coriolis/anabatic/src/GraphicAnabaticEngine.cpp

528 lines
16 KiB
C++
Raw Normal View History

// -*- C++ -*-
//
// This file is part of the Coriolis Software.
// Copyright (c) UPMC 2016-2016, All Rights Reserved
//
// +-----------------------------------------------------------------+
// | C O R I O L I S |
// | A n a b a t i c - Global Routing Toolbox |
// | |
// | Author : Jean-Paul Chaput |
// | E-mail : Jean-Paul.Chaput@lip6.fr |
// | =============================================================== |
// | C++ Module : "GraphicAnabaticEngine.cpp" |
// +-----------------------------------------------------------------+
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
#include <cmath>
#include <boost/bind.hpp>
#include <QAction>
#include <QMenu>
#include <QMenuBar>
#include <QApplication>
#include <hurricane/Warning.h>
#include <hurricane/Error.h>
#include <hurricane/Breakpoint.h>
#include <hurricane/DebugSession.h>
#include <hurricane/UpdateSession.h>
#include <hurricane/Go.h>
#include <hurricane/Net.h>
#include <hurricane/Cell.h>
#include <hurricane/viewer/Graphics.h>
#include <hurricane/viewer/CellWidget.h>
#include <hurricane/viewer/CellViewer.h>
#include <hurricane/viewer/ControllerWidget.h>
#include <hurricane/viewer/ExceptionWidget.h>
#include <crlcore/Utilities.h>
#include <crlcore/AllianceFramework.h>
#include "anabatic/GCell.h"
#include "anabatic/GraphicAnabaticEngine.h"
#include "anabatic/Dijkstra.h"
namespace Anabatic {
using namespace std;
using Hurricane::Error;
using Hurricane::Warning;
using Hurricane::Exception;
using Hurricane::Breakpoint;
using Hurricane::DebugSession;
using Hurricane::UpdateSession;
using Hurricane::Point;
using Hurricane::Entity;
using Hurricane::Net;
using Hurricane::Graphics;
using Hurricane::ColorScale;
using Hurricane::DisplayStyle;
using Hurricane::DrawingStyle;
using Hurricane::ControllerWidget;
using Hurricane::ExceptionWidget;
using CRL::Catalog;
using CRL::AllianceFramework;
// -------------------------------------------------------------------
// Test functions.
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
class DigitalDistance {
public:
inline DigitalDistance ( float h, float k );
DbU::Unit operator() ( const Vertex* source ,const Vertex* target,const Edge* edge ) const;
private:
// For an explanation of h & k parameters, see:
// "KNIK, routeur global pour la plateforme Coriolis", p. 52.
float _h;
float _k;
};
inline DigitalDistance::DigitalDistance ( float h, float k ) : _h(h), _k(k) { }
DbU::Unit DigitalDistance::operator() ( const Vertex* source ,const Vertex* target,const Edge* edge ) const
{
if (edge->getCapacity() <= 0) return Vertex::unreached;
float congestion = (float)edge->getRealOccupancy() / (float)edge->getCapacity();
float congestionCost = 1.0 + _h / (1.0 + std::exp(_k * (congestion - 1.0)));
float distance = (float)source->getDistance() + congestionCost * (float)edge->getDistance();
// Edge* sourceFrom = source->getFrom();
// if (sourceFrom) {
// distance += ((sourceFrom->isHorizontal() xor edge->isHorizontal()) ? 3.0 : 0.0) * (float)Edge::unity;
// }
cdebug_log(112,0) << "cong:" << congestion
<< " ccost:" << congestionCost
<< " digitalDistance:" << DbU::getValueString((DbU::Unit)distance) << endl;
return (DbU::Unit)distance;
}
void anabaticTest_1 ( AnabaticEngine* engine )
{
engine->getSouthWestGCell()->doGrid();
Point position ( DbU::fromLambda(100.0), DbU::fromLambda(100.0) );
GCell* gcell = engine->getGCellUnder( position );
cerr << "Gcell under:" << position << " is " << gcell << endl;
}
void anabaticTest_2 ( AnabaticEngine* engine )
{
UpdateSession::open();
GCell* row0 = engine->getSouthWestGCell();
DbU::Unit xcorner = engine->getCell()->getAbutmentBox().getXMin();
DbU::Unit ycorner = engine->getCell()->getAbutmentBox().getYMin();
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_log(119,1) << "row0: " << row0 << endl;
