Design Article
Tell us What You Think
We want to know what you thought about this Design. Let us know by adding a comment.
Removing pessimism and optimism in timing analysis
Naman Gupta - Freescale Semiconductor
10/1/2012 10:29 AM EDT
Clock reconvergence pessimism
Clock reconvergence pessimism (CPR)
Clock reconvergence pessimism arises when the launching and capturing clocks re-converge at a point. Consider the following test case:
Here, the clock first bifurcates into two combinatorial logic blocks and then re-converges through a multiplexer. Timing analysis for this case would depend on the clock architecture that the design is intended to support. For setup analysis, the most pessimistic check would be if launch is through the longest clock path and capture is through the shortest clock path. Timing engineers need to make sure if such paths are supported by the architecture. If they are supported, clock reconvergence pessimism should not be removed. However, if the design architecture forbids it, one must remove the clock reconvergence pessimism in order not to make the timing analysis pessimistic.
For hold analysis, the situation is different. Hold is generally checked on the same clock edge. Thus it is unlikely that in case of hold timing launch and capture would happen through different clock paths. So, clock reconvergence pessimism should be removed for hold timing.
This discussion about CPPR applies to the cases where the launch and capture flip-flops are working at the same edge of the clock. If the transition sense is different, there might be some pessimism in slack calculations.
Next: Calculating pessimism
Clock reconvergence pessimism (CPR)
Clock reconvergence pessimism arises when the launching and capturing clocks re-converge at a point. Consider the following test case:
Figure 3: Clock reconvergence pessimism
Here, the clock first bifurcates into two combinatorial logic blocks and then re-converges through a multiplexer. Timing analysis for this case would depend on the clock architecture that the design is intended to support. For setup analysis, the most pessimistic check would be if launch is through the longest clock path and capture is through the shortest clock path. Timing engineers need to make sure if such paths are supported by the architecture. If they are supported, clock reconvergence pessimism should not be removed. However, if the design architecture forbids it, one must remove the clock reconvergence pessimism in order not to make the timing analysis pessimistic.
For hold analysis, the situation is different. Hold is generally checked on the same clock edge. Thus it is unlikely that in case of hold timing launch and capture would happen through different clock paths. So, clock reconvergence pessimism should be removed for hold timing.
This discussion about CPPR applies to the cases where the launch and capture flip-flops are working at the same edge of the clock. If the transition sense is different, there might be some pessimism in slack calculations.
Next: Calculating pessimism
|
|
|
|
|
Navigate to related information