* 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.
* Change: In Hurricane Commons.h, even when cdebug print nothing, it slow
down the program (three times for Kite!). Create a macro cdebug_log
which calls cdebug *only* if the debug level is active.
* 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.
* Change: In Hurricane::Observer, simplification of the template so that
we can use it directly instead of creating a derived class with an
implementation of Observer::notify(). Now the implemenation is the
same for all and just calls a notify() method on the owner object.
Create a StaticObservable for fast access of the observers. This
is mainly a fixed size table (array<>) with oberservers at known
indexes. This way we can go from the obervable to the observer
owner in one indirect access (or so I hope).
* Change: In Hurricane::CellViewer, use the new observers.
* Change: In Hurricane::ControllerWidget, use the new observers.
* Change: In Hurricane::CellWidget, set the winidow initial size to a
more sensible 500px.
* Change: In CRL::LibraryManager, use the new observers.
* Change: In CRL::LibraryManager, use the new observers.
* Change: In CRL, in display.conf & technology.conf, modification for
the ST hcmos9gp capacitors.
* 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.
* Bug: In Unicorn, in cgt.py, do not import Anabatic as it is not present
yet in the devel branch.
* Change: In Hurricane, in Commons.h, in the vector slot template, start
index at 0 and not 1 (to closer to the internal representation).
* Change: In ExtensionGo, forgotten trace system calls to update.
* Change: In Hurricane, in Flags add operator overload for "int" type
and not only "unsigned int". Otherwise the compiler complaints about
ambiguous overload when using enum values which are considered as
"int".
Simpler code for the BaseFlags::contains() method.
Added implicit conversion from BaseFlags toward bool type.
* Change: In Hurricane, in Commons, complete replacement of the previous
two trace systems (trace & ltrace) by a stream-based one.
As it is a true object it is much less fragile than the one based
on defines (but maybe a little slower).
Define a reservation table for the trace levels for all the
Coriolis & Chams components.
* Change: All tools, use the new trace system.
* Change: In Bootstrap, in coriolisEnv.py no longer sets the environment
variable STRATUS_MAPPING_NAME (guessed by init).
* Change: In CRL Core, in System CTOR sets the default value for
"stratus1.mappingName" to "not_set" so the init can known it is not
set.
In etc/coriolis2/*/stratus1.conf, do not set "stratus1.mappingName".
In coriolisInit.py, now guess the value of "stratus1.mappingName"
if it has not already been set by the user. The complete setting
process is as follow (decreasing priority):
1. Setup by the user in his ".coriolis2/setting.py".
2. A "stratus.xml" file is detected in the current technological
directory (example: etc/coriolis2/65/cmos065/setting.xml).
3. Fallback value to "etc/coriolis2/stratus2sxlib.xml".
* Change: In Hurricane, in PyHurricane.h, in all DirectSet*() macros,
suppress the third parameter giving the function name. It is now
build from stringification of other parameters.
The various macros calls are also modificated all througout
Isobar.
Added PyAny_AsInt() to convert a Python Int or Long into an int.
Maybe we should manage only long, because the type difference seems
ambiguous in Python itself.
* Change: In Hurricane, in PyHurricane.h, in DirectGetBoolAttribute()
deduce the function name to display in messages from the arguments.
Idem for DirectIsAFromCStringAttribute(), DirectGetIntAttribute(),
DirectGetUIntAttribute(), DirectGetDoubleAttribute() and
DirectGetStringAttribute().
* New: In Hurricane, in PyHurricane.h, new macro DirectGetNameAttribute()
to build accessors for method returning a Name.
* New: In Hurricane, in PyHurricane.h, new macro DirectSetIntAttribute()
to build modificators taking int arguments.
* Bug: In Hurricane, in PyHurricane.h, in macro DirectSetBoolAttribute(),
the check for a boolean PyObject was wrong, leading the macro to
systematically emit a Python exception.
* Bug: In Hurricane, in PyInstance method table, PyLibrary_create() was
registered as a NOARGS function, but it *do* have arguments, change
to VARARGS.
