Has anyone noticed parasitic capacitance increased...
# openlane
m
Has anyone noticed parasitic capacitance increased between mpw8 and the latest tools? I'm running STA on the TinyTapeout design and traces routed on met1,2,3 between blocks has jumped from 0.03 to 0.06
image.png
tt02 layout
tt03 layout
My guess is that the values for capacitance have changed in the newer PDK, but I'm having a hard time testing it because I switched to the new JSON configuration for TT03 and it's making it much harder to try the old PDK and old openlane with the TT03 configuration
k
@Matt Venn is it possible to share both full reports ?
m
@Kareem Farid the reports are linked above
@Jeremy Birch has been helping me with this, and his thoughts are that the parasitics for TT02 were too low
k
@Matt Venn I am suspecting the following: There are differences in the wrapper level routing between TT02 and TT03, such as net
sw_245_clk_out
having differences in capacitances . If we take these variances and high fanout of scanchain blocks into account, they can propagate and lead to what we are seeing now. I will try to confirm this suspicion by tomorrow, as today is a National holiday in Egypt.
m
Thanks Kareem. I checked the nets and they looked very similar to me. Also, the fanout is low, as it's just one clock to the next. Max fanout is 1.
k
The fanout of a chain's root buffer is 10. This buffer is linked to chain and perhaps more sensitive towards subtle differences in capacitances of the link between chains. Will confirm tomorrow
m
Hi Kareem, just wondering if you made any progress
k
@Matt Venn what are the differences between tt02 and tt03 ? In terms of OpenLane version, generated files, etc ..
Can you compare the length of the same net in the two layouts for example
sw_244_clk_out
Also can you share the details of extraction off and on the flow?
What were the versions used to generate tt02 user project wrapper spef ?
m
check the files in the signoff directory for versions of pdk and openlane