* 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.
* Bug: In Hurricane, in Cell::flattenNets(), addition to the topHyperNets
vector was done *inside* the components loop, resulting in multiple
additions of the same top net. This was leading to the RoutingPads
created multiple times on the same connectors. Hence the conflict in
KiteEngine::protectRoutingPads().
* Change: In Hurricane, in Cell::flattenNets(), do not create RoutingPads
or build rings on already routed Nets. A Net is considered already
routed if it has at least one Segment. This way we avoid Rubbers to
be drawn over routed Nets.
* Change: In Hurricane, in DeepNet, do not build RoutingPads & rings on
already routed Nets (same condition as in Cell::flattenNets()).
* New: In Hurricane, in HyperNet, new collection of all component
occurrences of an HyperNet. May or may not (default) include
components from the leaf cells.
* Change: In Knik, in Vertex, add a "blocked" flag to signal disabled
vertexes in the grid (must not be used by the global router).
Modificate the Graph::getVertex() method so that when a vertex
is geometrically queried, if is a blocked one, return a non-blocked
neighbor. This mechanism is introduced to, at last, prevent the
global router to go *under* the pad in case of a commplete chip.
* New: In Katabatic, in AutoSegment, a new state has been added: "reduced".
A reduced segment is in the same layer as it's perpandiculars.
To be reduced, a segments has to be connected on source & target to
AutoContactTurn, both of the perpandiculars must be of the same layer
(below or above) and it's length must not exceed one pitch in the
perpandicular direction.
To reduce an AutoSegment, call ::reduce() and to revert the state,
call ::raise(). Two associated predicates are associated:
::canReduce() and ::mustRaise().
Note: No two adjacent segments can be reduced at the same time.
* Bug: In Katabatic, in GCellTopology, add a new method ::doRp_AccessPad()
to connect to the pads. Create wiring, fixed and non managed by
Katabatic, to connect the pad connector layer to the lowest routing
layers (depth 1 & 2). The former implementation was sometimes leading
to gaps (sheared contact) that *must not* occurs during the building
stage.
Remark: This bug did put under the light the fact that the initial
wiring must be created without gaps. Gaps are closed by making doglegs
on contacts. But this mechanism could only work when the database if
fully initialised (the cache is up to date). Otherwise various problems
arise, in the canonization process for example.
* New: In Katabatic, in AutoContactTerminal::getNativeConstraintBox(),
when anchored on a RoutingPad, now take account the potential rotation
of the Path's transformation. Here again, for the chip's pads.
* New: In Kite, support for reduced AutoSegment. TrackSegment associateds
to reduced AutoSegment are *not* inserted into track to become
effectively invisibles. When a segment becomes reduced, a TrackEvent
is generated to remove it. Conversely when it is raised a RoutingEvent
is created/rescheduled to insert it. All this is mostly managed inside
the Session::revalidate() method.
* New: In Kite, in KiteEngine::createGlobalGraph(), in case of a chip,
mark all global routing vertexes (Knik) that are under a pad, as blockeds.
* Bug: In Cumulus, in PadsCorona.Side.getAxis(), inversion between X and
Y coordinate of the chip size. Did not show until a non-square chip
was routed (i.e. our MIPS R3000).
* Change: In Stratus1, in st_placement.py add the ClockBuffer class for
backward compatibility with the MIPS32 bench. Have to review this
functionnality coming from the deprecated placeAndroute.py.
In st_instance.py, no longer creates the Plug ring of a Net.
In my opinion it just clutter the display until the P&R is called.
Can re-enable later as an option (in Unicorn).
* Change: In Unicorn, in cgt.py, more reliable way of loading then running
user supplied scripts. Borrowed from alliance-checker-toolkit doChip.py .
* Change: In Hurricane, the NetRoutingProperty is moved into Hurricane
from Katabatic. Needed for Knik to be able to access thoses
informations.
* Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in
case of identical area, select the component of lowest id.
This should not be needed if the component ordering was fully
deterministic as it should be (will investigate later).
This is to ensure that the choosen component is always the
same, especially between save/load of a global routing.
* Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate
the segments only if the topology is valid (no NULL in the
cached segments).
* Bug: In Katabatic, in GCellTopology::construct(), throw an error
if the topology is bad instead of trying to continue (and core
dump later... ).
* Bug: In Kite, in BuildPowerRails, distinguish the name of the master
net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the
name of the net in the *chip* netlist. Must use the later to make
comparison as they may differs.
* Change: In Knik, in save/load solution, exclude nets that are not
globally routed by Knik. That is which NetRoutingProperty is not
*Automatic*.
