General Description: Cases have been seen where PAR crashes during the 2nd pass of an MPPR run during placer phase 2.2. The source of this failure has been identified as a software bug, and a fix has been scheduled.
This problem will be fixed in 6.1i Service Pack 3 which is currently scheduled for December 11, 2003.
Meanwhile a work-around is to create a command line script that runs multiple single pass PAR jobs using different cost tables, for example, "-t 5" will run cost table 5.
The phase involved has to do with calculating the placement of clock domains. If the clock domains are already area constrained, or if the number and placement of BUFGMUXs involved is such that there is no chance of a primary/secondary clock conflict, then it is possible to bypass the failing placer phase by setting the environment variable XIL_PAR_SKIPAUTOCLOCKPLACEMENT.