* New: In Katana::Configuration, added "katana.profileEventCosts" to
triggers the event's profiling.
* New: In Katana::NegociateWindow::_negociate(), save a profiling
trace of all the events and their priority, separated by metal,
for later analysis (see doChip.py in alliance-check-toolkit).
* New: In Katana::RoutingEvent::Key::Compare(), start implementing
new segment freedom degree functions.
* Change: In Hurricane::Net::_getString(), put a more complete
information about the Net instead of just only it's name.
* Bug: In Katana, reorder the various stages so that they are
executed in the exact same sequence as in "doChip.py" so now
routing in graphic mode and text mode gives exactly the same
results.
* Bug: In Katana::PyKatanaEngine, runGlobalRouter do not take any
argument.
* Change: In Hurrican::Component::_preDestroy(), the cascaded destruction
of slaves Components and subsequent ring merge and Rubber creation
was using set<> sorted on pointer values. This was making the order
of destruction of the slaves components non-deterministic. Now the
sets are sorted according to their id (see Entity).
* Change: In Hurricane::SharedProperty, the _ownerSet is now a vector<>
instead of a set<>. This will slow the search process, but hopefully
this is needed only during construction/destruction of a Property
so the slow down shouldn't be too much.
This has also an impact on Hurricane::Relation.
* New: In Katana::SegmentFsm, the object is now able to handle two events
at the same time. The master and it's symmetric. When there is no
symmetric, the corresponing data is just left blank. This makes a
bigger object, but as there is only one when running, it is not an
issue.
Candidates tracks are now an vector of array<2> (pairs), the
TrackCost::Compare() functor has to be wrapped through CompareCostArray.
The compined TrackCost of the two tracks is accumulated into the first
element.
Everything related to events gets duplicated: _event is now _event1
and _event2, and so on.
As there can be now two Manipulator actions done with SegmentFsm,
this class now completly hide the Manipulator level from the
RoutingEvent processing.
New function ::bindToTrack() to perform the track insertion.
* New: In Katana::TrackCost, add a new ::merge() function.
* New: In Katana::TrackElement and Katana::TrackSegment, add symmetric
management. Allows to know if a TrackElement has a symmetric and to
access it.
* New: In Katana::DataSymmetric, add new overload for ::getSymmetrical()
to handle DbU::Unit and intervals.
* Change: In Katana::RoutingEvent, remove all direct uses of Manipulator
objects. Now any change to the event associated segment must go through
call to Segment Fsm.
* Change: In Katana, adjust the debug level so internal informations are
put below level 156.
* New: In Hurricane::DbU, in ::getValueString(), special display when the
value is Min or Max (more helpful than a gigantic number).
- Data management concerning Analog GCells optimized
- Improvements of analog paths estimation:
- Save intervals to memorize paths
- Handle equal distance paths
* Change: In Hurricane::BaseFlags, store flags in uint64_t instead of
unsigned int because we start to need more than 32 different flags
in some tools.
* New: In ::getString() & ::getRecord() templates, add support for
std::array<>.
* Change: In CRL::ToolEngine, add support for timer (time & memory
measurements) displaced from Katabatic. This way all ToolEngine
can use this feature. The _postCreate() method display the
memory just after ToolEngine allocation.
* Change: In Etesian::EtesianEngine, make use of the ToolEngine
builtin timer (remove the local one). Forgot to call the base
class _postCreate() and _preDestroy().
* Change: In Anabatic::AnabaticEngine, make use of the ToolEngine
builtin timer (remove the local one).
* New: In Anabatic, new AutoSegments_Connecteds() collection. This
Collection allows a deterministic walkthough *all* the AutoSegments
connected either to source or target of one AutoSegment.
* New: In Anabatic::AutoContactTerminal::isEndPoint() to check if an
AutoContactTerminal is the *only one* anchored on a RoutingPad,
thus being a true "end point" and not a kind of feed-through.
* New: In Katana::KatanaEngine, added support for symmetric nets.
Created new class DataSymmetric to store symmetric information
of a net (mainly the paired AutoSegments).
Added KatanaEngine::runSymmetricRouter(), for now only build
the DataSymmetric informations. More to come...
