summaryrefslogtreecommitdiffstats
path: root/src/aig/gia
diff options
context:
space:
mode:
authorAlan Mishchenko <alanmi@berkeley.edu>2014-06-23 13:11:59 -0700
committerAlan Mishchenko <alanmi@berkeley.edu>2014-06-23 13:11:59 -0700
commit44d9c7e54307f64cbcdc7c8cd17ff6e219e13b55 (patch)
tree2ae1aa5fef6ca8198f0c2fb6123c3d1494dbe2b4 /src/aig/gia
parentf93e5244219b75224df0c75b424654bd2424852b (diff)
downloadabc-44d9c7e54307f64cbcdc7c8cd17ff6e219e13b55.tar.gz
abc-44d9c7e54307f64cbcdc7c8cd17ff6e219e13b55.tar.bz2
abc-44d9c7e54307f64cbcdc7c8cd17ff6e219e13b55.zip
Improvements to CNF generation.
Diffstat (limited to 'src/aig/gia')
-rw-r--r--src/aig/gia/giaJf.c13
-rw-r--r--src/aig/gia/giaMf.c1582
2 files changed, 1588 insertions, 7 deletions
diff --git a/src/aig/gia/giaJf.c b/src/aig/gia/giaJf.c
index b5aa7787..0a0e7d48 100644
--- a/src/aig/gia/giaJf.c
+++ b/src/aig/gia/giaJf.c
@@ -1750,23 +1750,30 @@ Gia_Man_t * Jf_ManDeriveCnf( Gia_Man_t * p, int fCnfObjIds )
pPars->fCnfObjIds = fCnfObjIds;
return Jf_ManPerformMapping( p, pPars );
}
-Gia_Man_t * Jf_ManDeriveCnfMiter( Gia_Man_t * p )
+Gia_Man_t * Jf_ManDeriveCnfMiter( Gia_Man_t * p, int fVerbose )
{
Jf_Par_t Pars, * pPars = &Pars;
Jf_ManSetDefaultPars( pPars );
pPars->fGenCnf = 1;
pPars->fCnfObjIds = 0;
pPars->fAddOrCla = 1;
+ pPars->fVerbose = fVerbose;
return Jf_ManPerformMapping( p, pPars );
}
-void Jf_ManDumpCnf( Gia_Man_t * p, char * pFileName )
+void Jf_ManDumpCnf( Gia_Man_t * p, char * pFileName, int fVerbose )
{
+ abctime clk = Abc_Clock();
Gia_Man_t * pNew;
Cnf_Dat_t * pCnf;
- pNew = Jf_ManDeriveCnfMiter( p );
+ pNew = Jf_ManDeriveCnfMiter( p, fVerbose );
pCnf = (Cnf_Dat_t *)pNew->pData; pNew->pData = NULL;
Cnf_DataWriteIntoFile( pCnf, pFileName, 0, NULL, NULL );
Gia_ManStop( pNew );
+// if ( fVerbose )
+ {
+ printf( "CNF stats: Vars = %6d. Clauses = %7d. Literals = %8d. ", pCnf->nVars, pCnf->nClauses, pCnf->nLiterals );
+ Abc_PrintTime( 1, "Time", Abc_Clock() - clk );
+ }
Cnf_DataFree(pCnf);
}
diff --git a/src/aig/gia/giaMf.c b/src/aig/gia/giaMf.c
index c43b6aec..1a4da7a5 100644
--- a/src/aig/gia/giaMf.c
+++ b/src/aig/gia/giaMf.c
@@ -8,7 +8,7 @@
Synopsis [Cut computation.]
- Author [Alan Mishchenko]
+ Author [Alan Mishchenko]`
Affiliation [UC Berkeley]
@@ -19,7 +19,10 @@
***********************************************************************/
#include "gia.h"
-#include "misc/vec/vecSet.h"
+#include "misc/vec/vecMem.h"
+#include "misc/util/utilTruth.h"
+#include "misc/extra/extra.h"
+#include "sat/cnf/cnf.h"
ABC_NAMESPACE_IMPL_START
@@ -27,11 +30,1314 @@ ABC_NAMESPACE_IMPL_START
/// DECLARATIONS ///
////////////////////////////////////////////////////////////////////////
+#define MF_LEAF_MAX 10
+#define MF_CUT_MAX 16
+#define MF_LOG_PAGE 12
+#define MF_NO_LEAF 31
+#define MF_TT_WORDS ((MF_LEAF_MAX > 6) ? 1 << (MF_LEAF_MAX-6) : 1)
+
+typedef struct Mf_Cut_t_ Mf_Cut_t;
+struct Mf_Cut_t_
+{
+ word Sign; // signature
+ int Delay; // delay
+ float Flow; // flow
+ unsigned iFunc : 27; // function
+ unsigned nLeaves : 5; // leaf number (MF_NO_LEAF)
+ int pLeaves[MF_LEAF_MAX+1]; // leaves
+};
+typedef struct Mf_Obj_t_ Mf_Obj_t;
+struct Mf_Obj_t_
+{
+ int iCutSet; // cutset
+ float Flow; // area
+ float nFlowRefs; // flow references
+ unsigned Delay : 16; // delay
+ unsigned nMapRefs : 16; // map references
+};
+typedef struct Mf_Man_t_ Mf_Man_t;
+struct Mf_Man_t_
+{
+ // user data
+ Gia_Man_t * pGia0; // original manager
+ Gia_Man_t * pGia; // derived manager
+ Jf_Par_t * pPars; // parameters
+ // cut data
+ Mf_Obj_t * pLfObjs; // best cuts
+ Vec_Ptr_t vPages; // cut memory
+ Vec_Mem_t * vTtMem; // truth tables
+ Vec_Int_t vCnfSizes; // handles to CNF
+ Vec_Int_t vCnfMem; // memory for CNF
+ int iCur; // current position
+ int Iter; // mapping iterations
+ int fUseEla; // use exact area
+ // statistics
+ abctime clkStart; // starting time
+ double CutCount[4]; // cut counts
+ int nCutCounts[MF_LEAF_MAX+1];
+};
+
+static inline Mf_Obj_t * Mf_ManObj( Mf_Man_t * p, int i ) { return p->pLfObjs + i; }
+static inline int * Mf_ManCutSet( Mf_Man_t * p, int i ) { return (int *)Vec_PtrEntry(&p->vPages, i >> 16) + (i & 0xFFFF); }
+static inline int * Mf_ObjCutSet( Mf_Man_t * p, int i ) { return Mf_ManCutSet(p, Mf_ManObj(p, i)->iCutSet); }
+static inline int * Mf_ObjCutBest( Mf_Man_t * p, int i ) { return Mf_ObjCutSet(p, i) + 1; }
+
+static inline int Mf_ObjMapRefNum( Mf_Man_t * p, int i ) { return Mf_ManObj(p, i)->nMapRefs; }
+static inline int Mf_ObjMapRefInc( Mf_Man_t * p, int i ) { return Mf_ManObj(p, i)->nMapRefs++; }
+static inline int Mf_ObjMapRefDec( Mf_Man_t * p, int i ) { return --Mf_ManObj(p, i)->nMapRefs; }
+
+static inline int Mf_CutSize( int * pCut ) { return pCut[0] & MF_NO_LEAF; }
+static inline int Mf_CutFunc( int * pCut ) { return ((unsigned)pCut[0] >> 5); }
+static inline int Mf_CutSetBoth( int n, int f ) { return n | (f << 5); }
+static inline int Mf_CutIsTriv( int * pCut, int i ) { return Mf_CutSize(pCut) == 1 && pCut[1] == i; }
+
+#define Mf_SetForEachCut( pList, pCut, i ) for ( i = 0, pCut = pList + 1; i < pList[0]; i++, pCut += Mf_CutSize(pCut) + 1 )
