>>>> models.
>>>>>>>>> Explore
>>>>>>>>> techniques for threading, error checking, porting, and tuning.
>>>>>>>>> Get the
>>>>>>>>> most
>>>>>>>>> from the latest Intel processors and coprocessors. See
>>>>>>>>> abstracts and
>>>>>>>>> register
>>>>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>>>>>>>>> _______________________________________________
>>>>>>>>> VirtualGL-Users mailing list
>>>>>>>>> Vir...@li...
>>>>>>>>> <mailto:Vir...@li...>
>>>>>>>>> <mailto:Vir...@li...>
>>>>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> ------------------------------------------------------------------------------
>>>>>>>> November Webinars for C, C++, Fortran Developers
>>>>>>>> Accelerate application performance with scalable programming
>>>>>>>> models. Explore
>>>>>>>> techniques for threading, error checking, porting, and tuning.
>>>>>>>> Get the most
>>>>>>>> from the latest Intel processors and coprocessors. See abstracts
>>>>>>>> and register
>>>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> VirtualGL-Users mailing list
>>>>>>>> Vir...@li...
>>>>>>>> <mailto:Vir...@li...>
>>>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>>>>>
>>>>>>> ------------------------------------------------------------------------------
>>>>>>> November Webinars for C, C++, Fortran Developers
>>>>>>> Accelerate application performance with scalable programming
>>>>>>> models. Explore
>>>>>>> techniques for threading, error checking, porting, and tuning.
>>>>>>> Get the most
>>>>>>> from the latest Intel processors and coprocessors. See abstracts
>>>>>>> and register
>>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
>>>>>>> _______________________________________________
>>>>>>> VirtualGL-Users mailing list
>>>>>>> Vir...@li...
>>>>>>> <mailto:Vir...@li...>
>>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>>>>
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
>>>>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
>>>>> Critical Workloads, Development Environments & Everything In Between.
>>>>> Get a Quote or Start a Free Trial Today.
>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>>>>> _______________________________________________
>>>>> VirtualGL-Users mailing list
>>>>> Vir...@li...
>>>>> <mailto:Vir...@li...>
>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> Kevin Van Workum, PhD
>>>> Sabalcore Computing Inc.
>>>> "Where Data Becomes Discovery"
>>>> http://www.sabalcore.com
>>>> 877-492-8027 ext. 11
>>>>
>>>> ------------------------------------------------------------------------------
>>>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
>>>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
>>>> Critical Workloads, Development Environments & Everything In Between.
>>>> Get a Quote or Start a Free Trial Today.
>>>> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>>>> _______________________________________________
>>>> VirtualGL-Users mailing list
>>>> Vir...@li...
>>>> <mailto:Vir...@li...>
>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
>>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
>>> Critical Workloads, Development Environments & Everything In Between.
>>> Get a Quote or Start a Free Trial Today.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> VirtualGL-Users mailing list
>>> Vir...@li...
>>> <mailto:Vir...@li...>
>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>>>
>>
>>
>>
>> --
>> Kevin Van Workum, PhD
>> Sabalcore Computing Inc.
>> "Where Data Becomes Discovery"
>> http://www.sabalcore.com
>> 877-492-8027 ext. 11
>>
>> ------------------------------------------------------------------------------
>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
>> Critical Workloads, Development Environments & Everything In Between.
>> Get a Quote or Start a Free Trial Today.
>> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>> _______________________________________________
>> VirtualGL-Users mailing list
>> Vir...@li...
>> <mailto:Vir...@li...>
>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>
>
>
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>
>
>
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[Kevin V. W. <va...@sa...>]

Can you explain why you're using Xvfb? If you're just trying to make a
video of your 3D application with ffmpeg, can't you just get it from the 3D
X server directly? For example:

ffmpeg -f x11grab -video_size cif -framerate 25 -i :0.0 /tmp/out.mpg


On Fri, Jan 10, 2014 at 3:26 PM, DRC <dco...@us...>wrote:

> No, not the same display.  The 3D X server and 2D X server need to be on
> different displays.  Usually the 3D X server is on :0 if you have just
> one of them, and the 2D X server is on some other display number.
>
> I would suggest looking at the X server log to see if perhaps the nVidia
> driver is having difficulty starting up.  I personally have no
> first-hand experience with EC2 instances.  I know through user reports
> that VGL works with them, but I haven't ever actually tested this myself.
>
>
> On 1/10/14 1:55 PM, Stealthy wrote:
> >
> > Sounds good. So I’ll just start my X server like usual and point my Xvfb
> > server to the same display?
> >
> > About the Mesa issue below. You see anything abnormal in my xorg.conf
> > file below? Don’t understand why it’s still using Mesa when it should be
> > using hardware acceleration through the GPU.
> >
> > Thanks for the help guys!
> >
> > --
> > Desmond
> >
> > On Friday, January 10, 2014 at 11:53 AM, Kevin Van Workum wrote:
> >
> >> DRC was correct. You need a 3D X server, which is your Xorg X11
> >> server. And you need a 2D server; either VNC or Xvfb, not both. If you
> >> goal is to use Xvfb to capture images, then use Xvfb as the 2D server.
> >>
> >>
> >> On Fri, Jan 10, 2014 at 1:53 PM, Stealthy <ste...@gm...
> >> <mailto:ste...@gm...>> wrote:
> >>> Checked my xorg.conf file and the driver is nvidia, see below.
> >>>
> >>> Section "Monitor"
> >>>
> >>> Identifier"Monitor0"
> >>>
> >>> 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"
> >>>
> >>> BusID          "PCI:0:3:0"
> >>>
> >>> EndSection
> >>>
> >>>
> >>>
> >>> Section "Screen"
> >>>
> >>> Identifier"Screen0"
> >>>
> >>> Device "Device0"
> >>>
> >>> Monitor "Monitor0"
> >>>
> >>> DefaultDepth    24
> >>>
> >>> Option"UseDisplayDevice" "none"
> >>>
> >>>     SubSection     "Display"
> >>>
> >>> Virtual     1024 768
> >>>
> >>> Depth       24
> >>>
> >>>     EndSubSection
> >>>
> >>> EndSection
> >>>
> >>>
> >>>  Not sure why it’s still using mesa.
> >>>
> >>>  As to the use of turbovnc this is what I was told when I was setting
> >>> this up:
> >>> "VirtualGL needs two X servers-- a "3D X server" and a "2D X server."
> >>> The 3D X
> >>>
> >>> server has to be a real X server attached to the GPU. It can be
> >>>
> >>> headless, but it can't be virtual. The 2D X server can be a virtual X
> >>>
> >>> server, such as VNC or Xvfb.”
> >>>
> >>>
> >>> Isn’t this true? Are you saying I can just run everything through
> >>> Xvfb and still have my application run through virtualgl and use the
> >>> gpu? That would make things quite easier if that is possible.
> >>>
> >>>
> >>> Thanks!
> >>> --
> >>> Desmond
> >>>
> >>> On Friday, January 10, 2014 at 3:53 AM, Kevin Van Workum wrote:
> >>>
> >>>> It looks like your 3D X server (on :0) is not using the driver for
> >>>> your card. Make sure the "Device" section of xorg.conf has the
> >>>> correct driver and BusID specified. Secondly, I don't understand why
> >>>> you are using turbovnc. If you're trying to do captures with Xvfb,
> >>>> what's turobvnc for?
> >>>>
> >>>>
> >>>> On Thu, Jan 9, 2014 at 7:36 PM, Stealthy <ste...@gm...
> >>>> <mailto:ste...@gm...>> wrote:
> >>>>> Hi guys,
> >>>>>
> >>>>> So I am trying to run a 3d application using OPENGL through an EC2
> >>>>> server. The ultimate goal being to capture the screen using Xvfb.
> >>>>>
> >>>>> I installed virtualgl and turbovnc. when I run
> >>>>> /opt/VirtualGL/bin/glxinfo -display :0 -c
> >>>>>
> >>>>> I get this
> >>>>>
> >>>>> name of display: :0
> >>>>>
> >>>>> display: :0  screen: 0
> >>>>>
> >>>>> direct rendering: Yes
> >>>>>
> >>>>> server glx version string: 1.4 Mesa 9.2.4
> >>>>>
> >>>>> server glx extensions:
> >>>>>
> >>>>>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
> >>>>>
> >>>>>     GLX_EXT_visual_info, GLX_EXT_visual_rating,
> >>>>> GLX_MESA_copy_sub_buffer,
> >>>>>
> >>>>>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers,