* Change: In Katana::GraphicKatanaEngine::_runTest(), now perform
symmetric information building the non-symmetric routing.
- GlobalRouting takes into account self symmetries and symmetries in pair of nets according to an axis. Self symmetries are symmetries considering one net symmetric to an axis. During the materialize step, they are routed symmetrically according to the axis. Symmetries in pair of nets follow the same principle but with 2 nets.
In LocalGlobalRouting.cpp:
- Minor bug fixed
- Verbose commented
in NetRoutingProperty.cpp/.h:
- Some flags methods methods added
* Change: In Hurricane::Timer, measuring memory usage through the increase
of sbrk() gives erroneous results. Under Linux, now read the
/proc/[pid]/statm which gives the real size of the virtual memory
of the process. There is also code for reading /proc/[pid]/mmap
but it gives only the resident size of the process (i.e. swapped
pages are *not* accounteds).
* Change: In Katabatic, Anabatic, Kite & Katana now reports the total
memory size of the program.
- [Modified] Minor bug fixed
in Dijkstra:
- [Add]: Global routing for analog circuits is enhanced by intervals definition to estimate more precise wirelengh and wire positions. The distance calculation for analog cases has been moved to the horus tool in AnalogGlobalRoute.h.
* Change: In Anabatic::Dijkstra the _distanceCb attribute got copied
from the setDistance() argument. To be able to access the duplicated
distance (and therefore access/modify it) the function now return a
pointer to it. Done with the "target()" accessor of std::function<>.
* Change: In Katana::GlobalRouting::runGlobalRouter() use the new
Disjkstra::setDistance() to be able to call setNet() on the right
object.
* Bug: In Anabatic::singleGCell(), in some analogic design, there may be
not a single METAL1 in the GCell. Issue a clean warning instead of
going ashtray.
* Change: In CRL.etc/technology.cong & CRL.etc/display.conf, add a "metbot_r"
as a real layer when "metbot" symbolic layer is distinct from a normal
metal.
* New: In Anabatic::AutoContact, added forceOnGrid() method to allow
a AutoContactTerminal to be set (and fixed) outside it's constraint
Box. Be aware that this may generate gaps in the routing.
Stored using the UserNativeConstraints mechanism.
* New: In Anabatic::Session() new method getNearestgridpoint() to
easily compute the nearest grid point from a position.
* Change: In Anabatic::AutoSegment::_revalidate(), use layer mask to
compute the spin of the segment instead of the layers (more robust
for analogic).
* New: In ::GCellTopology::doRp_Access() new method to create terminal
contacts in _doDevice(). Issue more warnings and force the terminal
to on grid (even if it create gaps). This a temporary hack.
* New: In Anabatic::AutoSegment, introduce a the kind (associated to a
flag) "LongLocal". Analog GCells can be very wide, so at least some
carefuly choosen long local segments must be took into account as
attractors in the computation of the optimal axis.
* New: In Anabatic::AutoSegment::computeOptimal(), take LongLocal into
account as attractors.
* Change: In ::GCellTopology constructors compare the layers of the
RoutingPads using layer masks instead of Layer pointers. Allows to
find both "METALx" (symbolic) and "metalX" (real).
* Change: In ::GCellTopology::_doHChannel(), _doChannel(), _doStrut()
and _doDevice(), tag long locals as "LongLocal". This need to be
reviewed as it as bind done a bit too quickly.
* Change: In Anabatic::AutoSegment, due too a much bigger span of the
analogic GCells the _optimalMin & _optimalMax bitfields must use
16 bits instead of 8 (they where overflowed).
* New: In Katana, reorganisation of the initialization procedure to fit
both digital and analogic cases. Create an analogInit() method.
* Change: In Katana::RoutingEvent, the _tracksNb and _tracksFree bitfields
where too short for the Analog GCell size, now uses 16 bits instead of
6.
* Bug: In Katana::GraphicKatanEngine::drawGCell(), skip drawing of a
GCell if *both* width and height are under 150 pixels.
* New: In Katana::Session, add a new isOpen() method.
Restriction on vertex are now correctly done.
* Change: Constants
There is only one type of Strut.