+#define Mf_ObjForEachCut( pCuts, i, nCuts ) for ( i = 0, i < nCuts; i++ )
+
+extern int Kit_TruthToGia( Gia_Man_t * pMan, unsigned * pTruth, int nVars, Vec_Int_t * vMemory, Vec_Int_t * vLeaves, int fHash );
+extern void Dau_DsdPrintFromTruth( word * pTruth, int nVarsInit );
////////////////////////////////////////////////////////////////////////
/// FUNCTION DEFINITIONS ///
////////////////////////////////////////////////////////////////////////
+/**Function*************************************************************
+
+ Synopsis [Computing truth tables of useful DSD classes of 6-functions.]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+static int s_nCalls = 0;
+static Vec_Mem_t * s_vTtMem = NULL;
+int Mf_ManTruthCanonicize( word * t, int nVars )
+{
+ word Temp, Best = *t;
+ int r, i, Config = 0;
+ for ( r = 0; r < 1; r++ )
+ {
+ if ( Best > (Temp = ~Best) )
+ Best = Temp, Config ^= (1 << nVars);
+ for ( i = 0; i < nVars; i++ )
+ if ( Best > (Temp = Abc_Tt6Flip(Best, i)) )
+ Best = Temp, Config ^= (1 << i);
+ }
+ *t = Best;
+ if ( s_vTtMem == NULL )
+ s_vTtMem = Vec_MemAllocForTT( 6, 0 );
+ Vec_MemHashInsert( s_vTtMem, t );
+ s_nCalls++;
+ return Config;
+}
+void Mf_ManTruthQuit()
+{
+ if ( s_vTtMem == NULL )
+ return;
+ printf( "TT = %d (%.2f %%)\n", Vec_MemEntryNum(s_vTtMem), 100.0 * Vec_MemEntryNum(s_vTtMem) / s_nCalls );
+ Vec_MemHashFree( s_vTtMem );
+ Vec_MemFree( s_vTtMem );
+ s_vTtMem = NULL;
+ s_nCalls = 0;
+}
+
+Vec_Wrd_t * Mf_ManTruthCollect( int Limit )
+{
+ extern Vec_Wrd_t * Mpm_ManGetTruthWithCnf( int Limit );
+ int * pPerm = Extra_PermSchedule( 6 );
+ int * pComp = Extra_GreyCodeSchedule( 6 );
+ Vec_Wrd_t * vTruths = Mpm_ManGetTruthWithCnf( Limit );
+ Vec_Wrd_t * vResult = Vec_WrdAlloc( 1 << 20 );
+ word uTruth, tCur, tTemp1, tTemp2;
+ int i, p, c, k;
+ Vec_WrdForEachEntry( vTruths, uTruth, k )
+ {
+ for ( i = 0; i < 2; i++ )
+ {
+ tCur = i ? ~uTruth : uTruth;
+ tTemp1 = tCur;
+ for ( p = 0; p < 720; p++ )
+ {
+ tTemp2 = tCur;
+ for ( c = 0; c < 64; c++ )
+ {
+ tCur = Abc_Tt6Flip( tCur, pComp[c] );
+ Vec_WrdPush( vResult, tCur );
+ }
+ assert( tTemp2 == tCur );
+ tCur = Abc_Tt6SwapAdjacent( tCur, pPerm[p] );
+ }
+ assert( tTemp1 == tCur );
+ }
+ }
+ ABC_FREE( pPerm );
+ ABC_FREE( pComp );
+ printf( "Original = %d. ", Vec_WrdSize(vTruths) );
+ Vec_WrdFree( vTruths );
+ printf( "Total = %d. ", Vec_WrdSize(vResult) );
+ vTruths = Vec_WrdUniqifyHash( vResult, 1 );
+ Vec_WrdFree( vResult );
+ printf( "Unique = %d. ", Vec_WrdSize(vTruths) );
+ Vec_WrdForEachEntry( vTruths, uTruth, k )
+ {
+ Mf_ManTruthCanonicize( &uTruth, 6 );
+ Vec_WrdWriteEntry( vTruths, k, uTruth );
+ }
+ vResult = Vec_WrdUniqifyHash( vTruths, 1 );
+ Vec_WrdFree( vTruths );
+ printf( "Unique = %d. \n", Vec_WrdSize(vResult) );
+ return vResult;
+}
+int Mf_ManTruthCount()
+{
+ Vec_Wrd_t * vTruths = Mf_ManTruthCollect( 10 );
+ int RetValue = Vec_WrdSize( vTruths );
+ Vec_WrdFree( vTruths );
+ return RetValue;
+}
+
+/**Function*************************************************************
+
+ Synopsis [Collect truth tables used by the mapper.]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+void Mf_ManProfileTruths( Mf_Man_t * p )
+{
+ Vec_Int_t * vCounts;
+ int i, Entry, * pCut, Counter = 0;
+ vCounts = Vec_IntStart( Vec_IntSize(&p->vCnfSizes) );
+ Gia_ManForEachAndId( p->pGia, i )
+ {
+ if ( !Mf_ObjMapRefNum(p, i) )
+ continue;
+ pCut = Mf_ObjCutBest( p, i );
+ Vec_IntAddToEntry( vCounts, Abc_Lit2Var(Mf_CutFunc(pCut)), 1 );
+ }
+ Vec_IntForEachEntry( vCounts, Entry, i )
+ {
+ if ( Entry == 0 )
+ continue;
+ printf( "%6d : ", Counter++ );
+ printf( "%6d : ", i );
+ printf( "Occur = %4d ", Entry );
+ printf( "CNF size = %2d ", Vec_IntEntry(&p->vCnfSizes, i) );
+ Dau_DsdPrintFromTruth( Vec_MemReadEntry(p->vTtMem, i), p->pPars->nLutSize );
+ }
+ Vec_IntFree( vCounts );
+}
+
+/**Function*************************************************************
+
+ Synopsis [Derives CNFs for each function used in the mapping.]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+static inline int Mf_CubeLit( int Cube, int iVar ) { return (Cube >> (iVar << 1)) & 3; }
+static inline int Mf_ManCountLits( int * pCnf, int nCubes, int nVars )
+{
+ int i, k, nLits = nCubes;
+ for ( i = 0; i < nCubes; i++ )
+ for ( k = 0; k < nVars; k++ )
+ if ( Mf_CubeLit(pCnf[i], k) )
+ nLits++;
+ return nLits;
+}
+Vec_Int_t * Mf_ManDeriveCnfs( Mf_Man_t * p, int * pnVars, int * pnClas, int * pnLits )
+{
+ int i, k, iFunc, nCubes, nLits, * pCut, pCnf[512];
+ Vec_Int_t * vLits = Vec_IntStart( Vec_IntSize(&p->vCnfSizes) );
+ Vec_Int_t * vCnfs = Vec_IntAlloc( 3 * Vec_IntSize(&p->vCnfSizes) );
+ Vec_IntFill( vCnfs, Vec_IntSize(&p->vCnfSizes), -1 );
+ assert( p->pPars->nLutSize <= 8 );
+ // constant/buffer
+ for ( iFunc = 0; iFunc < 2; iFunc++ )
+ {
+ if ( p->pPars->nLutSize <= 6 )
+ nCubes = Abc_Tt6Cnf( *Vec_MemReadEntry(p->vTtMem, iFunc), iFunc, pCnf );
+ else
+ nCubes = Abc_Tt8Cnf( Vec_MemReadEntry(p->vTtMem, iFunc), iFunc, pCnf );
+ nLits = Mf_ManCountLits( pCnf, nCubes, iFunc );
+ Vec_IntWriteEntry( vLits, iFunc, nLits );
+ Vec_IntWriteEntry( vCnfs, iFunc, Vec_IntSize(vCnfs) );
+ Vec_IntPush( vCnfs, nCubes );
