Hello, any tips to reduce the time it takes LVS to...
# gf180mcu
b
Hello, any tips to reduce the time it takes LVS to run? Been looking through this readme https://github.com/efabless/mpw_precheck/blob/main/checks/be_checks/README.md and its not entirely clear to me what options can make it run faster. Thanks in advance!
m
If you have any srams, LVS is probably stuck in a loop until magic is updated to 8.3.442 or better. Are you running locally or on the platform?
b
On the platform as in?
I'm going to guess I'm running locally
m
Running locally is done with
make run-precheck
. On the platform, is done on the web.
Are you using the
gfmpw-1c
Makefile?
b
I'm running locally then. And yes, using the gfmpw-1c Makefile
m
Can you check the magic version used in the precheck docker? Look in the
precheck_results/<tag>/logs
directory for any magic logs.
b
In /precheck_results/<tag>/logs/tools.info, I'm seeing Magic: 8.3.442. Not seeing any magic logs though.
m
Can you check your precheck version with
git log
? Installed in
~/mpw_precheck
by default or
PRECHECK_ROOT
if that’s set. If it’s not
gfmpw-1c
the run
make precheck
. Is the LVS job completing but taking a long time?
b
Precheck version is gfmpw-1c. It's completed on user_proj_example, but I've implemented a small test project that it hasn't completed on.
Let it run for a few hours and hasn't completed
m
Can you tell if it’s stuck on extraction or LVS?
b
The last output is the LVS output directory
From ext.log, looks like extraction has finished
m
LVS can take a while if there’s a serious mis-match. You can look at the current report file to (hopefully) find a clue.