CRPR calculation for Half-Cycle path

half cycle paths

Hi,


Will PrimeTime or any timing sign-off tool calculate CRPR differently for Half-cycle as compared as Full-cycle path.


Regards,

crpr half cycle

yes, differently!

calculation of half path in one complete path

Please will you tell me how different for Half-Cycle path.

Thanks in advance.

Regards.

prime time crpr

delays are naturally different for rising vs. falling arcs propagating through a cell. The tool will not remove differences in the common part of the clock path for half cycle paths because they are expected to be different

crpr prime time

Hi shelby,


I did not understand the this below statement . Please will you explain me again.

**The tool will not remove differences in the common part of the clock path for half cycle paths because they are expected to be different.

Regards.

crpr primetime

The delay through a cell is differect for a rising vs a falling arc. If both a rising and falling clock edge is propogated through the common part of the clock tree (for a half cycle timing check), the tool cannot remove the difference as it would in a full-cycle check.

crpr clock

Hi Shelby,

Thanks.

I have copied snip from the PrimeTime user guide regarding calculation of CRPR for half-cycle path.

"To specify whether to perform clock reconvergence pessimism removal on clock paths that have opposite-sense transitions in a shared path segment, set the following variable:

set timing_clock_reconvergence_pessimism normal

You can set this variable to normal (the default setting) or same_transition .
For the default setting, PrimeTime still performs clock reconvergence pessimism removal with opposite sense transitions in the shared path segment.
In that case, if the two transition types produce different correction values, the smaller of the two values is used."


One more help I needed from you. Please will you upload celtic noise report file if you have?


Regards,

Sorry but i got a little confused, so does it means that PrimeTime:
1. by default (timing_clock_reconvergence_pessimism = normal) does NOT care about differences between rise & fall transition in determining common-clock path (treat full- & half-cycle paths in same way);
2. BUT we can configure (by setting timing_clock_reconvergence_pessimism = same_transition) it to differentiate it (treat full- & half-cycle differently)?