This article applies to iGrafx Process and iGrafx Process for Six Sigma Client tools
iGrafx simulation is generally very fast considering the amount of calculations and real-world detail it must often keep track of. However, there are various factors that can affect the performance of iGrafx simulation. If you want your simulation to run more quickly, consider the following factors:
- Hardware: Get a machine with a fast processor and 4GB (or more) RAM (within the limits of your operating system or 32-bit applications; see KB article "Are there limits built into the Simulator?"). While iGrafx may have a minimum system configuration required to make it run properly, that is not the recommend configuration. We recommend as fast a machine as your budget will allow, within the limits of a 32-bit program, particularly if performance is of key concern and you will be running many simulations.
- Modeling Resources: Resources and schedules are 'expensive' for simulation; the more resources and schedules the simulator has to track, the longer simulation will take.
- Sheet Windows Open: If you are using functions to write array or scalar values to a Sheet component, ensure the sheet is not open. The screen refresh of the open sheet as values are written will slow down simulation; even if the sheet is not visible in your current window size, it is still refreshing values.
- Intelligent Shapes (iShapes): Turn off any and all "iShapes" (Intelligent Modeling/Analysis shapes that update during simulation). They slow performance down.
If, for some reason, the above suggestions do not help increase performance significantly enough for your purposes, then consider alternate modeling approaches that give the simulator less work to do:
- Change what objects (how much work) each transaction represents. One method to do this is to have each transaction represent 10 (or 100 or 1000) items of work, and scale the time it takes to process that amount of work accordingly; e.g. instead of 1 transaction taking 5 seconds, 1 transaction takes 50 seconds and represents a collection of 10 work items processed in serial (interruptions or breaks can still be modeled with resource schedules and other methods).
- Run simulation for a shorter time and extrapolate the answer over a longer period of time. In other words, consider simulation a statistical sample, and change the sample size that you can derive performance from.