+ for ( k = 0; k < nCubes; k++ )
+ Vec_IntPush( vCnfs, pCnf[k] );
+ }
+ // other functions
+ *pnVars = 1 + Gia_ManCiNum(p->pGia) + Gia_ManCoNum(p->pGia);
+ *pnClas = 1 + 2 * Gia_ManCoNum(p->pGia);
+ *pnLits = 1 + 4 * Gia_ManCoNum(p->pGia);
+ Gia_ManForEachAndId( p->pGia, i )
+ {
+ if ( !Mf_ObjMapRefNum(p, i) )
+ continue;
+ pCut = Mf_ObjCutBest( p, i );
+ iFunc = Abc_Lit2Var( Mf_CutFunc(pCut) );
+ if ( Vec_IntEntry(vCnfs, iFunc) == -1 )
+ {
+ if ( p->pPars->nLutSize <= 6 )
+ nCubes = Abc_Tt6Cnf( *Vec_MemReadEntry(p->vTtMem, iFunc), Mf_CutSize(pCut), pCnf );
+ else
+ nCubes = Abc_Tt8Cnf( Vec_MemReadEntry(p->vTtMem, iFunc), Mf_CutSize(pCut), pCnf );
+ assert( nCubes == Vec_IntEntry(&p->vCnfSizes, iFunc) );
+ nLits = Mf_ManCountLits( pCnf, nCubes, Mf_CutSize(pCut) );
+ // save CNF
+ Vec_IntWriteEntry( vLits, iFunc, nLits );
+ Vec_IntWriteEntry( vCnfs, iFunc, Vec_IntSize(vCnfs) );
+ Vec_IntPush( vCnfs, nCubes );
+ for ( k = 0; k < nCubes; k++ )
+ Vec_IntPush( vCnfs, pCnf[k] );
+ }
+ *pnVars += 1;
+ *pnClas += Vec_IntEntry(&p->vCnfSizes, iFunc);
+ *pnLits += Vec_IntEntry(vLits, iFunc);
+ }
+ Vec_IntFree( vLits );
+ return vCnfs;
+}
+
+/**Function*************************************************************
+
+ Synopsis [Derives CNF for the AIG using the mapping.]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+Cnf_Dat_t * Mf_ManDeriveCnf( Mf_Man_t * p, int fCnfObjIds, int fAddOrCla )
+{
+ Cnf_Dat_t * pCnf;
+ Gia_Obj_t * pObj;
+ int Id, DriId, nVars, nClas, nLits, iVar = 1, iCla = 0, iLit = 0;
+ Vec_Int_t * vCnfs = Mf_ManDeriveCnfs( p, &nVars, &nClas, &nLits );
+ Vec_Int_t * vCnfIds = Vec_IntStartFull( Gia_ManObjNum(p->pGia) );
+ int pFanins[16], * pCut, * pCnfIds = Vec_IntArray( vCnfIds );
+ int i, k, c, iFunc, nCubes, * pCubes, fComplLast;
+ nVars++; // zero-ID to remain unused
+ if ( fAddOrCla )
+ {
+ nClas++;
+ nLits += Gia_ManCoNum(p->pGia);
+ }
+ // create CNF IDs
+ if ( fCnfObjIds )
+ {
+ int iVar = 1;
+ iVar += 1 + Gia_ManCiNum(p->pGia) + Gia_ManCoNum(p->pGia);
+ Gia_ManForEachCoId( p->pGia, Id, i )
+ Vec_IntWriteEntry( vCnfIds, Id, Id );
+ Gia_ManForEachAndReverseId( p->pGia, Id )
+ if ( Mf_ObjMapRefNum(p, Id) )
+ Vec_IntWriteEntry( vCnfIds, Id, Id ), iVar++;
+ Gia_ManForEachCiId( p->pGia, Id, i )
+ Vec_IntWriteEntry( vCnfIds, Id, Id );
+ Vec_IntWriteEntry( vCnfIds, 0, 0 );
+ assert( iVar == nVars );
+ }
+ else
+ {
+ int iVar = 1;
+ Gia_ManForEachCoId( p->pGia, Id, i )
+ Vec_IntWriteEntry( vCnfIds, Id, iVar++ );
+ Gia_ManForEachAndReverseId( p->pGia, Id )
+ if ( Mf_ObjMapRefNum(p, Id) )
+ Vec_IntWriteEntry( vCnfIds, Id, iVar++ );
+ Gia_ManForEachCiId( p->pGia, Id, i )
+ Vec_IntWriteEntry( vCnfIds, Id, iVar++ );
+ Vec_IntWriteEntry( vCnfIds, 0, iVar++ );
+ assert( iVar == nVars );
+ }
+ // generate CNF
+ pCnf = ABC_CALLOC( Cnf_Dat_t, 1 );
+ pCnf->pMan = (Aig_Man_t *)p->pGia;
+ pCnf->nVars = nVars;
+ pCnf->nLiterals = nLits;
+ pCnf->nClauses = nClas;
+ pCnf->pClauses = ABC_ALLOC( int *, nClas+1 );
+ pCnf->pClauses[0] = ABC_ALLOC( int, nLits );
+ // add last clause
+ if ( fAddOrCla )
+ {
+ pCnf->pClauses[iCla++] = pCnf->pClauses[0] + iLit;
+ Gia_ManForEachCoId( p->pGia, Id, i )
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit(pCnfIds[Id], 0);
+ }
+ // add clauses for the COs
+ Gia_ManForEachCo( p->pGia, pObj, i )
+ {
+ Id = Gia_ObjId( p->pGia, pObj );
+ DriId = Gia_ObjFaninId0( pObj, Id );
+
+ pCnf->pClauses[iCla++] = pCnf->pClauses[0] + iLit;
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit(pCnfIds[Id], 0);
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit(pCnfIds[DriId], !Gia_ObjFaninC0(pObj));
+
+ pCnf->pClauses[iCla++] = pCnf->pClauses[0] + iLit;
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit(pCnfIds[Id], 1);
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit(pCnfIds[DriId], Gia_ObjFaninC0(pObj));
+ }
+ // add clauses for the mapping
+ Gia_ManForEachAndReverseId( p->pGia, Id )
+ {
+ if ( !Mf_ObjMapRefNum(p, Id) )
+ continue;
+ pCut = Mf_ObjCutBest( p, Id );
+ iFunc = Abc_Lit2Var( Mf_CutFunc(pCut) );
+ //Dau_DsdPrintFromTruth( Vec_MemReadEntry(p->vTtMem, iFunc), 3 );
+ fComplLast = Abc_LitIsCompl( Mf_CutFunc(pCut) );
+ for ( k = 0; k < Mf_CutSize(pCut); k++ )
+ pFanins[k] = pCnfIds[pCut[k+1]];
+ pFanins[k++] = pCnfIds[Id];
+ // get clauses
+ pCubes = Vec_IntEntryP( vCnfs, Vec_IntEntry(vCnfs, iFunc) );
+ nCubes = *pCubes++;
+ for ( c = 0; c < nCubes; c++ )
+ {
+ pCnf->pClauses[iCla++] = pCnf->pClauses[0] + iLit;
+ k = Mf_CutSize(pCut);
+ assert( Mf_CubeLit(pCubes[c], k) );
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit( pFanins[k], (Mf_CubeLit(pCubes[c], k) == 2) ^ fComplLast );
+ for ( k = 0; k < Mf_CutSize(pCut); k++ )
+ if ( Mf_CubeLit(pCubes[c], k) )
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit( pFanins[k], Mf_CubeLit(pCubes[c], k) == 2 );
+ }
+ }
+ // constant clause
+ pCnf->pClauses[iCla++] = pCnf->pClauses[0] + iLit;
+ pCnf->pClauses[0][iLit++] = Abc_Var2Lit(pCnfIds[0], 1);
+ assert( iCla == nClas );
+ assert( iLit == nLits );
+ // add closing pointer
+ pCnf->pClauses[iCla++] = pCnf->pClauses[0] + iLit;
+ // cleanup
+ Vec_IntFree( vCnfs );
