Anyone happen to know if there are any problems with sdf simulation for Spartan-3? This works a lot of the time for me, but I get occasional setup failures, with messages like
# ** Warning: /X_SFF SETUP Low VIOLATION ON I WITH RESPECT TO CLK;
Generally, everything else fails soon after a setup violation.
The sdf file doesn't contain any triplets, despite various docs saying that it does. It contains single delay values, which are presumably the maximums. That means that I can't try sdftyp or sdfmax options in the simulator.
What I'm doing is:
- STA passes. I'm driving the device inputs with lots of slack over the worst-case numbers reported by trce
- I'm running netgen with default options on the post-PAR NCD
- sdf options to ModelSim are:
# vsim +no_path_edge +ntc_warn +splitsuh +pulse_int_e/0 +pulse_int_r/0
+multisource_int_delays +transport_int_delays +nowarnTFMPC +pulse_e/0 +pulse_r/0 +sdf_nocheck_celltype +sdf_verbose +transport_path_delays-quiet -sdfmax /TESTBENCH/DUT=dev_timing.sdf -t ps ...etc.
- XC3S1000-FG676-4, Webpack 8.2i, no service packs
Thanks -
Evan