When I run PAR and look at the timing statistics in the PAR report (.par), there are a large number of long pin delays. However, when I look in the delay report (.dly), there are only a short list of delays that do not seem to match up.
What does the large list of pin delays indicate? Why is there a difference between the pin delays in the ".par" report and the net delays in the ".dly" report?
The timing result overviews in these files differ because each file contains timing information related to either pin delay (from the PAR report) or net delay (from the delay report).
The ".par" results file is connection-based or pin-based, as is stated in the header for that portion of the report. This report includes each driver-to-load connection. If a net has a fanout of 10 (10 loads for one driver) that are in a given range of delay, the PAR report counts all 10 in its delay statistics.
By contrast, for the same net, the ".dly" file indicates that only one net has a delay in the same range. The reported value for this net is the longest delay path for the one net; this is the worst path of the 10 possible paths for this net.