+ // create mapping of objects into their clauses
+ if ( fCnfObjIds )
+ {
+ pCnf->pObj2Clause = ABC_FALLOC( int, Gia_ManObjNum(p->pGia) );
+ pCnf->pObj2Count = ABC_FALLOC( int, Gia_ManObjNum(p->pGia) );
+ for ( i = 0; i < pCnf->nClauses; i++ )
+ {
+ Id = Abc_Lit2Var(pCnf->pClauses[i][0]);
+ if ( pCnf->pObj2Clause[Id] == -1 )
+ {
+ pCnf->pObj2Clause[Id] = i;
+ pCnf->pObj2Count[Id] = 1;
+ }
+ else
+ {
+ assert( pCnf->pObj2Count[Id] > 0 );
+ pCnf->pObj2Count[Id]++;
+ }
+ }
+ }
+ else
+ {
+ if ( p->pGia != p->pGia0 ) // diff managers - create map for CIs/COs
+ {
+ pCnf->pVarNums = ABC_FALLOC( int, Gia_ManObjNum(p->pGia0) );
+ Gia_ManForEachCiId( p->pGia0, Id, i )
+ pCnf->pVarNums[Id] = pCnfIds[Gia_ManCiIdToId(p->pGia, i)];
+ Gia_ManForEachCoId( p->pGia0, Id, i )
+ pCnf->pVarNums[Id] = pCnfIds[Gia_ManCoIdToId(p->pGia, i)];
+ }
+ else
+ pCnf->pVarNums = Vec_IntReleaseArray(vCnfIds);
+ }
+ Vec_IntFree( vCnfIds );
+ return pCnf;
+}
+
+/**Function*************************************************************
+
+ Synopsis []
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+static inline int Mf_CutComputeTruth6( Mf_Man_t * p, Mf_Cut_t * pCut0, Mf_Cut_t * pCut1, int fCompl0, int fCompl1, Mf_Cut_t * pCutR, int fIsXor )
+{
+// extern int Mf_ManTruthCanonicize( word * t, int nVars );
+ int nOldSupp = pCutR->nLeaves, nCubes = 0, truthId, fCompl; word t;
+ word t0 = *Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut0->iFunc));
+ word t1 = *Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut1->iFunc));
+ if ( Abc_LitIsCompl(pCut0->iFunc) ^ fCompl0 ) t0 = ~t0;
+ if ( Abc_LitIsCompl(pCut1->iFunc) ^ fCompl1 ) t1 = ~t1;
+ t0 = Abc_Tt6Expand( t0, pCut0->pLeaves, pCut0->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ t1 = Abc_Tt6Expand( t1, pCut1->pLeaves, pCut1->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ t = fIsXor ? t0 ^ t1 : t0 & t1;
+ if ( (fCompl = (int)(t & 1)) ) t = ~t;
+ pCutR->nLeaves = Abc_Tt6MinBase( &t, pCutR->pLeaves, pCutR->nLeaves );
+ assert( (int)(t & 1) == 0 );
+ truthId = Vec_MemHashInsert(p->vTtMem, &t);
+ pCutR->iFunc = Abc_Var2Lit( truthId, fCompl );
+ if ( p->pPars->fGenCnf && truthId == Vec_IntSize(&p->vCnfSizes) )
+ Vec_IntPush( &p->vCnfSizes, Abc_Tt6CnfSize(t, pCutR->nLeaves) );
+// p->nCutMux += Mf_ManTtIsMux( t );
+ assert( (int)pCutR->nLeaves <= nOldSupp );
+// Mf_ManTruthCanonicize( &t, pCutR->nLeaves );
+ return (int)pCutR->nLeaves < nOldSupp;
+}
+static inline int Mf_CutComputeTruth( Mf_Man_t * p, Mf_Cut_t * pCut0, Mf_Cut_t * pCut1, int fCompl0, int fCompl1, Mf_Cut_t * pCutR, int fIsXor )
+{
+ if ( p->pPars->nLutSize <= 6 )
+ return Mf_CutComputeTruth6( p, pCut0, pCut1, fCompl0, fCompl1, pCutR, fIsXor );
+ {
+ word uTruth[MF_TT_WORDS], uTruth0[MF_TT_WORDS], uTruth1[MF_TT_WORDS];
+ int nOldSupp = pCutR->nLeaves, truthId;
+ int LutSize = p->pPars->nLutSize, fCompl;
+ int nWords = Abc_Truth6WordNum(LutSize);
+ word * pTruth0 = Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut0->iFunc));
+ word * pTruth1 = Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut1->iFunc));
+ Abc_TtCopy( uTruth0, pTruth0, nWords, Abc_LitIsCompl(pCut0->iFunc) ^ fCompl0 );
+ Abc_TtCopy( uTruth1, pTruth1, nWords, Abc_LitIsCompl(pCut1->iFunc) ^ fCompl1 );
+ Abc_TtExpand( uTruth0, LutSize, pCut0->pLeaves, pCut0->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ Abc_TtExpand( uTruth1, LutSize, pCut1->pLeaves, pCut1->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ if ( fIsXor )
+ Abc_TtXor( uTruth, uTruth0, uTruth1, nWords, (fCompl = (int)((uTruth0[0] ^ uTruth1[0]) & 1)) );
+ else
+ Abc_TtAnd( uTruth, uTruth0, uTruth1, nWords, (fCompl = (int)((uTruth0[0] & uTruth1[0]) & 1)) );
+ pCutR->nLeaves = Abc_TtMinBase( uTruth, pCutR->pLeaves, pCutR->nLeaves, LutSize );
+ assert( (uTruth[0] & 1) == 0 );
+//Kit_DsdPrintFromTruth( uTruth, pCutR->nLeaves ), printf("\n" ), printf("\n" );
+ truthId = Vec_MemHashInsert(p->vTtMem, uTruth);
+ pCutR->iFunc = Abc_Var2Lit( truthId, fCompl );
+ if ( p->pPars->fGenCnf && truthId == Vec_IntSize(&p->vCnfSizes) && LutSize <= 8 )
+ Vec_IntPush( &p->vCnfSizes, Abc_Tt8CnfSize(uTruth, pCutR->nLeaves) );
+ assert( (int)pCutR->nLeaves <= nOldSupp );
+ return (int)pCutR->nLeaves < nOldSupp;
+ }
+}
+static inline int Mf_CutComputeTruthMux6( Mf_Man_t * p, Mf_Cut_t * pCut0, Mf_Cut_t * pCut1, Mf_Cut_t * pCutC, int fCompl0, int fCompl1, int fComplC, Mf_Cut_t * pCutR )
+{
+ int nOldSupp = pCutR->nLeaves, nCubes = 0, truthId, fCompl; word t;
+ word t0 = *Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut0->iFunc));
+ word t1 = *Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut1->iFunc));
+ word tC = *Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCutC->iFunc));
+ if ( Abc_LitIsCompl(pCut0->iFunc) ^ fCompl0 ) t0 = ~t0;
+ if ( Abc_LitIsCompl(pCut1->iFunc) ^ fCompl1 ) t1 = ~t1;
+ if ( Abc_LitIsCompl(pCutC->iFunc) ^ fComplC ) tC = ~tC;
+ t0 = Abc_Tt6Expand( t0, pCut0->pLeaves, pCut0->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ t1 = Abc_Tt6Expand( t1, pCut1->pLeaves, pCut1->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ tC = Abc_Tt6Expand( tC, pCutC->pLeaves, pCutC->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ t = (tC & t1) | (~tC & t0);