* Bug: In Knik, in Vertex the _flags attribute was unitialized.
This flag is, in particular, telling if the vertex is blocked.
That it was taking a random value, was leading to errors and
unreproductible behaviors. I hope this will make the global
router deterministic at last. This problem did show in the
Stratus FIFO generator.
* Bug: In Knik, in Vertex, remove the unused _boundingBox attribute.
* Bug: In Katabatic, in AutoHorizontal::_slacken() when slackening
from a METAL2 terminal no longer set the terminal still on the
terminal fixed. Instead use a the new feature of
CntUserNativeConstraints.
* Bug: In Katabatic, in AutoContact, new flag CntUsernativeconstraints
so the native constraint box is specified by the user constraints.
This new possibility is needed because, if a Contact is fixed
the native constraint box *is* bound to it's current position.
But if the related segment is not fixed, and moved, it may move
the Contact with him, loosing the original fixed position.
Now we can remember the wanted position through the user
constraints.
* New: In Katabatic, in GCellConfiguration new explicit management
for 4 globals and one METAL2 (occured for the first time in the
RAM generator).
* Bug: In CRL Core, in scmos_deep_018/technology.conf, the extension
width in PTRANS/NTRANS active layer was too big. 4.0l instead of
3.0l. This was preventing seeing gaps in the active layer by cgt.
They did show up under Graal though...
* Change: In Stratus, in dpgen_RAM, the cell height was fixed to 50l.
Now read the prech[0] to set up the cell height (useful for MOSIS
cells).
* New: In Stratus, in buildModel(), the stand alone generator caller
now support a className, a modelName and a set of parameters to
pass on to the model. This is very useful when the module name
(the file), the class name and the model name are not the same.
* In Katabatic, in GCellConfiguration, added support to 3G_2M2 (generic).
* In Katabatic, in GCellGrid, in getUSide() we were not using the correct
contant set (Constant namespace instead of Kb...) so the size returned
was always X. It doesn't show if the chip is square...
* In Kite, in Track::getBeginIndex(), if the position requested was enclosed
in same net segments, we were going one index too far in the "previous"
direction.
* In Kite, in Track::getOverlapBounds() we where going one index too far
in the "next" direction while computing the "end" index.
* Bug: In Hurricane, in PyHurricane.h, the macro GENERIC_METHOD_HEAD was
not checking at all that the underlying Hurricane object was not NULL.
This may have lead to core-dumps.
* New: In CRL Core, in RoutingLayerGauge, adds a new parameter "obstacleDw".
It represents the distance to add from the *edge* of a blockage segment
to the next *axis* of a routing track. This was previously computed as
the pitch minus the mimimum half wire width. This works if the pitch
is *equal* to the wire width plus the minimal distance (edge to edge).
But if the wire width is less, than it leads to too great a distance
around obstacles, making them bigger than they are. And potentially
blocking everything. To summarize:
before: obstacleDw = pitch - wireWidth/2 (deduced)
after: obstacleDw = explicitly setup, usually:
pitch - max(wireWidth)/2
Modify the Alliance configuration helper and the various kite.conf
file accordingly.
* New: In Kite, in BuildPowerRails, make use of "obstacleDw" to compute
the footprint of a blockage.
* In CRL Core, in BlifParser, when an input terminal of an instance is
either connected to power or ground, insert a zero_x0 or one_x0
Cell to avoid direct connection to the supply (the router is not
able to do it). The names and terminals of the intermediate cells
are hard-wired for now (to SxLib).
When merging Nets, always merge internal nets into external ones
as the other way around is not always legal.
* In Hurricane, in Cell::flattenedNets(), if the cell has already been
flattened, do not issue warnings about duplictated flattened nets.
* In Hurricane, in Net, allow Net::setName() to switch the main name
for an alias without complaining about an already used name.
* In Cumulus, in RSavePlugin.py, when "views" is supplied in keywords
(kw) arguments, override the default instead of merging with it.
This is to allow scripts to save exactly what views they want.