GCell* row1 = row0->hcut( ycorner+DbU::fromLambda(50.0) );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_tabw(119,-1);
cdebug_log(119,1) << "row1: " << row1 << endl;
GCell* row2 = row1->hcut( ycorner+DbU::fromLambda(2*50.0) );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_tabw(119,-1);
cdebug_log(119,1) << "row2: " << row2 << endl;
row0 = row0->vcut( xcorner+DbU::fromLambda(50.0) );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_tabw(119,-1);
cdebug_log(119,1) << "row0+1: " << row0 << endl;
row0 = row0->vcut( xcorner+DbU::fromLambda(3*50.0) );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_tabw(119,-1);
cdebug_log(119,1) << "row0+2: " << row0 << endl;
row0 = row0->vcut( xcorner+DbU::fromLambda(5*50.0) );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_tabw(119,-1);
cdebug_log(119,1) << "row0+3: " << row0 << endl;
row1 = row1->vcut( xcorner+DbU::fromLambda(2*50.0) );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_tabw(119,-1);
cdebug_log(119,1) << "row1+1: " << row1 << endl;
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_tabw(119,-1);
UpdateSession::close();
}
void anabaticTest_3 ( AnabaticEngine* engine )
{
for ( int i=0 ; i<4 ; ++i ) {
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_log(110,1) << "Running test 3, loop:" << i << " ..." << endl;
if ( i%2) anabaticTest_1( engine );
else anabaticTest_2( engine );
engine->reset();
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
cdebug_log(110,1) << "Test 3, loop:" << i << " completed." << endl;
cdebug_tabw(110,-1);
}
}
void anabaticTest_4 ( AnabaticEngine* engine )
{
Cell* cell = engine->getCell();
cell->flattenNets( Cell::Flags::BuildRings );
cell->createRoutingPadRings( Cell::Flags::BuildRings );
engine->getSouthWestGCell()->doGrid();
vector<Net*> nets;
nets.push_back( cell->getNet( "ialu.inv_x2_3_sig" ) );
nets.push_back( cell->getNet( "iram.na3_x1_13_sig" ) );
nets.push_back( cell->getNet( "iram.ram_idx_7(1)" ) );
Dijkstra* dijkstra = new Dijkstra ( engine );
for ( Net* net : nets ) {
dijkstra->load( net );
dijkstra->run();
}
delete dijkstra;
}
void anabaticTest_5 ( AnabaticEngine* engine )
{
Cell* cell = engine->getCell();
cell->flattenNets( Cell::Flags::BuildRings );
cell->createRoutingPadRings( Cell::Flags::BuildRings );
//DebugSession::addToTrace( cell->getNet("alu_out(3)") );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
//DebugSession::addToTrace( cell->getNet("imuxe.not_i(1)") );
Anabatic transient commit 10. Ripup & reroute support in Dijsktra. * 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.
2016-06-26 07:32:32 -05:00
//DebugSession::addToTrace( cell->getNet("r(0)") );
//DebugSession::addToTrace( cell->getNet("ialu.not_aux104") );
DebugSession::addToTrace( cell->getNet("mips_r3000_1m_dp_shift32_rshift_se_muxoutput(159)") );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
engine->startMeasures();
cmess1 << " o Building regular grid..." << endl;
UpdateSession::open();
engine->getSouthWestGCell()->doGrid();
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
UpdateSession::close();
engine->stopMeasures();
engine->printMeasures( "Anbatic Grid" );
engine->startMeasures();
cmess1 << " o Running global routing..." << endl;
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
UpdateSession::open();
Dijkstra* dijkstra = new Dijkstra ( engine );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
dijkstra->setDistance( DigitalDistance( engine->getConfiguration()->getEdgeCostH()
, engine->getConfiguration()->getEdgeCostK() ) );
for ( Net* net : cell->getNets() ) {
if (net->isPower() or net->isClock()) continue;
dijkstra->load( net );
dijkstra->run();
}
UpdateSession::close();
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
engine->stopMeasures();
engine->printMeasures( "Dijkstra" );
Anabatic transient commit 10. Ripup & reroute support in Dijsktra. * 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.