* New: LoadGlobalRouting
Detailed Routing can be performed for analog circuits.
The router does not take into account analog constraints yet.
Warnings on "metal2" should be corrected in the future.
Types of GCell considered by the router are: Device, VChannel/HChannel and Strut.
The analog circuit MUST be organized in a slicing structure.
* Bug: In Anabatic::AutoHorizontal::getGCells(), the y coordinate used
to perform the line probing for neighboring GCells, cannot use the
segment constraints because they are not correctly set this early.
Instead we have to rely on the global router having put them in the
correct position (i.e. at the center of the GCell side).
The y position must also be epsilon-decreased if it is on the
border of the nothernmost GCell.
This apply to AutoVertical as well.
* Change: In Anabatic::GCellTopology CTOR, when counting the RoutingPads
of a GCell, compare gauge layers using their masks instead of the
Layer itself. This allow for Analogic layers to be recognizeds.
* Change: In Anabatic::Autocontact, replace getMinDepth() and
getMaxDepth() by getDepthSpan().
* New: In Anabatic::AutoSegment::canMoveUp(), add an optional check of
low up density (Flags::CheckLowUpDensity). Allows to move up a
segment if the up density is (very) low, and in this case it's more
efficient than breaking it to fit in the lower layer.
canMoveUp() is now able to perform the same work as canPivotUp()
if *not* supplied the flag Flags::IgnoreContacts.
* New: In Katana, in GlobalRouting::DigitalDistance() now take into
account the cost of a VIA (currently set to 2.5). Need to known the
Net currently routed in the DigitalDistance object itself.
* Change: In Katana::Track::Element::canPivotUp(), now also takes a flag
parameter.
* Change: In Katana::Manipulator, new flag IgnoreContacts to mirror the
one of Anabatic.
* Change: In Katana::SegmentFsm, allocate once a Manipulator object instead
of many times on the fly.
In SegmentFsm::_slackenGlobal(), in the slacken state, if the up
density is (very) low, bypass to move up instead of slackening.
This solve better the routing of the control part of the register file.
The register file having a pathological case of terminal placement:
many punctual terminals aligneds in METAL2 *and* a grid of METAL2 and
METAL3 blockages near below...
* Bug: In Katana::Session::_revalidate(), after removing the zero-length
segments, forgot to re-order the track, leading to many stranges effects
as the indexes where no longer coherent in the Track.
* Bug: In Cumulus & Stratus, now that creation of Hurricane objects must be
enclosed in UpdateSession, has to adds them into Python code.
* Bug: In Anabatic & Katana, in the global routing stage suspend the timed
bewteen ripup steps instead of resetting it.
* Bug: In Anabatic::RawGCellsUnder, *again*, the north and east borders
of the whole area *are* includeds (shut up disgraceful warning).
* New: In Anabatic::GCell, add a display mode to select between boundary
display (for analogic) and density display (for numeric).
* Bug: In KatanaEngine::runGlobalRouter(), do not check if an Edge cannot
be desatured when the vector or overloaded Edges is empty.
(one less disgraceful warning)
* New: In GraphicKatanaEngine::drawGCell(), support for drawing the GCells
in density mode. Use the fire scale and the MaxDensity mode.
Setup the GCell drawing mode in GraphicKatanaEngine::initGCell().
* Change: In GraphicKatanaEngine::drawEdge(), adjust the various thresholds
for showing the Edge and its label.
* New: In CRL Core, adds "anabatic.gcell.displayMode" to the set of Anabatic
parameters.
Adjust the showing threshold for GCells in "display.conf" so when the
zoom level is low, we still can see the density map.
* Change: In Anabatic, remove GraphicAnabaticEngine, PyGraphicAnabaticEngine,
PyAnabaticEngine (but keep PyAnabatic for constants exportation)
and GlobalRoute (moved to Katana). Drawing methods for Edge & GCell
are moved into GraphicKatanaEngine.
* Bug: In Anabatic, in AutoHorizontal & AutoVertical, native constraints
must use GCell::getConstraintsUMax().
* Bug: In Anabatic::GCell, correct terribly bad implementation of
both getConstraintUmax().
In getSide(), use getConstraintUMax().