+ if ( (fCompl = (int)(t & 1)) ) t = ~t;
+ pCutR->nLeaves = Abc_Tt6MinBase( &t, pCutR->pLeaves, pCutR->nLeaves );
+ assert( (int)(t & 1) == 0 );
+ truthId = Vec_MemHashInsert(p->vTtMem, &t);
+ pCutR->iFunc = Abc_Var2Lit( truthId, fCompl );
+ if ( p->pPars->fGenCnf && truthId == Vec_IntSize(&p->vCnfSizes) )
+ Vec_IntPush( &p->vCnfSizes, Abc_Tt6CnfSize(t, pCutR->nLeaves) );
+ assert( (int)pCutR->nLeaves <= nOldSupp );
+ return (int)pCutR->nLeaves < nOldSupp;
+}
+static inline int Mf_CutComputeTruthMux( Mf_Man_t * p, Mf_Cut_t * pCut0, Mf_Cut_t * pCut1, Mf_Cut_t * pCutC, int fCompl0, int fCompl1, int fComplC, Mf_Cut_t * pCutR )
+{
+ if ( p->pPars->nLutSize <= 6 )
+ return Mf_CutComputeTruthMux6( p, pCut0, pCut1, pCutC, fCompl0, fCompl1, fComplC, pCutR );
+ {
+ word uTruth[MF_TT_WORDS], uTruth0[MF_TT_WORDS], uTruth1[MF_TT_WORDS], uTruthC[MF_TT_WORDS];
+ int nOldSupp = pCutR->nLeaves, truthId;
+ int LutSize = p->pPars->nLutSize, fCompl;
+ int nWords = Abc_Truth6WordNum(LutSize);
+ word * pTruth0 = Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut0->iFunc));
+ word * pTruth1 = Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut1->iFunc));
+ word * pTruthC = Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCutC->iFunc));
+ Abc_TtCopy( uTruth0, pTruth0, nWords, Abc_LitIsCompl(pCut0->iFunc) ^ fCompl0 );
+ Abc_TtCopy( uTruth1, pTruth1, nWords, Abc_LitIsCompl(pCut1->iFunc) ^ fCompl1 );
+ Abc_TtCopy( uTruthC, pTruthC, nWords, Abc_LitIsCompl(pCutC->iFunc) ^ fComplC );
+ Abc_TtExpand( uTruth0, LutSize, pCut0->pLeaves, pCut0->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ Abc_TtExpand( uTruth1, LutSize, pCut1->pLeaves, pCut1->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ Abc_TtExpand( uTruthC, LutSize, pCutC->pLeaves, pCutC->nLeaves, pCutR->pLeaves, pCutR->nLeaves );
+ Abc_TtMux( uTruth, uTruthC, uTruth1, uTruth0, nWords );
+ fCompl = (int)(uTruth[0] & 1);
+ if ( fCompl ) Abc_TtNot( uTruth, nWords );
+ pCutR->nLeaves = Abc_TtMinBase( uTruth, pCutR->pLeaves, pCutR->nLeaves, LutSize );
+ assert( (uTruth[0] & 1) == 0 );
+ truthId = Vec_MemHashInsert(p->vTtMem, uTruth);
+ pCutR->iFunc = Abc_Var2Lit( truthId, fCompl );
+ if ( p->pPars->fGenCnf && truthId == Vec_IntSize(&p->vCnfSizes) && LutSize <= 8 )
+ Vec_IntPush( &p->vCnfSizes, Abc_Tt8CnfSize(uTruth, pCutR->nLeaves) );
+ assert( (int)pCutR->nLeaves <= nOldSupp );
+ return (int)pCutR->nLeaves < nOldSupp;
+ }
+}
+
+
+/**Function*************************************************************
+
+ Synopsis []
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+static inline int Mf_CutCountBits( word i )
+{
+ i = i - ((i >> 1) & 0x5555555555555555);
+ i = (i & 0x3333333333333333) + ((i >> 2) & 0x3333333333333333);
+ i = ((i + (i >> 4)) & 0x0F0F0F0F0F0F0F0F);
+ return (i*(0x0101010101010101))>>56;
+}
+static inline word Mf_CutGetSign( int * pLeaves, int nLeaves )
+{
+ word Sign = 0; int i;
+ for ( i = 0; i < nLeaves; i++ )
+ Sign |= ((word)1) << (pLeaves[i] & 0x3F);
+ return Sign;
+}
+static inline int Mf_CutCreateUnit( Mf_Cut_t * p, int i )
+{
+ p->Delay = 0;
+ p->Flow = 0;
+ p->iFunc = 2;
+ p->nLeaves = 1;
+ p->pLeaves[0] = i;
+ p->Sign = ((word)1) << (i & 0x3F);
+ return 1;
+}
+static inline void Mf_CutPrint( Mf_Man_t * p, Mf_Cut_t * pCut )
+{
+ int i, nDigits = Abc_Base10Log(Gia_ManObjNum(p->pGia));
+ printf( "%d {", pCut->nLeaves );
+ for ( i = 0; i < (int)pCut->nLeaves; i++ )
+ printf( " %*d", nDigits, pCut->pLeaves[i] );
+ for ( ; i < (int)p->pPars->nLutSize; i++ )
+ printf( " %*s", nDigits, " " );
+ printf( " } D = %4d A = %9.4f F = %6d ",
+ pCut->Delay, pCut->Flow, pCut->iFunc );
+ if ( p->vTtMem )
+ {
+ if ( p->pPars->fGenCnf )
+ printf( "CNF = %2d ", Vec_IntEntry(&p->vCnfSizes, Abc_Lit2Var(pCut->iFunc)) );
+ Dau_DsdPrintFromTruth( Vec_MemReadEntry(p->vTtMem, Abc_Lit2Var(pCut->iFunc)), pCut->nLeaves );
+ }
+ else
+ printf( "\n" );
+}
+static inline int Mf_ManPrepareCuts( Mf_Cut_t * pCuts, Mf_Man_t * p, int iObj )
+{
+ if ( Mf_ManObj(p, iObj)->iCutSet )
+ {
+ Mf_Cut_t * pMfCut = pCuts;
+ int i, * pCut, * pList = Mf_ObjCutSet(p, iObj);
+ Mf_SetForEachCut( pList, pCut, i )
+ {
+ pMfCut->Delay = 0;
+ pMfCut->Flow = 0;
+ pMfCut->iFunc = Mf_CutFunc( pCut );
+ pMfCut->nLeaves = Mf_CutSize( pCut );
+ pMfCut->Sign = Mf_CutGetSign( pCut+1, Mf_CutSize(pCut) );
+ memcpy( pMfCut->pLeaves, pCut+1, sizeof(int) * Mf_CutSize(pCut) );
+ pMfCut++;
+ }
+ if ( pCuts->nLeaves > 1 )
+ return pList[0] + Mf_CutCreateUnit( pMfCut, iObj );
+ return pList[0];
+ }
+ return Mf_CutCreateUnit( pCuts, iObj );
+}
+static inline int Mf_ManSaveCuts( Mf_Man_t * p, Mf_Cut_t ** pCuts, int nCuts )
+{
+ int i, * pPlace, iCur, nInts = 1;
+ for ( i = 0; i < nCuts; i++ )
+ nInts += pCuts[i]->nLeaves + 1;
+ if ( (p->iCur & 0xFFFF) + nInts > 0xFFFF )
+ p->iCur = ((p->iCur >> 16) + 1) << 16;
+ if ( Vec_PtrSize(&p->vPages) == (p->iCur >> 16) )
+ Vec_PtrPush( &p->vPages, ABC_ALLOC(int, (1<<16)) );
+ iCur = p->iCur; p->iCur += nInts;
+ pPlace = Mf_ManCutSet( p, iCur );
+ *pPlace++ = nCuts;
+ for ( i = 0; i < nCuts; i++ )
+ {