* Bug: In Cumulus, in chip.BlockPower take account of the layer
width extention to sligthy shrink the connector thus avoiding a
notch with standart cell in some cases.
* Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner
trees for the leaf clocks, as it seems overconstrained for the
router. First move was to lower them in M2/M3 (instead of M3/M4)
but that was not sufficent.
* New: In Cumulus, RSavePlugin for recursively saving a physical
hierarchy.
* New: In documentation, first embryo for RDS file. Should have been
in Alliance git, but I prefer to keep newest doc in Coriolis.
* New: In CRL Core, in helpers & alliance.conf, set and read a "PAD"
variable to define the pad model name extension ("px" for "sxlib
and "pxr" for vsxlib, this is provisional).
* New: In CRL Core, in plugin.conf, add parameters to define the name
of used for power & clock supply. We may remove the extention in
the future (to be more coherent with the previous modification).
* New: In Cumulus, in chip.Configuration.GaugeConf._rpAccess(), no
longer place the accessing contact *at the center* of the
RoutingPad. It works under sxlib because buffers & registers all
have same size terminals. But this is not true under vsxlib,
leading to misaligned contacts & wires. Now systematically place
on the slice midlle track (maybe with one pitch above or below).
This is still very weak as we do not check if the terminal
reach were the contact is being put. Has to be strenthened in
the future.
* New: In Cumulus, in chip.Configuration.ChipConf, read the new
clock & power pad parameters.
* Change: In Isobar (and all other Python wrappers), uses PyLong instead
of PyInt for DbU conversions. In PyHurricane argument converter,
automatically check for both PyLong and then PyInt.
* Change: In Cumulus, in chip.PadsCorona, more accurate error message
in case of discrepency in global net connections (i.e. no net
of the same name in instance model and instance model owner.
* Change: In Kite, in BuildPowerRails, when looking up at the pads
model name to find "pck_" or "pvddeck_", do not compare the
extension part. But we still use hard-coded stem pad names,
maybe we shouldn't.
* Bug: In Katabatic, in GCellConfiguration::_do_xG_xM1_xM3(), there
was a loop in the search of the best N/E initial RoutingPad.
* Bug: In Kite, in KiteEngine::protectRoutingPads(), *do not* protect
RoutingPads of fixed nets, they are already through the
BuildPowerRails stage (and it's causing scary overlap warning
messages).
* Bug: In Cumulus, in ClockTree.HTreeNode.addLeaf(), do not create
deep-plug when the core is flat (not sub-modules). All the new
nets are at core level.
* Bug: In Cumulus, in ChipPlugin.PlaceCore.doFloorplan(), ensure
that the core is aligned on the GCell grid (i.e. the slice
grid of the overall chip).
* Bug: In Kite, in GCellTopology::_do_xG_xM1_xM3(), infinite loop
while looking for the bigger N-E RoutingPad. Forgot to decrement
the index...
In Katabatic & Kite, remove all hard-coded values related to track pitches.
* New: In <Session>, add more convenience function to access RoutingGauge
characteristics.
* New: In <AutoSegment>, <AutoContact>, add support for the "depth spin",
that is, if the source/target contacts are going "top" or "down".
Used to compute the perpandicular pitch. Need a small modification
of the revalidation mechanism. The observers of <AutoSegment> are
notified when the spin changes.
* New: In <AutoSegment>, the getPPitch() method allow to compute the
"perpandicular pitch". For now it is simply the greatest from the
source perpandicular pitch and the target perpandicular pitch.
Make uses of the "depth spin".
* New: In <TrackElement>, <TrackSegment>, cache the perpandicular pitch.
Updated through the notification from the observable.
- Change: In RoutingEvent::State::conflictSolve1_v1b(), when finding obstacles
in candidates tracks, no longer ignore blockages (segments from pads that
connot be broken because they see *no* conflicts...).
- New: In BuildPowerRails, special processing for the power ring segments.
The "diagonal" of vias at each corner is causing a misbehavior of the
routing algorithm (due to fully saturated GCells in one direction).
As a temporary fix, extend the segments so they form a "square corner".
(problem arise on "d_in_i(22)").
- New: In RoutingEvent::_processNegociate, disable the "isForcedToHint()"
feature. No noticeable loss of quality or speed.
- New: In TrackElement/TrackSegment, wraps the AutoSegment parent's mechanism.
Allows to gets the DataNegociate of either the segment or it's parent.
- New: State::solveFullBlockages(), dedicated method to solves the case when
all the allowed tracks of a segment are blocked, tries to moves up
local segments and to break-up global ones.