* Bug: In Anabatic::AutoContact, setCBUMax() must be set using the GCell
getContraintUMax().
* New: In PyAnabatic, export the various Engine constants.
* Change: In Katana, in TrackElement and derived classes, add a getSegment()
method for easier access to the underlying Hurricane segment.
* New: In PyKatanaEngine, export the loadGlobalRouting method which was
missing (for use by Unicorn/cgt).
* Bug: In Dijkstra::_materialize(), in some special case a connex component
can is reached by two (or more) vertexes. But of course it will be
connected only through one, all others must be discarteds. The right
one will have it's connex id set, all the others will stay to -1.
So stop any backtrack when the first neighbor of a starting point
has it's connex id *not* set.
* Bug: In Anabatic::GCell, new functions getConstraintUMax() to return the
correct bounding box for the last north/east GCell. The GCells do not
own their north/east edges, except for the last ones. Thoses functions
must distinct from getUMax() because for them, the shrink is always
applied. getConstraintBox() uses the "getConstraintUMax()" functions.
* Bug: In Anabatic::AutoContact, initialize the constraint box from the
new GCell::getConstraintUMax(), to manage correctly the boundaries
of the nothernmost & easternmost GCells. The problem was showing for
pins on the north/east side of a circuit, the constraint
* Bug: In Anabatic::AutoContactHTee, Anabatic::AutoContactVTee and
Anabatic::AutoContactTerminal forgot to call the _preCreate() function
in the various create().
* Bug: In Matrix::getUnder(), if on the N/E boundary, decrease by one to
be enclosed in it.
- Modified: Matrix is now correctly updated for irregular floorplans.
Dijkstra:
- Modified: Source and Target contacts are now correctly chosen in _materialize.
* Bug: In Hurricane, in StaticObservable::getObserver(), if the slot
pointer is NULL, do not try to access the owner. Returns NULL, so
the caller can be aware of the situation...
* Change: In Hurricane, in BreakpointWidget & ExceptionWidget some
cosmetic changes (fonts and window sizes).
* Bug: In Anabatic, In AutoHorizontal::getConstraints(), take into account
the constraints from the source AutoContact, as it holds the constraints
transmitted by the RoutingPads and sets up by propageConstraintsFromRp().
It is likely to be a bug affecting the original Katabatic as well.
* Change: In Anabatic, in RawGCellsUnder(), check that the segment is not
completly oustside the cell abutment box and truncate the coordinates
to the part that is inside. Use the "shrink" if we reach the east/north
border.
* Change: In Anabatic, in Configuration, no more decorator because we will
use a true derived relationship. Katana *derives* from *Anabatic* and do
not *decorate* it, so the Configuration can do the same. It also implies
that we directly create a Katana engine, not an Anabatic one.
* Change: In Anabatic, in Session, do not allow the opening of the Session
in a standalone fashion (with a static method). Instead it must be opened
using the relevant method of the Anabatic/Katana engine. This ensure we
are opening the right Session type.
* Change: In Anabatic, in AutoSegment_Aligneds() collection the seed segment
is not part of the collection by default, but will be included if the
Flags::WithSelf is set.
* Change: In Configuration, all the flags value are now defined in two steps.
Declared in the header and initialized in the module. This is to prevent
the fact that on some cases, in relation with the Python "extern C" part
modules, we need a true allocated variable. It was causing weird linking
problems.
A side effect is that they can no longer be used as entry is switches,
have to replace them by if/else.
* New: In Anabatic, new GCell::getNeighborAt() utility function.
* Bug: In Anabatic, in GCell::doGrid(), tag all the GCells of the grid with
the grid type... Back annote all the edges capacity (north & east) with
the reserved local capacity.
* New: Complete portage of Kite over Anabatic. The new engine is christened
"Katana" for Kite-Analogic. When it's capabilities and performances
will be on a part with Kite, it is to completly replace it (and take
back the "Kite" name). Preliminary tests seems to show that, contrary
to intuition (because built on a more complex/slower grid), it is even
slightly faster than Kite 8-).
* Bug: In Anabatic::GCell, when width and/or heigh is null, do not try to
compute densities as it will trigger a divide by zero (and do not make
much sense anyway).
