Rebase analog fixes #110
Merged
Rebase analog fixes #110
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
* Returns the set of biggest rectangles inscribed inside the Rectilinear. This is *not* a partition, the rectangles overlaps. This is intended for the router to select the biggest rectangle in the direction he sees fit and making it a RoutingPad.
This is pursing me!
* Instead of choosing the kind of segments in accordance to the rounting gauge, use the form factor of the rectangle.
* In AnabaticEngine::loadGlobalRouting(), call relaxOverConstraineds() only if M1 is vertical (may write another one for M1 horizontal). * In AnabaticEngine::computeEdgeCapacities(), don't complain for offgrid M1 if it is allowed... * In AutoContactTerminal::updateGeometry(), implement dragging on horizontal M1. * In AutoSegment::getExtensionCap(), manage the case when neither source nor target is requested. Return the bottom layer cap in that instance (for M1 terminals). * In AutoVertical::updatePositions(), on non-preferred segments, the extention must be further extendend as to include perpendiculars axis of tracks that we may impact. * In AutoHorizontal::updatePositions(), idem. * In Configuration::selectRpComponent(), detect M1 offgrid components and flag them. HSmall & VSmall are now forced to be flagged as *punctual*. * In NetBuilderHV, support for M1 offgrid terminals. Systematically use doRp_Accesss() and never call directly doRp_AutoContacts(). * In NetBuilderVH, general overhaul & bug fixes. * In RoutingPad, add a flag to tell that a component has been selected. * In RoutingPlane::getTrackByPosition(), add failsafe when we are before the first track axis or after the last.
* TrackBlockage do not rely on an existing AutoSegment, or even a Segment, it is a portion of Track that we declare as blocked. * Used to manage the blockages coming for the LocalOnly planes, typically terminal planes. As it is not related to a Segment, we do not need to flatten the obstacles coming from the standard cells (would be a lot of them). * In PowerRails, also support blockages from Rectilinear, decomposed in vertical rectangles. * In Session, add a mechanism for TrackElement that have been put in their track *before* closing the session (used for TrackBlockage).
* Renaming TrackBlockage into a more flexible TrackFixedSpan than can
handle both blockage and terminal reservation.
Allow to change the associated net of a TrackFixedSegment, to
easily manage conflicts.
* Manage the segment unbreakble flag at the AutoSegment::create() flag
level (need to be added to Constant).
* In AutoSegment::makeDogleg(AutoContact*), don't check if the break
point is under the source/target GCell, in some case it could be
slightly outside (SegForOffgrid case).
* In AutoSegment::create(), get the correct wire with for metal1 segments.
* In RoutingPad, add the ability to manually select what is to be
considered as the center of the RP. See Configuration::selectRpComponent().
* In Configuration::selectRpComponent(), force the center of the
RoutingPad so it is in the GCell that have an accessible track
(due to offgrid).
* In Disjkstra::load(), use RoutingPad::getUserCenter() to locate the
relevant GCell.
* In NetBuilderVH::doRp_Access(), more complex creation of wiring for
offgrid metal1 terminals. Allow local metal2 to be one track outside
it's owning GCell (set the SegForOffgrid flag).
* In Anabatic::propagateConstraintsFromRp(), manage the case of offgrid
RoutingPad. Allow a one pitch margin on each side (widow of two pitchs
which should contains at least a grid point).
* In GlobalRoute::runGlobalRouter(), protect the RoutingPad in
Constant::LocalOnly layer gauge.
* In protectRoutingpads(), manage conflicts when a part of a track may
be reached by multiple terminals or overlap a blockage. Use a simple
strategy : in case of overlap, change into blockage.
Session must be revalidated after each terminal so the tracks are
up to date for overlap search for the next one.
* In RoutingEvent::revalidate(), expand the track axis constraint of
one pitch in case of an offgrid flagged segment (set in
NetBuilderVH::doRp_Access()).
* In Track::getBeginIndex(), finally manage the case where consecutive
segments of a same net are *not* overlapping (was assumed so before).
* In Track::expandFreeInterval(), fix wrong management of the case
BeginIsSegmentMax with a begin index of 0. This is a very corner
case causing error only at the very beginning of a track.
* New proxy method TrackElement::isForOffgrid().
* In AutoContact, add a flag to tag metal-only contact (for offgrid segments on metal1 mostly. * In DataNegociate::update(), when computing perpandicular constraints, when processing a non-pref, if S/T is metal-only, add a one pitch slack. * In PowerPlanes, when decomposing Rectilinear, do it *perpandicular* to the preferred routing direction. If it's horizontal, then slice vertically. * In protectRoutingpads(), full rework on how the track fixed span intervals are computeds. Must give a set of non-overlaping ones. So if they are changed into blockages they won't generate overlaps in tracks. * In Track, beginning of support for partial reorder (may give some speed-up in the end).