- New: RoutingEventLoop, a more sophisticated way to detect looping.
Maintain a dynamic histogram of the last N (default 10) segments routeds,
with the count of how many times they have occurred. If that count
exeed 40, we *may* be facing a loop.
- Change: In State::conflictSolve1, implement new policy. The global segments
no more can be broken by local ones. The idea behind is that breaking
a global on the request of a local will only produce more cluttering
in the GCell. Globals must be keep straigth pass through, especially
inside near-saturated GCells. Globals breaking however can occurs at
another global's request.
- Change: In TrackCost, implement the new policy about locals segments that
cannot break globals segments. The sorting class now accept flags to
modulate the sorting function. Two options avalaibles: IgnoreAxisWeigth
(to uses for strap segments) and DiscardGlobals (to uses with locals).
- Change: In TrackCost, the "distance to fixed" have now an upper bound of
50 lambdas (no need to be greater because it means it's outside the
begin & en GCells). Take account not only of fixed segment, but also
of placed segments which makes bound.
- Bug: In Track::_check(), while calling each individual TrackSegment check,
uses it as the *first* argument of the "or", otherwise it may not be
called.
- Bug: In ProtectRoutingPad, loop over segment Collections while modificating
it was producing non-deterministic results. The fact that a collection
must be not modificated while beeing iterated is becoming a more and more
painful problem.
- Change: Propagate renaming "obstacle" -> "blockage".
- Bug/Change: In Configuration, the value of the extensionCap was too big
(1.5 lambda), reduce to 0.5 lambda. This is a problem, the extension
should be coupled to the layer as it is not the same for each METAL.
- Bug: When using TrackElement, always uses the virtual "->isFixed()" method
instead of trying to access to "->base()->isFixed()" as the base may be
NULL in case of blockage/fixed segment.
- Change: Merge PowerRails & Blockage trans-hierarchical construction (into
PowerRails). All blockages are groupeds under "blockagenet".
Allows to remove TrackBlockage & BuildBlockages.
- Change: In KiteEngine::annotateGloblalGraph(), when routing a full chip,
ring power segments around the core must completly saturate the edges
in their segment direction. This is to prevent the global router to
use paths under the power/ground ring (may generate unsolvable configs).
- Change: In KiteEngine::annotateGloblalGraph(), when routing a full chip,
distinguish three areas: the core (65%), the corona (90%) and the pads
(100%). Capacities on the edges are sets accordingly.
- Change: In RoutingEvent, introduce an alternative algorithm for
conflictSolve1, FindPath which try to deduce the breakpoints from a
truly explorated path. Unfortunatly this gives worst results than the
Cs1Candidates method. The why should be investigated as it's a critical
point in the algorithm.
- Change: In Manipulator::ripupPerpandicular(), when a caged perpandicular
is encountered, instead of just "stopping", rip it up and change is
axis hint (actually increase) it's axis hint so it stands a chance to
go outside the track with an obstacle.
- Change: In RoutingEvent/State::slackenTopology(), allow move up of local
segments when they are tightly constrained *and* blocked (cageds).
Partial modification of functions calls from booleans to flags.
- Bug: In NegociateWindow::NegociateOverlapCost, check for fixed segments
before trying to get DataNegociate. The lack of DataNegociate cause the
TrackElement to be discarted. It's a failsafe behavior, but it leads to
overlaps.
- Bug: In ProtectRoutingPad, in Pad Cells only, *do not* protect RoutingPad
to avoid the edge capacity over the pad to decrease to zero. This is
due to unused RoutingPads being accounted as blockages.
- New: In BuilPowerRails & BuildBlockage, specific support for chip-level
design. The Alliance "top chip" design hierarchical structure is
hard-coded. Top level POWER/GROUND/CLOCK nets must always have the
<vsse>, <vdde>, <ck>, <vssi>, <vddi>, <ckc>, <cki> names.
Specific method to get trans-hierarchical root nets, espcially in
case of global ones (POWER/GROUND). Clock is *not* global.
Some correction in the manner obscured tracks are computeds.
- New: ProtectRoutingPad module that perform a more clean work for protecting
unused RoutingPad.
- Bug: In BuildPowerRails, uses stable_sort<> instead of sort<>, which
causes unexplained core dumps (seems to try to perform a comparison
using the "end" pseudo element). Already occured in Knik, no explanation
other than a STL bug.
- Change: Slight changes in the weights to move up. Now needs a full empty
track instead of a half one.
Jean-Paul Chaput
Gabriel Gouvine