* Change: In CRL Core, in cmos/alliance.conf, honor the ALLIANCE_TOP variable
if it is found in the user's environement.
* New: In Anabatic:
- In AnabaticEngine, create a new NetData information for storing
the static ordering of nets (for the global routing). The order
is the half perimeter of the search area divided by the number
of RoutingPads. The order is static once it as been computed in
the first place (even through ripup stages).
* Change: In Anabatic:
- Long wires, going straight accross multiple GCells were splitted
into a long sequence of segment+contact, one for each gone-through
GCell. It was memory-consuming and algorithmically inefficient.
Now there is only one straight wire.
As a consequence, Edges now store the list of all segments
going through them. That way we do not need to load a Net into
Dijkstra before riping up some of it's segment. The ripup mechanims
can now be implemented right at AnabaticEngine level, without
dependencies over Dijkstra. Note that long wires are automatically
broken in two if we need to attach a new branch on it (breakAt()).
And fused back into one if a branch is removed (unify()).
* Bug: In Hurricane, in tstream (Commons.h), in the *template* overload of
operator<<(), do not use the operator<<() of ostream as it will be
looked up in "stage 1" (template definition) and so will miss all the
overloads added later and built over getString<>(). Instead, make use
of getString<>(), which, as another template will be looked up in
"stage 2" (template instanciation) and at that point will have all the
needed template specialisation of getString<>().
We also need to define new stream manipulators to be able to create
a matching template overload not dependant from the implementation.
To avoid name clashes, we prefix a 't'. For now, only 'tsetw()' is
refined.
As a side effect, we cannot directly print bit-fields into the stream,
we must go through an intermediate variable (happens once in AutoContact).
* New: In Anabatic:
- In AnabaticEngine, keep track of overflowed edges.
- In AnabaticEngine, getNetsFromedge() to lookup all nets going
through an Edge.
- In Configuration, read the Kite "reserved local" parameter to
decrease the Edge capacity (it's a guessing of the cost of the
local routing).
- In Edge, add an attribute to know if there is an associated
segment of the current net (set by Dijkstra::_traceback()).
Transparently manage the overflowed edges.
- In GCell_Edges, correct a filtering bug when not all sides are
selecteds.
- New GCell::getEdgeTo() to find the edge between two adjacent
GCells.
- New GCell::unrefContact() to automatically removes global contacts
no longer used by any global segments (used during the ripup
step).
- In Dijkstra::load(), now able to "reload" and already partially
or completly routed net (look for Contact of "gcontact" layer
and their attached segments).
- In Dijkstra, keep the last net loaded until the next one is.
Put the cleanup operations in an isolated function "_cleanup()".
- In Dijkstra::_selectFirstsource() and run(), load first source
component made of multiple vertexes.
- In Dijkstra::_trackback(), link the Net segments to the Edges.
- New Dijkstra::ripup(), Dijkstra::_propagateRipup() to perform
the ripup of one edge of a Net (must be loaded in Dijkstra first).
Dijkstra::_tagConnecteds() setup the connexId of a set of Vertexes
- that are connecteds through edges *with* segments.
- In GraphicAnabaticengine & GlobalRoute.cpp, embryo of a global
routing tool with ripup & reroute.
- (Add) HurricaneAMS includes
in Dijkstra:
- (Add+Modify) Handle restriction rules in Vertex for analog devices
in GCell:
- (Add) Check neighboring cells (North/South/East/West)
* Change: In Anabatic:
- In Dijkstra, replace _toWires() by _trackback() which is called
immediatly on reaching a target.
- In Constants & GCell, added flags for typing the GCells. We now
have:
* DeviceGCell, for analogic devices.
* ChannelGCell, for resizable routing channels.
* StrutGCell, for fixed spaces in analog designs.
* MatrixGCell, for square tile in digital designs.
* Bug: In Anabatic:
- In _propagate(), on reaching a target, forgot to remove it from
the queue before pushing it back with the new distance. It also
simplificate the core algorithm as target as treated normal nodes.
* New: In Anabatic:
- Update cdebug to use the fastest macro version.
