Explanation of solver options
On this page:
What is the difference between standard, advanced, and multi-threaded solvers?
Answer: The differences between these solver options are described as follows:
Solver Options
Standard solver
best for small problems
utilizes only single core/CPU
provides full instability information, which is useful for checking model stability before a long analysis
Advanced solver
default setting
best for medium to the largest problems
can utilize all cores/CPUs
utilizes disk to handle very large models
provides limited instability information
Multi-threaded solver
best for medium to large problems
fully utilizes all cores/CPUs
fully runs in RAM for speed
provides no instability information
Recommendations:
Use the Standard or Advanced (default) solver to check for stability early in the development of models.
Modal analysis using eigen vectors can help find instabilities.
Switch to the Multi-threaded solver for speed when the model is well developed and stable.
Use the Advanced solver instead if the model is too large and the Multi-threaded solver reports memory limitations.
Analysis Process Options
GUI process
best for small problems
analysis runs within the software, such as with SAFE.exe
benefit: less disk operations (I/O) are performed
drawback: the software itself consumes memory, leaving less available for analysis, which slows operations and prevents the ability to run larger models
Separate process
best for medium to large problems
the analysis model is written to the disk and read by CSI.SAPFire.Driver.exe, then analysis is run within CSI.SAPFire.Driver.exe
benefit: the analysis engine has access to more memory, therefore larger problems can be solved and analysis runs faster
drawback: time is lost to write and read the analysis model
Auto
default setting
the necessary memory is estimated, then compared to the physical ram available. If enough RAM is available, the analysis runs in GUI process. If not, it is shelled out to CSI.SAPFire.Driver.exe.
Parallel Load Case Options
If the Analysis Process option is set to “Separate Process”, it is possible to run up to eight load cases in parallel and significantly reduce the total analysis time for models with many long-running load cases.
Running multiple load cases in parallel puts a higher demand on system resources (e.g. memory, CPU, and disk) than running them sequentially, and can slow down other tasks running on the system, hence is best for fast, dedicated analysis machines.
Running more load cases than the number of physical cores in parallel generally does not speed up the analysis, hence is not recommended.
See Also
Parallel processing article