+ *pPlace++ = Mf_CutSetBoth(pCuts[i]->nLeaves, pCuts[i]->iFunc);
+ memcpy( pPlace, pCuts[i]->pLeaves, sizeof(int) * pCuts[i]->nLeaves );
+ pPlace += pCuts[i]->nLeaves;
+ }
+ return iCur;
+}
+static inline void Mf_ObjSetBestCut( int * pCuts, int * pCut )
+{
+ assert( pCuts < pCut );
+ if ( ++pCuts < pCut )
+ {
+ int pTemp[MF_CUT_MAX*(MF_LEAF_MAX+2)];
+ int nBlock = pCut - pCuts;
+ int nSize = Mf_CutSize(pCut) + 1;
+ memmove( pTemp, pCuts, sizeof(int) * nBlock );
+ memmove( pCuts, pCut, sizeof(int) * nSize );
+ memmove( pCuts + nSize, pTemp, sizeof(int) * nBlock );
+ }
+}
+
+/**Function*************************************************************
+
+ Synopsis [Check correctness of cuts.]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+static inline int Mf_CutCheck( Mf_Cut_t * pBase, Mf_Cut_t * pCut ) // check if pCut is contained in pBase
+{
+ int nSizeB = pBase->nLeaves;
+ int nSizeC = pCut->nLeaves;
+ int i, * pB = pBase->pLeaves;
+ int k, * pC = pCut->pLeaves;
+ for ( i = 0; i < nSizeC; i++ )
+ {
+ for ( k = 0; k < nSizeB; k++ )
+ if ( pC[i] == pB[k] )
+ break;
+ if ( k == nSizeB )
+ return 0;
+ }
+ return 1;
+}
+static inline int Mf_SetCheckArray( Mf_Cut_t ** ppCuts, int nCuts )
+{
+ Mf_Cut_t * pCut0, * pCut1;
+ int i, k, m, n, Value;
+ assert( nCuts > 0 );
+ for ( i = 0; i < nCuts; i++ )
+ {
+ pCut0 = ppCuts[i];
+ assert( pCut0->nLeaves <= MF_LEAF_MAX );
+ assert( pCut0->Sign == Mf_CutGetSign(pCut0->pLeaves, pCut0->nLeaves) );
+ // check duplicates
+ for ( m = 0; m < (int)pCut0->nLeaves; m++ )
+ for ( n = m + 1; n < (int)pCut0->nLeaves; n++ )
+ assert( pCut0->pLeaves[m] < pCut0->pLeaves[n] );
+ // check pairs
+ for ( k = 0; k < nCuts; k++ )
+ {
+ pCut1 = ppCuts[k];
+ if ( pCut0 == pCut1 )
+ continue;
+ // check containments
+ Value = Mf_CutCheck( pCut0, pCut1 );
+ assert( Value == 0 );
+ }
+ }
+ return 1;
+}
+
+
+/**Function*************************************************************
+
+ Synopsis []
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+static inline int Mf_CutMergeOrder( Mf_Cut_t * pCut0, Mf_Cut_t * pCut1, Mf_Cut_t * pCut, int nLutSize )
+{
+ int nSize0 = pCut0->nLeaves;
+ int nSize1 = pCut1->nLeaves;
+ int i, * pC0 = pCut0->pLeaves;
+ int k, * pC1 = pCut1->pLeaves;
+ int c, * pC = pCut->pLeaves;
+ // the case of the largest cut sizes
+ if ( nSize0 == nLutSize && nSize1 == nLutSize )
+ {
+ for ( i = 0; i < nSize0; i++ )
+ {
+ if ( pC0[i] != pC1[i] ) return 0;
+ pC[i] = pC0[i];
+ }
+ pCut->nLeaves = nLutSize;
+ pCut->iFunc = -1;
+ pCut->Sign = pCut0->Sign | pCut1->Sign;
+ return 1;
+ }
+ // compare two cuts with different numbers
+ i = k = c = 0;
+ if ( nSize0 == 0 ) goto FlushCut1;
+ if ( nSize1 == 0 ) goto FlushCut0;
+ while ( 1 )
+ {
+ if ( c == nLutSize ) return 0;
+ if ( pC0[i] < pC1[k] )
+ {
+ pC[c++] = pC0[i++];
+ if ( i >= nSize0 ) goto FlushCut1;
+ }
+ else if ( pC0[i] > pC1[k] )
+ {
+ pC[c++] = pC1[k++];
+ if ( k >= nSize1 ) goto FlushCut0;
+ }
+ else
+ {
+ pC[c++] = pC0[i++]; k++;
+ if ( i >= nSize0 ) goto FlushCut1;
+ if ( k >= nSize1 ) goto FlushCut0;
+ }
+ }
+
+FlushCut0:
+ if ( c + nSize0 > nLutSize + i ) return 0;
+ while ( i < nSize0 )
+ pC[c++] = pC0[i++];
+ pCut->nLeaves = c;
+ pCut->iFunc = -1;
+ pCut->Sign = pCut0->Sign | pCut1->Sign;
+ return 1;
+
+FlushCut1:
+ if ( c + nSize1 > nLutSize + k ) return 0;
+ while ( k < nSize1 )
+ pC[c++] = pC1[k++];
+ pCut->nLeaves = c;
+ pCut->iFunc = -1;
+ pCut->Sign = pCut0->Sign | pCut1->Sign;
+ return 1;
+}
+static inline int Mf_CutMergeOrderMux( Mf_Cut_t * pCut0, Mf_Cut_t * pCut1, Mf_Cut_t * pCut2, Mf_Cut_t * pCut, int nLutSize )
+{
+ int x0, i0 = 0, nSize0 = pCut0->nLeaves, * pC0 = pCut0->pLeaves;
+ int x1, i1 = 0, nSize1 = pCut1->nLeaves, * pC1 = pCut1->pLeaves;
+ int x2, i2 = 0, nSize2 = pCut2->nLeaves, * pC2 = pCut2->pLeaves;
+ int xMin, c = 0, * pC = pCut->pLeaves;
+ while ( 1 )
+ {
+ x0 = (i0 == nSize0) ? ABC_INFINITY : pC0[i0];
+ x1 = (i1 == nSize1) ? ABC_INFINITY : pC1[i1];
+ x2 = (i2 == nSize2) ? ABC_INFINITY : pC2[i2];
+ xMin = Abc_MinInt( Abc_MinInt(x0, x1), x2 );
+ if ( xMin == ABC_INFINITY ) break;
+ if ( c == nLutSize ) return 0;
+ pC[c++] = xMin;
+ if (x0 == xMin) i0++;
+ if (x1 == xMin) i1++;
+ if (x2 == xMin) i2++;
+ }
+ pCut->nLeaves = c;
+ pCut->iFunc = -1;
+ pCut->Sign = pCut0->Sign | pCut1->Sign | pCut2->Sign;
+ return 1;
+}
+static inline int Mf_SetCutIsContainedOrder( Mf_Cut_t * pBase, Mf_Cut_t * pCut ) // check if pCut is contained in pBase
+{
+ int i, nSizeB = pBase->nLeaves;
+ int k, nSizeC = pCut->nLeaves;
+ if ( nSizeB == nSizeC )
+ {
+ for ( i = 0; i < nSizeB; i++ )
+ if ( pBase->pLeaves[i] != pCut->pLeaves[i] )
+ return 0;
+ return 1;
+ }
+ assert( nSizeB > nSizeC );
+ if ( nSizeC == 0 )
+ return 1;
+ for ( i = k = 0; i < nSizeB; i++ )
+ {
+ if ( pBase->pLeaves[i] > pCut->pLeaves[k] )
+ return 0;
+ if ( pBase->pLeaves[i] == pCut->pLeaves[k] )
+ {
+ if ( ++k == nSizeC )
+ return 1;
+ }
+ }
+ return 0;
+}
+static inline int Mf_SetLastCutIsContained( Mf_Cut_t ** pCuts, int nCuts )
+{
+ int i;
+ for ( i = 0; i < nCuts; i++ )
+ if ( pCuts[i]->nLeaves <= pCuts[nCuts]->nLeaves && (pCuts[i]->Sign & pCuts[nCuts]->Sign) == pCuts[i]->Sign && Mf_SetCutIsContainedOrder(pCuts[nCuts], pCuts[i]) )
