Replies: 4 comments 4 replies
-
Seen from this angle, it is effectively a heatmap. This is with 10000 WNS paths. This is corroborating that global placement merits further study. Tinkering a bit more and checking that I'm getting roughly the right coordinate, I can see that the WNS is in the same location in the plot and in OpenROAD GUI Bumped to 1000000 paths: |
Beta Was this translation helpful? Give feedback.
-
I think you want the sta_to_db_* functions from dbSta.i |
Beta Was this translation helpful? Give feedback.
-
Those plots are labeled DELAY and Skew which shows some inconsistency. I think they are measuring delay from the driver to a sequential not really a skew (I am guessing). If you see two points at different heights there is no way to know whether a path exists. |
Beta Was this translation helpful? Give feedback.
-
Seems to work now, less slack:
I get:
|
Beta Was this translation helpful? Give feedback.
-
I would like to create a 3d graph in python for megaboom to see the skew as a function of x and y on the design as I was inspired by slide 20 in https://www.eng.biu.ac.il/temanad/files/2017/02/Lecture-8-CTS.pdf
(Of course this is just a heatmap, the 3d bit is just bling, but helps create a sense of what is going on)
Here is a first attempt, using slack instead of skew, I don't know how to get from a timing path to skew, although the OpenROAD GUI shows skew per path. Though from the slack and x,y I do get a sense of what is going on.
However, I haven't found a reasonable way to map a skew to an x,y location. The transition from find_timing_paths to the ODB objects is particularly awkward/hacky below...
Any pointers on how to perfom going from find_timing_slack to skew & x,y position in tcl?
Beta Was this translation helpful? Give feedback.
All reactions