* Restrict the RoutingPad boxes to the cell abutment box. * Check that there are overlaping bounds returned by Tracks::getOverlapBounds(). * Pass the current net to Track::getFreeInterval(), so it is not considered a blockage...
* Returns the set of biggest rectangles inscribed inside the Rectilinear. This is *not* a partition, the rectangles overlaps. This is intended for the router to select the biggest rectangle in the direction he sees fit and making it a RoutingPad.
* In AnabaticEngine::loadGlobalRouting(), call relaxOverConstraineds() only if M1 is vertical (may write another one for M1 horizontal). * In AnabaticEngine::computeEdgeCapacities(), don't complain for offgrid M1 if it is allowed... * In AutoContactTerminal::updateGeometry(), implement dragging on horizontal M1. * In AutoSegment::getExtensionCap(), manage the case when neither source nor target is requested. Return the bottom layer cap in that instance (for M1 terminals). * In AutoVertical::updatePositions(), on non-preferred segments, the extention must be further extendend as to include perpendiculars axis of tracks that we may impact. * In AutoHorizontal::updatePositions(), idem. * In Configuration::selectRpComponent(), detect M1 offgrid components and flag them. HSmall & VSmall are now forced to be flagged as *punctual*. * In NetBuilderHV, support for M1 offgrid terminals. Systematically use doRp_Accesss() and never call directly doRp_AutoContacts(). * In NetBuilderVH, general overhaul & bug fixes. * In RoutingPad, add a flag to tell that a component has been selected. * In RoutingPlane::getTrackByPosition(), add failsafe when we are before the first track axis or after the last.
Coloquinte
reviewed
Jan 17, 2024
| @@ -2822,7 +2829,7 @@ namespace Anabatic { | |||
| DbU::Unit verticalWidth = Session::getDVerticalWidth(); | |||
| if (not Session::getAnabatic()->getConfiguration()->isGMetal(hurricaneSegment->getLayer())) { | |||
| size_t depth = Session::getAnabatic()->getConfiguration()->getLayerDepth( hurricaneSegment->getLayer() ); | |||
| if (depth > 2) { | |||
| if ((depth == 0) or (depth > 2)) { | |||
There was a problem hiding this comment.
It's unclear to me what's going on here. Why is depth 1 special?
There was a problem hiding this comment.
Mmm. It means that if either we are at depth 0 (M1) or above depth 3 (M3), we use the specific wire width and perpandicular pitch of those layers. Otherwise (M2 or M3) we use the default defined wire width and perpandicular. Re-reading that code make me thinks I could do it in a more clear way.
There was a problem hiding this comment.
Maybe at least commenting it, as it's difficult to understand for someone that doesn't know the codebase
There was a problem hiding this comment.
Yes. But, this is typically something I should rewrite and cleanup.
Coloquinte
reviewed
Jan 17, 2024
| VIA5T = createVia( tech, 'VIA5T', 'Metal5', 'Via5', 'MetalTop', u(0.26) ) | ||
| setEnclosures( VIA5T, Metal5 , (u(0.01), u(0.06)) ) | ||
| setEnclosures( VIA5T, MetalTop, (u(0.06), u(0.01)) ) | ||
| VIAS( useHV ).setup( tech ) |
There was a problem hiding this comment.
Maybe remove the commented code?
There was a problem hiding this comment.
Yes, I will. But let this one be commited !
Coloquinte
approved these changes
Jan 17, 2024
| sliceHeight = spares.conf.sliceHeight | ||
| centerArea.inflate( 4*sliceHeight, sliceHeight ) | ||
| trace( 540, '\tQuadTree.isUnderArea(): {} {} of {}\n'.format(rtag,centerArea,spares.conf.cell) ) | ||
| if rtag[-2:] == 'bl': |
There was a problem hiding this comment.
Seems slightly fishy to me, as XCenter is present 12x but YCenter only 4x
There was a problem hiding this comment.
It was because, usually, the buffer cells are much wider than tall, but it was an educated guess with no warranty it would work in all the cases.
We would in touble if the edge of the blocked zone is too close of the footprint area anyway. Let's see when the problem arise...
| @@ -0,0 +1,197 @@ | |||
| .. -*- Mode: rst -*- | |||
There was a problem hiding this comment.
Thanks. Will try to keep it updated. Schematic are exported from my Boox e-ink tablet. My dream would be to port XFig on it...
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.
A fixed version of the
analog-fixespull request. With a cleanly rebased history onmain.