ParaView OptiX

ParaView with NVIDIA OptiX

Provides GPU accelerated ray-tracing technology within ParaView; offering enhanced visual cues and high performance rendering for large scale scenes.

NOTE: The HPC visualization containers require NVIDIA Docker 2.0 to be installed on your system. For installation instructions, see NVIDIA Docker 2.0 github.

For general information on HPC visualilzation containers, configuration of the client and server container in both GLX and EGL configuration, see the NGC User's guide.

See here for a document describing the steps to pull NGC containers.

1. Single-Machine Container with GLX

Pull the docker image:

docker pull nvcr.io/nvidia-hpcvis/paraview-optix:glx-17.11.13-beta

Set up X11 forwarding variables:

XSOCK=/tmp/.X11-unix; XAUTH=/tmp/.docker.xauth;
touch /tmp/.docker.xauth;
xauth nlist :0 | sed -e 's/^..../ffff/' | xauth -f /tmp/.docker.xauth nmerge -

Run the image. In this example, host system data in the current directory $(pwd) are mounted to both /work in the container. This should be modified as desired.

docker run --rm -it --runtime=nvidia -v /tmp/.X11-unix:/tmp/.X11-unix -v /tmp/.docker.xauth:/tmp/.docker.xauth -e XAUTHORITY=/tmp/.docker.xauth -e DISPLAY=:0 -v $(pwd):/work:rw nvcr.io/nvidia-hpcvis/paraview-optix:glx-17.11.13-beta sh -c paraview

2. Server Container with EGL

Launching a ParaView server on GPU HPC resources often requires EGL support, requiring a separate build of ParaView for which we have a distinct container, for which we have a separate container. Pull the container:

docker pull nvcr.io/nvidia-hpcvis/paraview-optix:egl-17.11.13-beta

Specify the connection port and launch the container as follows (in this example, we listen on the default port 11111):

docker run --runtime=nvidia -p 11111:11111 --rm -it \
nvcr.io/nvidia-hpcvis/paraview-optix:egl-17.11.13-beta sh -c pvserver

3. Running The GLX Client And Attaching To The Server

With the server launched, it is then straightforward to use the GLX image to run a client, and connect to the server as follows. Here we assume the server is listening on port 11111, addressable at your.server.address.

docker pull nvcr.io/nvidia-hpcvis/paraview-optix:glx-17.11.13-beta

XSOCK=/tmp/.X11-unix; XAUTH=/tmp/.docker.xauth
touch /tmp/.docker.xauth
xauth nlist :0 | sed -e 's/^..../ffff/' \
| xauth -f /tmp/.docker.xauth nmerge -

docker run --rm -it --runtime=nvidia \
-v /tmp/.X11-unix:/tmp/.X11-unix -v /tmp/.docker.xauth:/tmp/.docker.xauth \
-e XAUTHORITY=/tmp/.docker.xauth -e DISPLAY=:0 \
nvcr.io/nvidia-hpcvis/paraview-optix:glx-17.11.13-beta \
sh -c paraview\ --server-url=cs://your.server.address:11111

4. Example ParaView Pipeline With NVIDIA OptiX

1. Exit the splash screen.
2. Click Sources > Wavelet on the top pull-down menu. Click Apply on the left pane.
3. Select Filter > Common > Contour from the top pull-down menu. Click Apply again.
4. Select Filter > Common > Clip from the top pull down menu. Move the clip plane to the desired position and click Apply.
5. Under the Plane Parameters sub-pane on the left pane, uncheck Show Plane to hide the clip plane.
6. Scroll down on the left pane and select Enable OptiX.
7. Optional: Enable Shadows OptiX.
8. Optional: Enable 4 samples per pixel and 4 ambient samples in OptiX.
9. Optional: Click Add Light on the left pane and modify as desired. The result should appear as follows:

5. Suggested Reading

ParaView

NVIDIA OptiX SDK.