+ return 1;
+ return 0;
+}
+static inline int Mf_SetLastCutContainsArea( Mf_Cut_t ** pCuts, int nCuts )
+{
+ int i, k, fChanges = 0;
+ for ( i = 0; i < nCuts; i++ )
+ if ( pCuts[nCuts]->nLeaves < pCuts[i]->nLeaves && (pCuts[nCuts]->Sign & pCuts[i]->Sign) == pCuts[nCuts]->Sign && Mf_SetCutIsContainedOrder(pCuts[i], pCuts[nCuts]) )
+ pCuts[i]->nLeaves = MF_NO_LEAF, fChanges = 1;
+ if ( !fChanges )
+ return nCuts;
+ for ( i = k = 0; i <= nCuts; i++ )
+ {
+ if ( pCuts[i]->nLeaves == MF_NO_LEAF )
+ continue;
+ if ( k < i )
+ ABC_SWAP( Mf_Cut_t *, pCuts[k], pCuts[i] );
+ k++;
+ }
+ return k - 1;
+}
+static inline int Mf_CutCompareArea( Mf_Cut_t * pCut0, Mf_Cut_t * pCut1 )
+{
+ if ( pCut0->Flow < pCut1->Flow ) return -1;
+ if ( pCut0->Flow > pCut1->Flow ) return 1;
+ if ( pCut0->Delay < pCut1->Delay ) return -1;
+ if ( pCut0->Delay > pCut1->Delay ) return 1;
+ if ( pCut0->nLeaves < pCut1->nLeaves ) return -1;
+ if ( pCut0->nLeaves > pCut1->nLeaves ) return 1;
+ return 0;
+}
+static inline void Mf_SetSortByArea( Mf_Cut_t ** pCuts, int nCuts )
+{
+ int i;
+ for ( i = nCuts; i > 0; i-- )
+ {
+ if ( Mf_CutCompareArea(pCuts[i - 1], pCuts[i]) < 0 )//!= 1 )
+ return;
+ ABC_SWAP( Mf_Cut_t *, pCuts[i - 1], pCuts[i] );
+ }
+}
+static inline int Mf_SetAddCut( Mf_Cut_t ** pCuts, int nCuts, int nCutNum )
+{
+ if ( nCuts == 0 )
+ return 1;
+ nCuts = Mf_SetLastCutContainsArea(pCuts, nCuts);
+ Mf_SetSortByArea( pCuts, nCuts );
+ return Abc_MinInt( nCuts + 1, nCutNum - 1 );
+}
+static inline int Mf_CutArea( Mf_Man_t * p, int nLeaves, int iFunc )
+{
+ if ( nLeaves < 2 )
+ return 0;
+ if ( p->pPars->fGenCnf )
+ return Vec_IntEntry(&p->vCnfSizes, Abc_Lit2Var(iFunc));
+ if ( p->pPars->fOptEdge )
+ return nLeaves + 1;
+ return 1;
+}
+static inline void Mf_CutParams( Mf_Man_t * p, Mf_Cut_t * pCut, float FlowRefs )
+{
+ Mf_Obj_t * pBest;
+ int i, nLeaves = pCut->nLeaves;
+ assert( nLeaves <= p->pPars->nLutSize );
+ pCut->Delay = 0;
+ pCut->Flow = 0;
+ for ( i = 0; i < nLeaves; i++ )
+ {
+ pBest = Mf_ManObj(p, pCut->pLeaves[i]);
+ pCut->Delay = Abc_MaxInt( pCut->Delay, pBest->Delay );
+ pCut->Flow += pBest->Flow;
+ }
+ pCut->Delay += (int)(nLeaves > 1);
+ pCut->Flow = (pCut->Flow + Mf_CutArea(p, nLeaves, pCut->iFunc)) / FlowRefs;
+}
+void Mf_ObjMergeOrder( Mf_Man_t * p, int iObj )
+{
+ Mf_Cut_t pCuts0[MF_CUT_MAX], pCuts1[MF_CUT_MAX], pCuts[MF_CUT_MAX], * pCutsR[MF_CUT_MAX];
+ Gia_Obj_t * pObj = Gia_ManObj(p->pGia, iObj);
+ Mf_Obj_t * pBest = Mf_ManObj(p, iObj);
+ int nLutSize = p->pPars->nLutSize;
+ int nCutNum = p->pPars->nCutNum;
+ int nCuts0 = Mf_ManPrepareCuts(pCuts0, p, Gia_ObjFaninId0(pObj, iObj));
+ int nCuts1 = Mf_ManPrepareCuts(pCuts1, p, Gia_ObjFaninId1(pObj, iObj));
+ int fComp0 = Gia_ObjFaninC0(pObj);
+ int fComp1 = Gia_ObjFaninC1(pObj);
+ Mf_Cut_t * pCut0, * pCut1, * pCut0Lim = pCuts0 + nCuts0, * pCut1Lim = pCuts1 + nCuts1;
+ int i, nCutsR = 0;
+ for ( i = 0; i < nCutNum; i++ )
+ pCutsR[i] = pCuts + i;
+ if ( Gia_ObjIsMuxId(p->pGia, iObj) )
+ {
+ Mf_Cut_t pCuts2[MF_CUT_MAX];
+ int nCuts2 = Mf_ManPrepareCuts(pCuts2, p, Gia_ObjFaninId2(p->pGia, iObj));
+ int fComp2 = Gia_ObjFaninC2(p->pGia, pObj);
+ Mf_Cut_t * pCut2, * pCut2Lim = pCuts2 + nCuts2;
+ p->CutCount[0] += nCuts0 * nCuts1 * nCuts2;
+ for ( pCut0 = pCuts0; pCut0 < pCut0Lim; pCut0++ )
+ for ( pCut1 = pCuts1; pCut1 < pCut1Lim; pCut1++ )
+ for ( pCut2 = pCuts2; pCut2 < pCut2Lim; pCut2++ )
+ {
+ if ( Mf_CutCountBits(pCut0->Sign | pCut1->Sign | pCut2->Sign) > nLutSize )
+ continue;
+ p->CutCount[1]++;
+ if ( !Mf_CutMergeOrderMux(pCut0, pCut1, pCut2, pCutsR[nCutsR], nLutSize) )
+ continue;
+ if ( Mf_SetLastCutIsContained(pCutsR, nCutsR) )
+ continue;
+ p->CutCount[2]++;
+ if ( p->pPars->fCutMin && Mf_CutComputeTruthMux(p, pCut0, pCut1, pCut2, fComp0, fComp1, fComp2, pCutsR[nCutsR]) )
+ pCutsR[nCutsR]->Sign = Mf_CutGetSign(pCutsR[nCutsR]->pLeaves, pCutsR[nCutsR]->nLeaves);
+ Mf_CutParams( p, pCutsR[nCutsR], pBest->nFlowRefs );
+ nCutsR = Mf_SetAddCut( pCutsR, nCutsR, nCutNum );
+ }
+ }
+ else
+ {
+ int fIsXor = Gia_ObjIsXor(pObj);
+ p->CutCount[0] += nCuts0 * nCuts1;
+ for ( pCut0 = pCuts0; pCut0 < pCut0Lim; pCut0++ )
+ for ( pCut1 = pCuts1; pCut1 < pCut1Lim; pCut1++ )
+ {
+ if ( (int)(pCut0->nLeaves + pCut1->nLeaves) > nLutSize && Mf_CutCountBits(pCut0->Sign | pCut1->Sign) > nLutSize )
+ continue;
+ p->CutCount[1]++;
+ if ( !Mf_CutMergeOrder(pCut0, pCut1, pCutsR[nCutsR], nLutSize) )
+ continue;
+ if ( Mf_SetLastCutIsContained(pCutsR, nCutsR) )
+ continue;
+ p->CutCount[2]++;
+ if ( p->pPars->fCutMin && Mf_CutComputeTruth(p, pCut0, pCut1, fComp0, fComp1, pCutsR[nCutsR], fIsXor) )
+ pCutsR[nCutsR]->Sign = Mf_CutGetSign(pCutsR[nCutsR]->pLeaves, pCutsR[nCutsR]->nLeaves);
+ Mf_CutParams( p, pCutsR[nCutsR], pBest->nFlowRefs );
+ nCutsR = Mf_SetAddCut( pCutsR, nCutsR, nCutNum );
+ }
+ }
+ // debug printout
+ if ( 0 )
+// if ( iObj % 1000 == 0 )
+// if ( iObj == 474 )
+ {
+ printf( "*** Obj = %d FlowRefs = %.2f MapRefs = %2d\n", iObj, pBest->nFlowRefs, pBest->nMapRefs );
+ for ( i = 0; i < nCutsR; i++ )
+ Mf_CutPrint( p, pCutsR[i] );
+ printf( "\n" );
+ }
