Hello everyone, I had a mismatch error at lvs stag...
# sky130
b
Hello everyone, I had a mismatch error at lvs stage, when running user_project_wrapper. When I looked at lvs log file, I saw these power pins mismatch.
Copy code
vccd1_uq0                                  |vccd1 **Mismatch**        
vssd1_uq0                                  |vssd1 **Mismatch** 
vccd1                                      |(no matching pin)                          
vssd1                                      |(no matching pin)                          
vdda2_uq0                                  |(no matching pin)                          
vdda1_uq0                                  |(no matching pin)                          
vccd2_uq0                                  |(no matching pin)                          
vssd2_uq0                                  |(no matching pin)                          
vssa2_uq0                                  |(no matching pin)                          
vssa1_uq0                                  |(no matching pin)
The *_uq0 pins are created by the flow, not by me. How can I solve this mismatch? Reproducibles and log files can be downloaded here. I also opened a github issue in OpenLane repository, but it did not much help. For more explanation, you can see the github issue here.
m
Your result is matching circuits with mismatching pins, correct? Does this cause a problem with mpw precheck? The current flow often yields the similar results at this level. You could try
set ::env(LVS_CONNECT_BY_LABEL) 1
in the config.tcl file or
export LVS_CONNECT_BY_LABEL=1
in the shell before running lvs. You could also modify the extraction commands to use
extract unique notopports
instead of the current
extract unique
. See
openlane/scripts/tcl_commands/magic.tcl
d
@Burak Yakup Çakar I used to see this issue in my repo. Look like during PDN there are Isolated Power Mesh .. i.e some of vertical power net not connected horizontal net between hard-ip. These isolated power net are named with *_uq0,*_uq1 during spice extraction .. One way to resolve this issue is to give more space between hard-ip
b
Thank you both for your support. It helped a lot 🙂