2016-06-26 07:32:32 -05:00
const vector<Edge*>& ovEdges = engine->getOvEdges();
if (not ovEdges.empty()) {
size_t count = 0;
cmess1 << " - " << ovEdges.size() << " overloaded edges." << endl;
cmess1 << " " << ovEdges[0] << endl;
NetSet nets;
engine->getNetsFromEdge( ovEdges[0], nets );
for ( Net* net : nets ) {
cmess1 << " [" << setw(2) << count++ << "] " << net << endl;
}
UpdateSession::open();
Net* net = *nets.begin();
dijkstra->load( net );
dijkstra->ripup( ovEdges[0] );
UpdateSession::close();
}
UpdateSession::open();
delete dijkstra;
UpdateSession::close();
}
// -------------------------------------------------------------------
// Class : "Anabatic::GraphicAnabaticEngine".
size_t GraphicAnabaticEngine::_references = 0;
GraphicAnabaticEngine* GraphicAnabaticEngine::_singleton = NULL;
void GraphicAnabaticEngine::initGCell ( CellWidget* widget )
{
widget->getDrawingPlanes().setPen( Qt::NoPen );
}
void GraphicAnabaticEngine::drawGCell ( CellWidget* widget
, const Go* go
, const BasicLayer* basicLayer
, const Box& box
, const Transformation& transformation
)
{
const GCell* gcell = static_cast<const GCell*>(go);
QPainter& painter = widget->getPainter();
QPen pen = Graphics::getPen ("Anabatic::GCell",widget->getDarkening());
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
Box bb = gcell->getBoundingBox();
QRect pixelBb = widget->dbuToScreenRect(bb);
painter.setPen ( pen );
painter.setBrush( Graphics::getBrush("Anabatic::GCell",widget->getDarkening()) );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
painter.drawRect( pixelBb );
if (gcell->isFlat()) return;
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
if (pixelBb.width() > 150) {
QString text = QString("id:%1").arg(gcell->getId());
QFont font = Graphics::getFixedFont( QFont::Bold );
painter.setFont(font);
pen.setWidth( 1 );
painter.setPen( pen );
painter.save ();
painter.translate( widget->dbuToScreenPoint(bb.getCenter().getX(), bb.getCenter().getY()) );
painter.drawRect (QRect( -75, -25, 150, 50 ));
painter.drawText (QRect( -75, -25, 150, 50 )
, text
, QTextOption(Qt::AlignCenter)
);
painter.restore ();
}
}
void GraphicAnabaticEngine::initEdge ( CellWidget* widget )
{
widget->getDrawingPlanes().setPen( Qt::NoPen );
}
void GraphicAnabaticEngine::drawEdge ( CellWidget* widget
, const Go* go
, const BasicLayer* basicLayer
, const Box& box
, const Transformation& transformation
)
{
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
static QFont font = Graphics::getFixedFont( QFont::Bold );
static int fontHeight = QFontMetrics(font).height();
const Edge* edge = static_cast<const Edge*>(go);
if (edge) {
Box bb = edge->getBoundingBox();
unsigned int occupancy = 255;
if (edge->getRealOccupancy() < edge->getCapacity())
occupancy = (unsigned int)( 255.0 * ( (float)edge->getRealOccupancy() / (float)edge->getCapacity() ) );
QPainter& painter = widget->getPainter();
if (edge->getRealOccupancy() > edge->getCapacity()) {
QColor color ( Qt::cyan );
painter.setPen( DisplayStyle::darken(color,widget->getDarkening()) );
}
QBrush brush = QBrush( Qt::white, Qt::DiagCrossPattern );
if (edge->getCapacity() > 0.0)
brush = Graphics::getColorScale( ColorScale::Fire ).getBrush( occupancy, widget->getDarkening() );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
QRect pixelBb = widget->dbuToScreenRect( bb, false);
painter.setPen( Qt::NoPen );
painter.setBrush( brush );
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
painter.drawRect( pixelBb );
if (fontHeight > ((edge->isHorizontal()) ? pixelBb.height() : pixelBb.width()) + 4) return;
QString text = QString("%1/%2").arg(edge->getRealOccupancy()).arg(edge->getCapacity());
Anabatic transient commit 8. More Dijkstra bugs correcteds. * 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.
2016-06-17 06:09:34 -05:00
QColor color ( (occupancy > 170) ? Qt::black : Qt::white );
painter.setPen (DisplayStyle::darken(color,widget->getDarkening()));
painter.setFont(font);
if (edge->isVertical()) {
painter.save ();
painter.translate( widget->dbuToScreenPoint(bb.getXMin(), bb.getYMin()) );
painter.rotate ( -90 );
painter.drawText (QRect( 0
, 0
, widget->dbuToScreenLength(bb.getHeight())
, widget->dbuToScreenLength(bb.getWidth ()))
, text
, QTextOption(Qt::AlignCenter)
);
painter.restore ();
} else
painter.drawText( widget->dbuToScreenRect(bb,false ), text, QTextOption(Qt::AlignCenter) );
painter.setPen( Qt::NoPen );
}
}
AnabaticEngine* GraphicAnabaticEngine::createEngine ()
{
Cell* cell = getCell();
AnabaticEngine* engine = AnabaticEngine::get( cell );
if (not engine) {
engine = AnabaticEngine::create( cell );
} else
cerr << Warning( "%s already has a Anabatic engine.", getString(cell).c_str() ) << endl;
return engine;
}
AnabaticEngine* GraphicAnabaticEngine::getForFramework ( unsigned int flags )
{
// Currently, only one framework is avalaible: Alliance.
