Troubleshoot High Elapsed Run Times
One indication of computer system performance is the ratio of elapsed time to CPU time for Creo Simulate engine jobs. A high ratio of elapsed time to CPU time can indicate a performance problem, either with the settings you are using to run Creo Simulate or with your computer system.
For jobs running the direct solver, a ratio of elapsed time to CPU time much greater than 4 can indicate a problem. For jobs running the iterative solver, a ratio much greater than 7 can indicate a problem. The iterative solver generally has a higher ratio of elapsed time to CPU time because it does more I/O per calculation than the direct solver.
Following are possible causes of high elapsed times:
• Running more than one application at a time — This might not be a true performance problem unless the applications are competing with each other for access to the CPU, memory, or disk. If performance is lower than you expect, consider what other applications might have been competing for resources.
Determine whether the problem is repeatable by rerunning the job when there is little or no other activity on the machine.
• Setting the solver RAM (solram) value
too high
• Setting the solver RAM (solram) value
too low
• Starting the iterative solver after pass 3 or later, instead of after pass 2 — Unless you are solving a class of problems that consistently demonstrate better convergence by starting the iterative solver after pass 3, always start the iterative solver immediately after pass 2.
Starting the iterative solver after pass 3 or later frequently results in greatly increased I/O and elapsed time, with little or no reduction in the number of iterations required for convergence.
• Using remote NFS-mounted disks for your working directories — I/O to network disks can be up to 10 times slower than I/O to local disks. For example, a local disk might transfer data at 10 MB/sec, whereas a remotely mounted disk might transfer data at only 1 MB/sec. Use local disks whenever possible.
• Using swap space that is not local to the machine — Add local swap space and avoid using remote swap space.
• Using swap space or disk space that is not dedicated to that purpose — Some operating systems allow an area of a disk to be shared dynamically between temporary files and swap space. Switching from this type of dynamic sharing to dedicated disk space or dedicated swap space can provide much better performance. You can also gain performance by placing the swap space and the scratch files on separate physical disks.
• Running a large job with insufficient machine RAM — The performance of many programs suffers when the virtual memory usage exceeds the available machine RAM. However, Creo Simulate typically uses only a small part of memory at a time. Therefore, performance can remain stable even when the reported total memory usage is 2 or 3 times machine RAM, and sometimes as much as 5 times machine RAM.
If you follow the guidelines for all run-setting parameters and you still suspect poor performance, compare the ratio of elapsed time to CPU time and compare the memory usage for this job with previously run jobs using similar models. Make sure all runs were made on the same machine under the same work load conditions. Similar models should have the same element and analysis types and use the same type of solver.
If the ratios of elapsed time to CPU time and memory usage to machine RAM are both high, you might need to install more machine RAM in order to improve performance. In some cases, it helps to decrease the solram allocation or reduce the size of the model you are analyzing.
For more information on the use of virtual memory, see
Managing RAM, Solram and Swap Space.