- More readable drawings of GCells and Edges.
- Added timer support.
- The distance is now computed in DbU::Unit (aka long) and not in
normalized float.
- The distance function is now a callback (std::function<>) that
can be changed (a default is provided at initialization).
- New concept of branch in the agglomerated connex component.
Each trace back part create a "branch" (tagged with a "branchId").
When a node is reached with the same distance, but from two
different branches, choose the the branch that was lastly created.
This create a slightly different tree which grows outward from
the newest branches.
- Makes the horizontal edges *slightly* longer than the vertical ones
to skew the tree to use vertical edges, as it is usually less
congested than the horiontal one (due to metal1 cell terminals).
It is also my understanding that it is useful to reduce the
number of vias, whithout introducing a via cost.
* New: In Bootstrap:
- Script sprof.py to perform sprof & demangle libraries execution
profile.
* ToDo: In Anabatic:
- Corner optimization. Sometimes when two corners are possible, the
wrong one is choosen. That is, one of it's edge cannot be used for
further grow of the tree.
* Bug: In Anabatic:
- Activate the searchArea to limit the area explored by the Dijkstra.
Uses the bounding boxes of the vertexes/gcells instead of the ones
the RoutingPads.
- Always detach contacts from the rings in load(), even if the vertex
is already marked as target.
- On reaching a target (non-negative connexId), do not forget to update
the distance to source.
- In Edges locator, added support for side selection (needed by the
monotonic feature).
* New: In Anabatic:
- Hide the atomic Dijkstra methods (made private).
- Implementation of the monotonic mode in Dijkstra (for bipoints only).
- Added a separate function for distance computation between two vertexes,
now take account of the VIAs costs.
* Bug: In Anabatic:
- In Dijktra::load(), do not create Contact while looping over the Net's
components (for RoutingPads).
- In Dijkstra::propagate(), reset the connexId of reached Vertexes
on updating the stamp to avoid using connexId from previous runs.
Push the vertexes on the queue while backtracking (with a distance
of 0.0). Do not reset the "_from" as we need it in Dijkstra::toWires()
to know how were the routing is.
- In Edge::getDistance(), convert to float more early so the normalisation
do not always end up in zero.
- In GCell::_add(), when the axis of the new edge is equal to one already
on the given side, adds it *after* the existing edge and not before.
(to be coherent with the way GCells are split in two)
- In GCell::hcut() and GCell::vcut(), create an Edge if is equal to X/Y Max,
*but* the new chunk is flat. The new chunk is then expected to expand
"below".
- In GraphicsAnabaticEngine::drawGCell(), display the id of the GCell in
a small box at the center of it's bounding box. Nothing displayed for
flat GCells, to avoid cluttering.
* Bug: In Anabatic:
- In Dijkstra/Vertex, use the new Observer for fast access from the GCells.
- In Dijktra::propagate() loop detection.
- In Dijkstra, reset the "_from" of vertexes between two propagates on
the same Net (otherwise: loops...).
- In Matrix, there must be two different way of calculating an index from
a position. One for the min, which is rounded to superior. And one for
the max which is rounded to inferior.
- In Matrix::updateArea(), checks for invalid indexes.
* Bug: In Anabatic:
- In Edge, materialize was done before the Edge get it's final position.
Must be invalidated afterwards for the UpdateSession to put him in
the correct QuadTree.
- Functional (debugged) Dijkstra skeleton.
* Change: In Hurricane, in CellWidget, enable the drawing of ExtensionGos
according to the the threshold paramaters (as layers do).
* Change: In CRL Core, provide DrawingStyle for Anabatic::GCell &
Anabatic::Edge. Use the exact name of the ExtensionGo.
In kite.conf, add settings for the size of an edge as the optimal
size may change between analog & digital designs.
* New: In Anabatic:
- Support for fast position query using a matrix.
- Computation of the Edge capacity, based on the routing gauge parameters.
- Graphic display of the edges capacity. Added configuration
parameters to size the bounding box of an Edge.
* Bug: In Anabatic:
- GCell::_moveEdges was both too complex (some cases never arise the
way we divide GCells) and bugged.
EricLaoGitHub
Jean-Paul Chaput