Various fixes for IHP SG13G2 node. #139
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…nsion.
* Move updatePosition() & checkPosition() from AutoVertical & AutoHorizontal
into AutoSegment, as it appears now to be generic code.
* Fix: In AutoSegment::updatePosition(), for very short segment the
duSource (resp. duTarget) can exceed the contact cap size. When
source and target contact are close (or at the exact same position)
we check that the target cap extension is not smaller than the
source cap extention (on case of different VIAs on source and
target, with target bigger than source). We must check the overload
with the VIAs extension, not the duTarget, which, in some case *may*
be bigger (for minimal area wires). Same apply to target.
* Change: In TrackSegment::revalidate(), merge confitional code (stupid).
* Change : In AutoSegment::_getString(), display the source/target
track footprint interval.
|
A preview of 3c88c6e is uploaded and can be seen here: ✨ https://lip6.github.io/coriolis/pull/139/ ✨ Changes may take a few minutes to propagate. Since this is a preview of production, content with |
* Promote means changing a segment in non-preferred direction, say a
vertical in M2 (in a HV scheme) up to M3. This was only possible
for segments not conecteds to terminals (short jogs, usually for
terminal access in M1). This was done by moveUpToPref().
Now add a full support able to manage non-pref directly conecteds
to terminal. This allows to solve some bad configurations occuring
with IHP SG13G2. The previous function is now replaced by a
promoteToPref().
* We partially reuse the _makeDogleg mechanism has it can be
understood as an "incomplete" dogleg (no extra parallel is made).
* This is a tricky change because we need to *re-create* the
TrackSegment build upon the AutoSegment, from a TrackSegmentNonPref
to TrackSegmentRegular.
* Bug: In AutoSegment::getEndAxes(), this function may be called
*before* updateOrient(), so source axis *may* be greater than
target axis. Just check and re-order them so we always get a
positive or null length...
* Bug: In AutoVertical::updateOrient(), check the segment S/T
ordering with S/T axis, not sourceU/TargetU. If by some other
bug they are inverted (minimum area segment), then we got
incoherency.
* Change: In AutoVertical::setDuSource() and the three others
variant, issue a warning when the extension are positive
(source) or negative (target), that is, the segment is
shrunk. Should never occur.
* Change: In AutoSegment::unexpandToMinLength(), with IHP SG13G2
we hit a borderline case. The minimum length to ensure minimum
area of a segment *is equal* to the pitch. So, in the special
case where the segment is lengthened to *one* pitch *after*
it was a minimal length the S/T du are *not* reset to zero.
Change the test for "<" instead of "<=" and hope it do not
occur in some other tech (pitch *smaller* than minimal length
to ensure minimum area).
* Should use a "<=" and *not* "<" when looking for the leftmost (first)
overlapping tile in an IntervalTree. This was causing, under certain
case, egde-touching wires being seen as disconnected.
Example: and2_x0/i0 on GatPoly for IHPSG13G2.
…es"). * The GDS could be contained in one single GDS file, aka "gallery", instead of a set of individual ones. In that case, we can now preload the GDS gallery inside a sub-library, then, when a foreign directive is encountered, create an instance of the GDS model. This is much easier than performing an "on the fly" merge of the GDS and LEF cell contents.
* In AntennaProtect::antennaProtect(), no longer forbid segments part
of an antenna cluster to move up. This is genarating too much
constraints (oversaturated lower layers) and is likely not be
really needed.
Needs to be more finely understood.
* In AutoSegment::setLayer(), when the segment is moved up *and*
the state goes from non-pref to pref, the relevant flag was
not updated leading to discrepencies between Anabatatic & Katana.
was just ignored. Now, take into account if local and close to
the maximum ripup limit and if the interval is not empty.
If at maximum ripup limit, ignore it again.
ring, wifh was, in some rare configuration causing the power lines
from the core *not* to be connected to the ring.
Now, put it on the *outer* side.
* In Tramontana::Sweepline::loadNextWindow(), the last left edge tile
signaling that the window is fully processed was badly computed.
It is the last "leftmost" tile which left edge is *still inside*
the window which should be used.
|
In order for Coriolis to work on IHP, the |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.