> >>>>> GLX_SGIX_fbconfig,
> >>>>>
> >>>>>     GLX_SGIX_pbuffer
> >>>>>
> >>>>>
> >>>>> client glx version string: 1.4 Mesa 9.2.4
> >>>>>
> >>>>> client glx extensions:
> >>>>>
> >>>>>     GLX_ARB_get_proc_address, GLX_EXT_texture_from_pixmap,
> >>>>>
> >>>>>     GLX_EXT_visual_info, GLX_EXT_visual_rating,
> >>>>> GLX_MESA_copy_sub_buffer,
> >>>>>
> >>>>>     GLX_MESA_pixmap_colormap, GLX_MESA_release_buffers,
> >>>>> GLX_SGIX_fbconfig,
> >>>>>
> >>>>>     GLX_SGIX_pbuffer
> >>>>>
> >>>>>
> >>>>> Obviously I don’t want to use Mesa. Do I have to start the X server
> >>>>> in any specific manner to avoid using the MESA software renderer.
> >>>>> Did I miss a step.
> >>>>>
> >>>>> Thanks for your help guys!
> >>>>> --
> >>>>> Desmond
> >>>>>
> >>>>> On Thursday, November 7, 2013 at 10:03 AM, Stealthy wrote:
> >>>>>
> >>>>>> Thanks a lot for all the clarification, I’ll definitely spend more
> >>>>>> time in the manual. This is really helpful, I’ll take another stab
> >>>>>> at it.
> >>>>>>
> >>>>>> Thanks again!
> >>>>>>
> >>>>>> --
> >>>>>> Desmond
> >>>>>>
> >>>>>> On Wednesday, November 6, 2013 at 2:56 PM, DRC wrote:
> >>>>>>
> >>>>>>> Yes, you're missing quite a lot. Please read the manual. VirtualGL
> >>>>>>> needs two X servers-- a "3D X server" and a "2D X server." The 3D X
> >>>>>>> server has to be a real X server attached to the GPU. It can be
> >>>>>>> headless, but it can't be virtual. The 2D X server can be a
> >>>>>>> virtual X
> >>>>>>> server, such as VNC or Xvfb.
> >>>>>>>
> >>>>>>> Secondly, the way you launch a 3D application in VGL is by using
> >>>>>>> vglrun,
> >>>>>>> not vglclient. The instructions do not ever even specify calling
> >>>>>>> vglclient directly. vglclient is generally called from
> >>>>>>> vglconnect, but
> >>>>>>> you don't need to ever use either of those unless your 2D X
> >>>>>>> server and
> >>>>>>> 3D X server are on different machines (which they aren't.)
> >>>>>>>
> >>>>>>> I'll summarize the steps you need to take (but please do read the
> >>>>>>> manual.)
> >>>>>>> -- Set up the 3D X server, which should be a "real" X server
> >>>>>>> connected
> >>>>>>> to the GPU. Make sure the GLX extension works by running glxinfo
> >>>>>>> against the 3D X server and verifying that it is not using a
> >>>>>>> software
> >>>>>>> OpenGL renderer (Mesa.)
> >>>>>>> -- Run vglserver_config to give VirtualGL access to the 3D X server
> >>>>>>> (this is only necessary if the 3D X server is running some sort
> >>>>>>> of login
> >>>>>>> manager under the root account, like gdm or xdm. Otherwise, if
> >>>>>>> you are
> >>>>>>> just starting the 3D X server under your user account, then you
> >>>>>>> should
> >>>>>>> already have permission to access it.)
> >>>>>>> -- Start Xvfb (the 2D X server.)
> >>>>>>> -- Set DISPLAY to point to your Xvfb instance.
> >>>>>>> -- vglrun {application}
> >>>>>>>
> >>>>>>>
> >>>>>>> On 11/6/13 4:21 PM, Stealthy wrote:
> >>>>>>>> I am not sure I going through the right steps to enable
> >>>>>>>> virtualgl. Post
> >>>>>>>> installation I started the Xvfb server with this command:
> >>>>>>>>
> >>>>>>>> *Xvfb*:0.0 -screen 0 1280x720x24 -ac +extension GLX +render
> >>>>>>>> -noreset &
> >>>>>>>>
> >>>>>>>> I then started what I think is virtualgl with this command
> >>>>>>>> expecting any
> >>>>>>>> opengl application running after that would use the GPU:
> >>>>>>>>
> >>>>>>>> /opt/VirtualGL/bin/vglclient
> >>>>>>>>
> >>>>>>>> The application still only uses the CPU.
> >>>>>>>>
> >>>>>>>> Am I missing something?
> >>>>>>>>
> >>>>>>>> Thanks for the help btw.
> >>>>>>>> --
> >>>>>>>> Desmond
> >>>>>>>>
> >>>>>>>> On Wednesday, November 6, 2013 at 12:47 PM, DRC wrote:
> >>>>>>>>
> >>>>>>>>> "Trouble initializing VirtualGL with Xvfb" tells me nothing.
> >>>>>>>>> Please be
> >>>>>>>>> specific as to the error messages you are encountering or what
> >>>>>>>>> is not
> >>>>>>>>> working as you expect.
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>> On 11/6/13 2:16 PM, Stealthy wrote:
