On 20/11/19 5:04 pm, DRC wrote:
On Nov 19, 2019, at 8:36 PM, dave-ms
<[email protected]
<mailto:[email protected]>> wrote:
Through a bit of trial and error I think I've got it working in both
Direct and Raw modes.
OK, so the problem was not related to CUDA?
It would appear so, although I think I'm not at the bottom of it yet,
just didn't want you wasting time on a red herring.
One difference is that last week I was working remotely, so my
workstation (with the GPU) was sitting at the login window in the
office. This week I am in the office with a laptop (client) next to the
workstation (GPU) so the workstation is logged in to a desktop. I
presume this means the displays change number (0 -> 1) as well as
probably changing some authentication issues (since I'm targetting the
desktop display - vglrun -d :1).
Using vglclient:
|
|client>vglclient &client>ssh -X
<server>server>exportVGL_CLIENT=<client_ip>:0.0server>vglrun -d
:1/opt/VirtualGL/bin/glxspheres64 |
|
This is on Ubuntu 18.04 which seems to create two displays when
logged in on the server (one for gdm and one for the desktop).
It does that because Wayland is enabled. Recent versions of
vglserver_config disable Wayland logins in GDM for that reason.
Otherwise, if GDM is running in Wayland mode, the GDM scripts that
vglserver_config modifies in order to grant users access to the 3D X
server are never executed. Regardless, logging in locally when using
the 3D X server for VirtualGL is not supported. When you log out, the
X server will reset, which would have the effect of aborting every 3D
application currently running with VirtualGL. All the more reason why
I’m trying desperately to develop an EGL back end, but unfortunately
it has turned into more of a 300-hour project than the 50-hour project
I originally envisioned, and there will be some compatibility
limitations to that back end once it is finished.
So best (only?) practice is to have a workstation booted to the login
screen and use that display (:0) for vglrun.
Using vncserver:
|
server>/opt/TurboVNC/bin/vncserver
Desktop'TurboVNC: server:2 (dave)'started on display server:2
client>|/opt/TurboVNC/bin/vncviewer server:2 &
client> ssh server
server> DISPLAY=:2 vglrun -d :1 /opt/VirtualGL/bin/glxspheres64
Polygons in scene: 62464
Visual ID of window: 0xad
Context is Direct
OpenGL Renderer: GeForce RTX 2080
|
|
App is rendering in the vnc window on the client so I think I'm good.
Just to check,is it correct to use "ssh server" (no "-X") to hop to
the server to launch applications (in this case)?
Yes. -X is never needed with the official VirtualGL launch procedures.
If you’re using the VGL Transport (which hasn’t been called “Direct
Mode” for many years), then vglconnect will automatically enable X11
forwarding in the SSH connection (which is what -X does.) If you’re
using the X11 Transport with an X proxy, then X11 forwarding is not
needed.
Thanks. I got the "Direct Mode" terminology from here:
https://virtualgl.org/vgldoc/2_0/#hd007. I now see that there's a link
to the up-to-date docs on the virtualgl.org page. If I can make a
suggestion, it wasn't clear that the Documentation header in the
left-hand sidebar was a link, I thought it was just a header (like About
VirtualGL) with only two documents in that section. Perhaps make it a
plain header and have a link below. I ended up in the old docs via
google search, having looked for the official docs on the main page and
failing to find them.
So vncserver and vncviewer use VGL Transport, and vglconnect is used
implicitly (I'm not explicitly running that anywhere). Is that correct?
And, to get my terminology and understanding right, vncserver is an X
proxy, correct?
Thanks,
Dave.
On Friday, November 15, 2019 at 3:47:01 PM UTC+11, DRC wrote:
I’ll try to repro it. It may require a similar fix to the one I
made in 2.6.3 to support OpenGL/OpenCL interoperability.
On Nov 14, 2019, at 7:54 PM, dave <[email protected]> wrote:
I'm trying to run one of the NVidia CUDA sample programs on an
application server and view it in vncviewer.
I start the vnc servine in one ssh session|:|
|
|
$ /opt/TurboVNC/bin/vncserver
|
|
and in another run the application but get the following error:
|
$ vglrun -d :1./particles
CUDA ParticlesSimulationStarting...
grid:64x 64x 64=262144cells
particles:16384
GPU Device0:"GeForce RTX 2080 Ti"withcompute capability 7.5
CUDA error at
particleSystem_cuda.cu:79code=30(cudaErrorUnknown)"cudaGraphicsGLRegisterBuffer(cuda_vbo_resource,
vbo, cudaGraphicsMapFlagsNone)"
|
Any tips on how to progress?
Thanks, Dave
--
You received this message because you are subscribed to the
Google Groups "VirtualGL User Discussion/Support" group.
To unsubscribe from this group and stop receiving emails from
it, send an email to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/virtualgl-users/09ad303d-2dd5-4143-a7e7-c67072afb9b0%40googlegroups.com
<https://groups.google.com/d/msgid/virtualgl-users/09ad303d-2dd5-4143-a7e7-c67072afb9b0%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google
Groups "VirtualGL User Discussion/Support" group.
To unsubscribe from this group and stop receiving emails from it,
send an email to [email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/virtualgl-users/cd08376b-22e1-4ae6-8a13-672517ed859a%40googlegroups.com
<https://groups.google.com/d/msgid/virtualgl-users/cd08376b-22e1-4ae6-8a13-672517ed859a%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google
Groups "VirtualGL User Discussion/Support" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to [email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/virtualgl-users/339C09E1-21DB-427C-939B-51B4455BA221%40virtualgl.org
<https://groups.google.com/d/msgid/virtualgl-users/339C09E1-21DB-427C-939B-51B4455BA221%40virtualgl.org?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google Groups "VirtualGL
User Discussion/Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/virtualgl-users/ada8483e-4d2f-3464-8241-7316625cf412%40missionsystems.com.au.
