Hi folks,
I have some models I'm working with that I built in 3.5 that is having a pretty serious performance drop when I move it forward to 4.0. Right now I'm seeing an increase in run time from about 40 minutes (3.5) to over three hours (4.0). There seems to be a hangup after log files are written and the autostepper increases of about 20 seconds per step which appears to be the problem but I'm not sure what's going on - there's no memory, disk, or CPU usage spikes in that time that I've been able to observe. I've been having this issue with pardiso and the MKL DSS solvers.
I am getting some warnings that the fixed/prescribed boundary conditions are deprecated and I can update those, but the warning also throws that the parameters are getting mapped properly anyway so it's unclear to me if that's the issue. Any ideas on what I can do to get the runtimes back down? I'd like to try out the updated reactive viscoelasticity models in 4.0 but the current runtime isn't sustainable for my inverse process.
I have some models I'm working with that I built in 3.5 that is having a pretty serious performance drop when I move it forward to 4.0. Right now I'm seeing an increase in run time from about 40 minutes (3.5) to over three hours (4.0). There seems to be a hangup after log files are written and the autostepper increases of about 20 seconds per step which appears to be the problem but I'm not sure what's going on - there's no memory, disk, or CPU usage spikes in that time that I've been able to observe. I've been having this issue with pardiso and the MKL DSS solvers.
I am getting some warnings that the fixed/prescribed boundary conditions are deprecated and I can update those, but the warning also throws that the parameters are getting mapped properly anyway so it's unclear to me if that's the issue. Any ideas on what I can do to get the runtimes back down? I'd like to try out the updated reactive viscoelasticity models in 4.0 but the current runtime isn't sustainable for my inverse process.
Comment