* Change: In Hurricane, in DataBase::CellDepths() the recursion stop
criterion must be Cell::isLeaf() and not Cell::isTerminal() as
the second one can be used to hide some levels of hierarchy,
and we want all of them in a blob.
* New: In Hurricane, in Cell, create a new Slaveds relation to keep
track of all the Cells with a slaved abutment box. This work is
incomplete as we do not manage the behavior in case of merge or
Cell destruction or slaving Cells with aready slaveds ones.
Modify Cell::setAbutmentBox() to work in both autonomous and
slaved mode.
* New: In Hurricane, in Net, add a new type of Net: BLOCKAGE this
avoid us to be dependant on the framework pattern recognition.
(change propagated to the Python support)
* New: In CRL Core, in the various drivers, recognize blockage nets
as such and set their type accordingly.
* Change: In CRL, in Toolbox::deleteEmptyNets(), preserve blockage
Nets. This was the cause of crashs in Kite::BuildPowerRails() as
we where trying to use a deleted blockage net...
* Bug: In Hurricane, in NetAlias, do not write NetAlias as a name
but as a type. They were not read back and moreover staying in
the JSON parser stack.
* In Hurricane: In Instance::slaveAbutmentBox() and Instance::uniquify()
use isUnique() instead of isUniquified(). This way we do not clone
masterCells that are unique in the design. This was the reason
sometimes "holes" were appearing in the visualiser, as the AB were
not merged. The uniqification policy may need some refinement.
* Bug: In CRL Core, in BlifParser, recognize clocks (Alliance patterns).
* Change: In Cumulus, in RSavePlugin, "kw" manage a new "views" to
specify which views must be saved. Physical by default, but sometimes
we need logical as well. If the design contains uniquified cells,
save the logical view.
In ClockTree, abort the clock tree building if the design has no
top level clock.
* Change: In Katabatic, in GCellTopology, adds 2G_5M1 configuration.
* Bug: Kite, in BuildPowerRails, if we are not in a chip the nets
composing the H-Tree must be protecteds be blockages.
* Change: In Hurricane, in JsonReader::HurricaneHandler, add RawNumber().
Don't know what kind of data it parses, so for now if it's called,
just issue a warning.
* Bug: In documentation, in UsersGuide, add dependencies on doc_HTML and
doc_LaTeX so they are rebuild if the source files changes.
Some modifications to work with our new installation of TeXLive 2014.
* Bug: In Hurricane, in HApplication, ExceptionWidget ans PyHurricane.h,
forgot to catch the Hurricane::Bug exception which was leading to
terminate() without explanations.
* Change: In Katabatic, In GCellTopology, now support one GCell with
five metal1 terminal of the same net. This unlikely configuration
did appear in the SNX for the first time...
* In Katabatic, in AutoSegment::canMoveUp(), adds the ability to reject
the move up if the density of the depth we comes from is below 0.6
in all the GCells, activated with the KbChecklowDensity flag.
* In Katabatic, In LayerAssign, new method moveUpNetTrunk2() which,
unlike the previous one, move up only the part of the Net trunk
which is connex to the seed segment. Use the canMoveUp with
low density checking to avoid unbalancing higher metal layers.
* In Kite, in SegmentFsm, use the canMoveUp() with low density checking.
* Bug: In Hurricane, in Cell_HyperNetRootNetOccurrences::Locator, must
also check that the current Net is not a DeepNet in the constructor.
Take the occasion to prune Net that are automatic.
* Change: In Hurricane, in DeepNet::_createRoutingPads(), check for
already created RoutingPads. That is, the DeepNet must not have
RoutingPad components or Segments components.
* Change: In CRL Core, the Alliance VHDL (vst) driver was renaming the
names of Cells, Instances and Nets into their VHDL conterparts.
But if we still work on the Cell after saving it, the Net renaming
will cause touble, especially when there are DeepNets. The name
of the DeepNet is generated from the Occurrence name with the dot
separator which is *not* a VHDL valid character for name, thus
after that the DeepNet name has changed it cannot be reassociated
with the Occurrence path. This was causing double-flattening issues.
EricLaoGitHub
Jean-Paul Chaput
Gabriel Gouvine