I'm attempting to harden my design using option 1 ...
# openlane
m
I'm attempting to harden my design using option 1 in the caravel/openlane readme from the develop branch. I started out by copying the
user_proj_example/config.tcl
and modifying it appropriately. I then commented out
set ::env(DIE_AREA)
setting of
0 0 600 600
since that is fairly small (and the design needs more room). When I do this, the core area remains very small and modifying
set ::env(PL_TARGET_DENSITY
has no effect. Of course this then results in a RePlAce error saying
Filler area is negative!! Please put higher target density or Re-floorplan to have enough coreArea
. If I uncomment
set ::env(DIE_AREA)
and adjust manually, I get past the core area issues. So my questions are, 1.) where is the default core area parameter being set when
DIE_AREA
setting is commented out in the config.tcl? 2.) For this option (option 1) of hardening our design, is setting
DIE_AREA
to specific values required?
t
Density is not the area used. Try
FP_CORE_UTIL
And, no, setting a
DIE_AREA
for the "inside" design (i.e. what's going inside the wrapper) is not required.
m
Thanks for the reply. I've commented out
FP_SIZING
and
DIE_AREA
and added
FP_CORE_UTIL
. However, the utilization % (as reported in
replace.log
) is always 1% greater than the value I set for
FP_CORE_UTIL
. Again, if I change
DIE_AREA
from
0 0 600 600
to
0 0 1250 1250
, the RePlAce step succeeds. I'm trying to understand the correlation between
FP_CORE_UTIL
and the default core area used when
DIE_AREA
is not specified.
t
My understanding is that it'll just take the estimated gate are after synthesis and try to make that area
X
percent of the total area. Of course this is just an estimate at that point and so by the time you get to placement, there will be small differences.
m
Ok, I'll continue to manually specify
DIE_AREA
as that gets me past this issue.