[winswitch] How to use the Xdummy with Xpra?
antoine at nagafix.co.uk
Sat Dec 27 17:01:11 GMT 2014
On 27/12/14 23:53, Heng Zhou wrote:
> Thanks for the reply. I don't understand "if you have access to a real
> graphics card in the server system". By "server system" here do you
> mean the remote machine?
I said the "server", which may or may not be remote (which is not
> I asked the question in the original post just because the remote
> machine is not installed a physical graphics card,
Then you can skip VirtualGL, you will get unaccelerated software rendering.
> so I tried to use Xdummy (or Xvfb) to setup a virtual graphics card
> (or virtual frame buffer). So the settings of my question is that
> there is no physical graphics card on the remote machine, and I tried
> to run glxgears on it and return the animation back to me. You said
> your forwarding of glxgears works fine.
> Is it working on a remote machine without physical graphics card?
> If so, could you please share your settings and running steps?
Default settings, default steps.
> To be specific, is the remote machine running Red Hat?
Non-specific response: some are.
What does remote mean here? Client? Server?
> Is the following commands the sequence of steps you did:
> (1) xpra start :100
> (2) sudo DISPLAY=:100 Xorg -noreset +extension GLX :200
I have no idea what you are trying to accomplish with the command above.
xpra starts the display, or alternatively it can use an existing
display, trying to start the display after starting xpra makes no sense
> My remote ubuntu machine dies at this step.
Like I said: Xdummy is broken with Ubuntu.
It will not work.
> (3) Use Xpra Launcher in Windows to connect to the remote machine
> using display :200
> (4) sudo DISPLAY=:200 glxgears
> Then the Xpra window in WIndows should display the animation of
> rolling gears, right?
Assuming that you started xpra with Xdummy as its vfb server, yes.
> On Sat, Dec 27, 2014 at 11:16 AM, Antoine Martin
> <antoine at nagafix.co.uk <mailto:antoine at nagafix.co.uk>> wrote:
> On 27/12/14 22:45, Heng Zhou wrote:
> > Hi,
> > According to The webpage https://www.xpra.org/trac/wiki/Xdummy and
> > http://en.wikipedia.org/wiki/Xvfb, my understanding is that,
> using "Xdummy
> > with Xpra", we can run any program that produce graphical output
> on a
> > computer that does not install any physical graphics card and
> > screen, because the output of the program can be produced by
> Xdummy and
> > transfered to my local computer through Xpra. It is true for
> OpenGL program
> > in particular, assuming I have installed mesa in which a software
> > simulation of OpenGL rasterization is implemented. Am I right?
> If I am, how
> > can we run an OpenGL program, say, glxinfo or glmark2, on a
> remote machine
> > that has no physical graphics card (so /dev/dri/card0 and
> /dev/fb0 do not
> > exist as a result), e.g., the Amazon EC2 t2.micro instance
> running ubuntu
> > 14.04, and transfer the output (the animation) back to my local
> computer? I
> > can not grasp the idea of the example in the aforementioned
> Xdummy webpage,
> > and searching the internet (for a couple of days already) did
> not gives me
> > any useful hint. So could you please help me with this question?
> Thank you
> > very much.
> The default xpra installations using Xdummy support OpenGL as the Xorg
> command line includes "+extension GLX".
> Forwarding "glxgears" (or any other gl application) works fine out of
> the box, if you have access to a real graphics card in the server
> you can also use VirtualGL to get hardware acceleration.
> Unfortunately, Ubuntu does weird things with their Xorg server
> installation and so you may not be able to use Xdummy on Ubuntu.
> shifter-users mailing list
> shifter-users at lists.devloop.org.uk
> <mailto:shifter-users at lists.devloop.org.uk>
More information about the shifter-users