New faster computer. quadro m2200. Xeon 1505 v6 32GB ram. Everything runs much faster in flow, but running a particle study that used to take 1 minute now takes 15 minutes. I am running the certified gpu driver with the quadro m2200.
Anyone else noticed this when running flow? Im also using the latest SP. I had SP 2 on my old laptop that ran faster with a 3000m quadro.
Its a confidential design, so I cannot upload the model.I do find it strange that it takes longer for the particle sim. I read that one of the areas that a faster gpu should affect is the post processing features, like particle studies. I did sent it into goengineer and they turned this (and another issue) into sworks. I will post any conclusions.
Hey Monty, I hear you worked with Arvind on this. I believe you said you tried running a performance benchmark which kept hanging up, so you reinstalled all of your Quadro software and updated all of the Windows 10 updates whether they seemed relevant or not. Did you also perform a reinstall of SWx? After that you said you saw more reasonable solve times. Is that still true?
On my machine while running your model, for a larger (1000 - 5000) number of particles in a fine (300k+) mesh, I also saw about 2-3 minutes for a particle study solve time. Some quick specs of my computer: SWx 2017 SP2.0, Dell Precision M4800, i7 2.4GHz, 4 core, SSD, Quadro K2100M, driver 10.18.13.6213. I have had issues with flow solve times when my security program started checking each of the new result files as they were created, perhaps the issue was related to that?
I looked over the notes we had with SWx, they said that a particle study didn't use any hardware different from running a flow project (CPU for analysis, GPU for visualization). They offered some insight into how the particle paths are calculated, which was interesting even if not really relevant here. Here's a summary: a particle study integrates an equation like Newton's law, where it solves for the velocity of the particle and takes into account viscous drag (drag coefficient from Henderson's semi-empirical formula), other forces on the particle based on the flow project, gravity, density of the fluid & particle, and velocity of the fluid (more specifically, the instantaneous velocity perturbation which it will obtain from a time-based flow project, or recreate using something like a random walk model in a steady state analysis).
please upload your model so other users can check if it's model or system specific