I’m a little confused about what is required to do parallel data imports/rendering with ParaView. I’m running on a windows machine that has 36 cores (72 logical processors), an NVIDIA Quadro P4000 GPU, and 575GB of memory. I’m running ParaView 5.4.1-Qt5-OpenGL2-Windows-64bit install of the software. I know under Edit–>Settings–>General (Multicore Support), I can Enable Auto MPI and set it to 36 cores (or 72 processors?). I’ve also been able to launch a pvserver session and configure it so that it connects.
My main questions is if the version of ParaView that I run does not contain “MPI” in the build name, am I actually doing anything in parallel even if I’ve set up the pvserver and Enabled Auto MPI in the settings menu? When I’m testing the system on a “large” mesh job (12GB), it seems to run very slowly (takes a long time to plot a 2D surface and frequently freezes for long periods while loading). I see ParaView utilizing 230GB of memory, but it doesn’t seem to be utilizing much of the CPUs/GPUs when I check in the task manager. Since I’m new to using ParaView in parallel, I’m not sure if I’ve set things up correctly and how to manage expectations about what performance improvements I should be seeing. Any advice you can give me would be greatly appreciated.