RE: [sv-ec] Testbench examples ...#1step

From: Steven Sharp <sharp_at_.....>
Date: Mon Nov 06 2006 - 15:38:16 PST
>I agree that using #1step in anything other than a clocking block sample
>is of dubious value, but that's a methodology issue. We used a special
>timeunit to specify this special skew in order to allow the skew to be
>passed in as a parameter.

Another idea that does not work, if the module you are passing the value
into has a different timeunit.  The numeric value of 1step is dependent
on the timeunit.  If the global timeprecision is 1ps, then the value of
1step is 1.0 in a module with a timeunit of 1ps, and 0.001 in a module
with a timeunit of 1ns.  If you pass one of those values into the other
module and use it as a delay there, you will not get the effect of #1step.
If you pass 1.0 into the module with a timeunit of 1ns, you will get a
delay of 1ns, not 1ps.  If you pass 0.001 into the module with a timeunit
of 1ps, it will try to delay 0.001ps, which it will round to zero.

Steven Sharp
sharp@cadence.com
Received on Mon Nov 6 15:38:29 2006

This archive was generated by hypermail 2.1.8 : Mon Nov 06 2006 - 15:38:51 PST