Multiple Pass Layout attempts to improve layout quality by selecting from a greater number of Layout results. This is done by running individual place and route multiple times with varying placement seeds and measuring the best results with a specified criteria.
Note:
Before running Multiple Pass Layout, you need to save your design.
Multiple Pass Layout is supported in the following families: ProASIC3/E, Axcelerator, ProASICPLUS, ProASIC, SX-A, and eX.
Multiple Pass Layout saves your design file with the pass that has the best layout results. A corresponding timing report file for the best result, named <design_name>_timing.rpt is also saved to disk. If you want to preserve your existing design state, you should save your design file with a different name before proceeding. To do this, from the File menu, select Save As.
A timing report for each pass will be written out to the working directory to assist you in later analysis. The report files will be named <design_name>_timing.rpt.pass-number. Look at the <design_name>_iteration_summary.rpt for details of the saved files.
To configure your multiple pass options:
When running Layout, select Use Multiple Passes in the Layout Options dialog box.
Click Configure. The Multi-Pass Configuration dialog box appears.
Set the options and click OK.
Slowest Clock |
Select to use the slowest clock in the design in a given pass as the performance reference for the layout pass. |
Specific Clock |
Select to use a specific clock as the performance reference for all Layout passes. |
Timing Violations |
Select to use the pass that best meets the slack or timing-violations constraints. Note: You must enter your own timing constraints through the Timer or SDC. the "best" case is calculated by determining the design with the least amount of negative slack (or most amount of positive slack if all constraints are met). |
The extended_run_shell Tcl script enables you to run the multiple pass layout in batch mode from a command line. See the extended_run_shell script for more information.
See Also