AnabaticEngine* engine = AnabaticEngine::get( getCell() );
if (engine) return engine;
if (flags & CreateEngine) {
engine = createEngine();
if (not engine)
throw Error( "Failed to create Anabatic engine on %s.", getString(getCell()).c_str() );
} else {
throw Error( "AnabaticEngine not created yet, run the global router first." );
}
return engine;
}
void GraphicAnabaticEngine::_runTest ()
{
if (_viewer) _viewer->emitCellAboutToChange();
AnabaticEngine* engine = getForFramework( CreateEngine );
//anabaticTest_1( engine );
//anabaticTest_2( engine );
//anabaticTest_3( engine );
//anabaticTest_4( engine );
anabaticTest_5( engine );
if (_viewer) _viewer->emitCellChanged();
}
Anabatic transient commit 10. Ripup & reroute support in Dijsktra. * 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.
2016-06-26 07:32:32 -05:00
void GraphicAnabaticEngine::_globalRoute ()
{
AnabaticEngine* engine = getForFramework( CreateEngine );
engine->globalRoute();
}
void GraphicAnabaticEngine::_detailRoute ()
{
AnabaticEngine* engine = getForFramework( CreateEngine );
engine->loadGlobalRouting( EngineLoadGrByNet );
engine->layerAssign( EngineNoNetLayerAssign );
}
void GraphicAnabaticEngine::addToMenu ( CellViewer* viewer )
{
assert( _viewer == NULL );
_viewer = viewer;
Anabatic transient commit 10. Ripup & reroute support in Dijsktra. * 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.
2016-06-26 07:32:32 -05:00
if (_viewer->hasMenuAction("placeAndRoute.anabatic.globalRoute")) {
cerr << Warning( "GraphicAnabaticEngine::addToMenu() - Anabatic engine already hooked in." ) << endl;
return;
}
Anabatic transient commit 10. Ripup & reroute support in Dijsktra. * 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.
2016-06-26 07:32:32 -05:00
_viewer->addMenu ( "placeAndRoute.anabatic", "Anabatic" );
_viewer->addToMenu( "placeAndRoute.anabatic.runTest"
, "Anabatic - &Test Run"
, "Perform a test run of Anabatic on the design"
, std::bind(&GraphicAnabaticEngine::_runTest,this)
);
Anabatic transient commit 10. Ripup & reroute support in Dijsktra. * 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.
2016-06-26 07:32:32 -05:00
_viewer->addToMenu( "placeAndRoute.anabatic.globalRoute"
, "Anabatic - &Global Route"
, "Global Route"
, std::bind(&GraphicAnabaticEngine::_globalRoute,this)
);
_viewer->addToMenu( "placeAndRoute.anabatic.detailedRoute"
, "Anabatic - &Detailed Route"
, "Run the <b>Anabatic</b> detailed router"
, std::bind(&GraphicAnabaticEngine::_detailRoute,this)
);
}
const Name& GraphicAnabaticEngine::getName () const
{ return AnabaticEngine::staticGetName(); }
Cell* GraphicAnabaticEngine::getCell ()
{
if (_viewer == NULL) {
throw Error( "<b>Anabatic:</b> GraphicAnabaticEngine not bound to any Viewer." );
return NULL;
}
if (_viewer->getCell() == NULL) {
throw Error( "<b>Anabatic:</b> No Cell is loaded into the Viewer." );
return NULL;
}
return _viewer->getCell();
}
GraphicAnabaticEngine* GraphicAnabaticEngine::grab ()
{
if (not _references)
_singleton = new GraphicAnabaticEngine ();
_references++;
return _singleton;
}
size_t GraphicAnabaticEngine::release ()
{
--_references;
if (not _references) {
delete _singleton;
_singleton = NULL;
}
return _references;
}
GraphicAnabaticEngine::GraphicAnabaticEngine ()
: GraphicTool()
, _viewer (NULL)
{
addDrawGo( "Anabatic::GCell", initGCell, drawGCell );
addDrawGo( "Anabatic::Edge" , initEdge , drawEdge );
}
GraphicAnabaticEngine::~GraphicAnabaticEngine ()
{ }
} // Anabatic namespace.