> >>>>>>>>>>
> >>>>>>>>>>> Hi guys,
> >>>>>>>>>>>
> >>>>>>>>>>> Hope I can get some help on this one. I have an ec2 instance
> >>>>>>>>>>> running
> >>>>>>>>>>> using a GPU cluster. I am capturing the Xvfb virtual screen
> using
> >>>>>>>>>>> ffmpeg and would like to have my OPENGL enabled application
> >>>>>>>>>>> use the
> >>>>>>>>>>> GPU to render. It’s currently rendering everything using the
> CPU.
> >>>>>>>>>>>
> >>>>>>>>>>> I already installed virtualgl but I’m having some trouble
> >>>>>>>>>>> intializing
> >>>>>>>>>>> virtualgl with Xvfb and ensure the GPU is being used. Anybody
> >>>>>>>>>>> could
> >>>>>>>>>>> help on this?
> >>>>>>>>>>>
> >>>>>>>>>>> Thanks in advance
> >>>>>>>>>
> >>>>>>>>>
> ------------------------------------------------------------------------------
> >>>>>>>>> November Webinars for C, C++, Fortran Developers
> >>>>>>>>> Accelerate application performance with scalable programming
> >>>>>>>>> models.
> >>>>>>>>> Explore
> >>>>>>>>> techniques for threading, error checking, porting, and tuning.
> >>>>>>>>> Get the
> >>>>>>>>> most
> >>>>>>>>> from the latest Intel processors and coprocessors. See
> >>>>>>>>> abstracts and
> >>>>>>>>> register
> >>>>>>>>>
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> >>>>>>>>> _______________________________________________
> >>>>>>>>> VirtualGL-Users mailing list
> >>>>>>>>> Vir...@li...
> >>>>>>>>> <mailto:Vir...@li...>
> >>>>>>>>> <mailto:Vir...@li...>
> >>>>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> ------------------------------------------------------------------------------
> >>>>>>>> November Webinars for C, C++, Fortran Developers
> >>>>>>>> Accelerate application performance with scalable programming
> >>>>>>>> models. Explore
> >>>>>>>> techniques for threading, error checking, porting, and tuning.
> >>>>>>>> Get the most
> >>>>>>>> from the latest Intel processors and coprocessors. See abstracts
> >>>>>>>> and register
> >>>>>>>>
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> _______________________________________________
> >>>>>>>> VirtualGL-Users mailing list
> >>>>>>>> Vir...@li...
> >>>>>>>> <mailto:Vir...@li...>
> >>>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >>>>>>>
> >>>>>>>
> ------------------------------------------------------------------------------
> >>>>>>> November Webinars for C, C++, Fortran Developers
> >>>>>>> Accelerate application performance with scalable programming
> >>>>>>> models. Explore
> >>>>>>> techniques for threading, error checking, porting, and tuning.
> >>>>>>> Get the most
> >>>>>>> from the latest Intel processors and coprocessors. See abstracts
> >>>>>>> and register
> >>>>>>>
> http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
> >>>>>>> _______________________________________________
> >>>>>>> VirtualGL-Users mailing list
> >>>>>>> Vir...@li...
> >>>>>>> <mailto:Vir...@li...>
> >>>>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >>>>>>
> >>>>>
> >>>>>
> >>>>>
> ------------------------------------------------------------------------------
> >>>>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> >>>>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> >>>>> Critical Workloads, Development Environments & Everything In Between.
> >>>>> Get a Quote or Start a Free Trial Today.
> >>>>>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> >>>>> _______________________________________________
> >>>>> VirtualGL-Users mailing list
> >>>>> Vir...@li...
> >>>>> <mailto:Vir...@li...>
> >>>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >>>>>
> >>>>
> >>>>
> >>>>
> >>>> --
> >>>> Kevin Van Workum, PhD
> >>>> Sabalcore Computing Inc.
> >>>> "Where Data Becomes Discovery"
> >>>> http://www.sabalcore.com
> >>>> 877-492-8027 ext. 11
> >>>>
> >>>>
> ------------------------------------------------------------------------------
