Hello all, while tying to do a parasitic resistanc...
# magic
v
Hello all, while tying to do a parasitic resistance extraction using extresist after running ext2sim command, my magic window closes abruptly and reports segmentation fault. The .sim and .node files are being created in the work folder. When I try to reopen magic and run the same command, it throws out an error " Couldn't find def corresponding to box" How to sort this out?
t
For the segmentation fault, you might try updating to the most recent version of magic, because I discovered an issue in extresist that will cause a segfault and fixed it about a week ago. But the error only happens when doing parasitic resistance extraction on a hierarchical layout, that that isn't supported yet (or at least won't yield meaningful results). Make sure you flatten the circuit before doing "extresist". For the "Couldn't find def" issue, make sure you select the circuit before doing "extresist" (as a command, "select top cell" after "load"). The same thing applies when flattening; if you don't load the flattened cell and select, then "extresist" will end up running on the original hierarchical cell.
v
@Tim Edwards Thank you, I have few questions. For updating to newer version of magic, should i reinstall by uninstalling the previous version that i use? or can i update using a command in the terminal window? For circuit flattening , is it sufficient that i use the ext2sim command?
I mean , is it sufficient to run extresist after doing an ext2sim?
t
If you compiled and installed from source, especially if you got magic from the git site, then just do a git pull and recompile/reinstall.
The set of commands would be something like this:
Copy code
load mycell
flatten mycell_flat
load mycell_flat
select top cell
extract do local
extract all
ext2sim labels on
ext2sim
extresist tolerance 10
extresist
ext2spice lvs
ext2spice cthresh 0
ext2spice extresist on
ext2spice
v
Thanks a lot, will these set of commands work even if i don't update to a newer version?
t
Depends. What version are you using?
v
8.3.210
t
That should work. I've made a few updates since then, but mostly just refining algorithms to remove some bottlenecks (but bottlenecks definitely still exist---don't try to run RCX on a microprocessor).
v
okay thanks, i just tried the above commands, the extresist still didnt work out. I just wanted to confirm the command that you have mentioned in the fourth line. Is that select top mycell or select mycell?
t
Just literally "select top cell", verbatim.
v
oh okay!thanks
It is working! Thanks a lot
t
Yeah, it's a goal of mine to simplify the RCX extraction. I am at the point now where I can pull out the dependence on .sim files, and once that's done, I should be able to make it work properly on hierarchy, which will further simplify it.
a
@Tim Edwards I have the 8.3.213 version, but Magic keeps on crashing when I try to do RCX extraction.
t
@Anchit Proch: Are you flattening before extracting? There are always opportunities for fatal bugs in newly revised code like this, so if you suspect that it's a bug, please send a reproducible example.
v
@Tim Edwards I am using Magic 8.3.221. I too get Segmentation Fault when I try to do "extresist". I have attached my layout.
t
@vasanthramuk: Thanks, I can reproduce the segfault, and I am investigating.
👍 1
I just pushed a fix to magic (now version 8.3.222) on opencircuitdesign.com; there was a fixed array of layers per contact that was being exceeded. It needs a "proper" fix but raising the array size from 3 to 4 covers the contacts defined in the sky130A techfile, so it suffices for now. The github repository mirror will update overnight.
v
Thanks Tim