Form: Advanced SAPFire Options
Use the Analyze menu > Advanced SapFire Options command to access the Advanced SapFire Options form and select which numerical solver and analysis process should be used when running the analysis.
Solver Options:
Choose from Standard Solver, Advanced Solver, or Multi-threaded Solver. All three solvers perform numerical operations in a different order, and it is possible that numerically sensitive problems may yield slightly different results due to numerical round off after switching solvers. In extremely sensitive, nonlinear, history-dependent problems the differences may be more pronounced.
The Standard Solver is the slowest of the three options, but offers the most extensive reporting on instabilities that are encountered.
The Advanced Solver is the default solver and is considerably (one or two orders of magnitude) faster than the Standard Solver, and it also uses less disk space. Although the Advanced Solver will report instabilities in the model, it will not provide any information about their location. The Advanced Solver is based on proprietary technology that uses, in part, code derived from the TAUCS family of solvers. With the Advanced Solver option, it is possible to select a load case for which to generate mass and stiffness text output.
The Multi-threaded Solver is the fastest of the three solvers, offering the ability to utilize multiple processors & cores, but issues no warning about instabilities. Therefore, if you are concerned that you might have instabilities in your model, it is probably best to start with either the Standard or Advanced Solvers until you are confident in the analysis.
Analysis Process Options:
Choose from Auto, GUI Process or Separate Process. The Auto option is the default setting, and allows the program to determine whether the analysis should be run as part of the GUI process, or run outside the GUI as a separate process.
Note: It is typically recommended that you select the Auto option.
Parallel Load Case Options:
The Use up to All Physical Cores option is the default and allows the program to run up to the number of physical cores load cases simultaneously. The actual number of physical cores present on the system is displayed next to the option.
The Use up to a Fixed Number of Cores option enables you to explicitly specify the maximum number of load cases that can be run simultaneously. Running load cases in parallel requires more memory than running them sequentially, hence must be taken into consideration when specifying this value.
Note: It is typically recommended that only for fast, dedicated analysis machines should you enable the Use up to All Physical Cores option, as it can slow down other tasks running on the system. If you need to perform other work on the machine while the analysis is running, choose a fixed number of cores that is less than the number physical cores available. Running more load cases than the number of physical cores in parallel generally does not speed up the analysis, hence is not recommended.
Sets the maximum number of threads a single analysis instance/process can use. Choose from Program Determined, or User Specified.
The Program Determined option is the default and allows the analysis to use up to the number of physical cores simultaneous threads. The actual number of threads the analysis will use is displayed next to the option.
The User Specified option enables you to explicitly specify the number of threads the analysis can use. Each additional thread used for analysis requires additional memory, hence must be taken into consideration when specifying this value. It is typically recommended not to exceed the number of physical cores present on the system.
Note: It is typically recommended that you select the Program Determined option.
The Machine Override field displays the number of threads the analysis will use if the setting specified in the model is overridden by the environment variable SAPFIRE_NUM_THREADS, and None if the environment variable is not set. If you want all models run on your machine to behave the same, set SAPFIRE_NUM_THREADS to 0 for Program Determined, or a positive integer for User Specified.
For most users, the use of the environment variable is not needed.
Sets the maximum size of a response file in Megabytes (MBs) before a new response file is created. Choose from Program Determined, or User Specified.
The Program Determined option is the default and allows the analysis to create response files up to a size of ~100 Gigabytes (GBs). The actual file size limit is displayed next to the option.
The User Specified option enables you to explicitly specify the limit for the size of response files the analysis creates.
Note: It is typically recommended to use Program Determined unless the file system does not support files larger than 4 GBs.
The Machine Override field displays the response file size limit if the setting specified in the model is overridden by the environment variable SAPFIRE_FILESIZE_MB, and None if the environment variable is not set. If you want all models run on your machine to behave the same, set SAPFIRE_FILESIZE_MB to 0 for Program Determined, or a positive integer for User Specified.
For most users, the use of the environment variable is not needed.