> >>>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> >>>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> >>>> Critical Workloads, Development Environments & Everything In Between.
> >>>> Get a Quote or Start a Free Trial Today.
> >>>>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> >>>> _______________________________________________
> >>>> VirtualGL-Users mailing list
> >>>> Vir...@li...
> >>>> <mailto:Vir...@li...>
> >>>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >>>
> >>>
> >>>
> ------------------------------------------------------------------------------
> >>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> >>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> >>> Critical Workloads, Development Environments & Everything In Between.
> >>> Get a Quote or Start a Free Trial Today.
> >>>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> >>> _______________________________________________
> >>> VirtualGL-Users mailing list
> >>> Vir...@li...
> >>> <mailto:Vir...@li...>
> >>> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >>>
> >>
> >>
> >>
> >> --
> >> Kevin Van Workum, PhD
> >> Sabalcore Computing Inc.
> >> "Where Data Becomes Discovery"
> >> http://www.sabalcore.com
> >> 877-492-8027 ext. 11
> >>
> >>
> ------------------------------------------------------------------------------
> >> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> >> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> >> Critical Workloads, Development Environments & Everything In Between.
> >> Get a Quote or Start a Free Trial Today.
> >>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> >> _______________________________________________
> >> VirtualGL-Users mailing list
> >> Vir...@li...
> >> <mailto:Vir...@li...>
> >> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >
> >
> >
> >
> ------------------------------------------------------------------------------
> > CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> > Learn Why More Businesses Are Choosing CenturyLink Cloud For
> > Critical Workloads, Development Environments & Everything In Between.
> > Get a Quote or Start a Free Trial Today.
> >
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> >
> >
> >
> > _______________________________________________
> > VirtualGL-Users mailing list
> > Vir...@li...
> > https://lists.sourceforge.net/lists/listinfo/virtualgl-users
> >
>
>
> ------------------------------------------------------------------------------
> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
> Learn Why More Businesses Are Choosing CenturyLink Cloud For
> Critical Workloads, Development Environments & Everything In Between.
> Get a Quote or Start a Free Trial Today.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
> _______________________________________________
> VirtualGL-Users mailing list
> Vir...@li...
> https://lists.sourceforge.net/lists/listinfo/virtualgl-users
>



-- 
Kevin Van Workum, PhD
Sabalcore Computing Inc.
"Where Data Becomes Discovery"
http://www.sabalcore.com
877-492-8027 ext. 11

-- 

Re: [VirtualGL-Users] Running virtualgl with Xvfb on ec2 ubuntu headless gpu instance

[DRC <dco...@us...>]

On 1/10/14 2:38 PM, Kevin Van Workum wrote:
> Can you explain why you're using Xvfb? If you're just trying to make a
> video of your 3D application with ffmpeg, can't you just get it from the
> 3D X server directly? For example:
>
> ffmpeg -f x11grab -video_size cif -framerate 25 -i :0.0 /tmp/out.mpg

I am no expert with ffmpeg, but I could envision this not working with 
3D apps on the 3D X server, because when you are using DRI to render 
OpenGL, the 3D pixels are bypassing the X server.  Thus, it may be 
necessary to use Xvfb and VGL.  It does seem inefficient, though, 
because you're reading back the pixels from the 3D X server, blitting 
them to the 2D X server, and reading them back yet again.  Depending on 
the nature of the solution, it may be better to address the problem 
using a VirtualGL image transport plugin.