+ // store the cutset
+ pBest->Flow = pCutsR[0]->Flow;
+ pBest->Delay = pCutsR[0]->Delay;
+ pBest->iCutSet = Mf_ManSaveCuts( p, pCutsR, nCutsR );
+ // verify
+ assert( nCutsR > 0 && nCutsR < nCutNum );
+ assert( Mf_SetCheckArray(pCutsR, nCutsR) );
+ p->nCutCounts[pCutsR[0]->nLeaves]++;
+ p->CutCount[3] += nCutsR;
+}
+
+
+/**Function*************************************************************
+
+ Synopsis []
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+void Mf_ManSetFlowRefs( Gia_Man_t * p, Vec_Int_t * vRefs )
+{
+ int fDiscount = 1;
+ Gia_Obj_t * pObj, * pCtrl, * pData0, * pData1;
+ int i, Id;
+ Vec_IntFill( vRefs, Gia_ManObjNum(p), 0 );
+ Gia_ManForEachAnd( p, pObj, i )
+ {
+ if ( Gia_ObjIsAnd(Gia_ObjFanin0(pObj)) )
+ Vec_IntAddToEntry( vRefs, Gia_ObjFaninId0(pObj, i), 1 );
+ if ( Gia_ObjIsAnd(Gia_ObjFanin1(pObj)) )
+ Vec_IntAddToEntry( vRefs, Gia_ObjFaninId1(pObj, i), 1 );
+ if ( p->pMuxes )
+ {
+ if ( Gia_ObjIsMuxId(p, i) && Gia_ObjIsAnd(Gia_ObjFanin2(p, pObj)) )
+ Vec_IntAddToEntry( vRefs, Gia_ObjFaninId2(p, i), 1 );
+ }
+ else if ( fDiscount && Gia_ObjIsMuxType(pObj) ) // discount XOR/MUX
+ {
+ pCtrl = Gia_Regular(Gia_ObjRecognizeMux(pObj, &pData1, &pData0));
+ pData0 = Gia_Regular(pData0);
+ pData1 = Gia_Regular(pData1);
+ if ( Gia_ObjIsAnd(pCtrl) )
+ Vec_IntAddToEntry( vRefs, Gia_ObjId(p, pCtrl), -1 );
+ if ( pData0 == pData1 && Gia_ObjIsAnd(pData0) )
+ Vec_IntAddToEntry( vRefs, Gia_ObjId(p, pData0), -1 );
+ }
+ }
+ Gia_ManForEachCoDriverId( p, Id, i )
+ if ( Gia_ObjIsAnd(Gia_ManObj(p, Id)) )
+ Vec_IntAddToEntry( vRefs, Id, 1 );
+ for ( i = 0; i < Vec_IntSize(vRefs); i++ )
+ Vec_IntUpdateEntry( vRefs, i, 1 );
+}
+int Mf_ManSetMapRefs( Mf_Man_t * p )
+{
+ float Coef = 1.0 / (1.0 + (p->Iter + 1) * (p->Iter + 1));
+ int * pCut, i, k, Id;
+ // compute delay
+ int Delay = 0;
+ Gia_ManForEachCoDriverId( p->pGia, Id, i )
+ Delay = Abc_MaxInt( Delay, Mf_ManObj(p, Id)->Delay );
+ // check delay target
+ if ( p->pPars->DelayTarget == -1 && p->pPars->nRelaxRatio )
+ p->pPars->DelayTarget = (int)((float)Delay * (100.0 + p->pPars->nRelaxRatio) / 100.0);
+ if ( p->pPars->DelayTarget != -1 )
+ {
+ if ( Delay < p->pPars->DelayTarget + 0.01 )
+ Delay = p->pPars->DelayTarget;
+ else if ( p->pPars->nRelaxRatio == 0 )
+ Abc_Print( 0, "Relaxing user-specified delay target from %d to %d.\n", p->pPars->DelayTarget, Delay );
+ }
+ p->pPars->Delay = Delay;
+ // check references
+// Gia_ManForEachAndId( p->pGia, i )
+// assert( Mf_ManObj(p, i)->nMapRefs == 0 );
+ // compute area and edges
+ if ( !p->fUseEla )
+ Gia_ManForEachCoDriverId( p->pGia, Id, i )
+ Mf_ObjMapRefInc( p, Id );
+ p->pPars->Area = p->pPars->Edge = p->pPars->Clause = 0;
+ Gia_ManForEachAndReverseId( p->pGia, i )
+ {
+ if ( !Mf_ObjMapRefNum(p, i) )
+ continue;
+ pCut = Mf_ObjCutBest( p, i );
+ if ( !p->fUseEla )
+ for ( k = 1; k <= Mf_CutSize(pCut); k++ )
+ Mf_ObjMapRefInc( p, pCut[k] );
+ p->pPars->Edge += Mf_CutSize(pCut);
+ p->pPars->Area++;
+ if ( p->pPars->fGenCnf )
+ p->pPars->Clause += Mf_CutArea(p, Mf_CutSize(pCut), Mf_CutFunc(pCut));
+ }
+ // blend references
+ for ( i = 0; i < Gia_ManObjNum(p->pGia); i++ )
+ p->pLfObjs[i].nFlowRefs = Coef * p->pLfObjs[i].nFlowRefs + (1.0 - Coef) * Abc_MaxFloat(1, p->pLfObjs[i].nMapRefs);
+// p->pLfObjs[i]. = 0.2 * p->pLfObjs[i]. + 0.8 * Abc_MaxFloat(1, p->pLfObjs[i].nMapRefs);
+ return p->pPars->Area;
+}
+
+/**Function*************************************************************
+
+ Synopsis []
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+Gia_Man_t * Mf_ManDeriveMapping( Mf_Man_t * p )
+{
+ Vec_Int_t * vMapping;
+ int i, k, * pCut;
+ assert( !p->pPars->fCutMin && p->pGia->vMapping == NULL );
+ vMapping = Vec_IntAlloc( Gia_ManObjNum(p->pGia) + (int)p->pPars->Edge + (int)p->pPars->Area * 2 );
+ Vec_IntFill( vMapping, Gia_ManObjNum(p->pGia), 0 );
+ Gia_ManForEachAndId( p->pGia, i )
+ {
+ if ( !Mf_ObjMapRefNum(p, i) )
+ continue;
+ pCut = Mf_ObjCutBest( p, i );
+ Vec_IntWriteEntry( vMapping, i, Vec_IntSize(vMapping) );
+ Vec_IntPush( vMapping, Mf_CutSize(pCut) );
+ for ( k = 1; k <= Mf_CutSize(pCut); k++ )
+ Vec_IntPush( vMapping, pCut[k] );
+ Vec_IntPush( vMapping, i );
+ }
+ assert( Vec_IntCap(vMapping) == 16 || Vec_IntSize(vMapping) == Vec_IntCap(vMapping) );
+ p->pGia->vMapping = vMapping;
+ return p->pGia;
+}
+Gia_Man_t * Mf_ManDeriveMappingCoarse( Mf_Man_t * p )
+{
+ Gia_Man_t * pNew, * pGia = p->pGia;
+ Gia_Obj_t * pObj;
+ int i, k, * pCut;
+ assert( !p->pPars->fCutMin && pGia->pMuxes );
+ // create new manager
+ pNew = Gia_ManStart( Gia_ManObjNum(pGia) );
+ pNew->pName = Abc_UtilStrsav( pGia->pName );
+ pNew->pSpec = Abc_UtilStrsav( pGia->pSpec );
+ // map primary inputs
+ Gia_ManConst0(pGia)->Value = 0;
+ Gia_ManForEachCi( pGia, pObj, i )
+ pObj->Value = Gia_ManAppendCi( pNew );
+ // start mapping
+ pNew->vMapping = Vec_IntAlloc( Gia_ManObjNum(pGia) + 2*Gia_ManXorNum(pGia) + 2*Gia_ManMuxNum(pGia) + (int)p->pPars->Edge + (int)p->pPars->Area * 2 );
+ Vec_IntFill( pNew->vMapping, Gia_ManObjNum(pGia) + 2*Gia_ManXorNum(pGia) + 2*Gia_ManMuxNum(pGia), 0 );
+ // iterate through nodes used in the mapping
+ Gia_ManForEachAnd( pGia, pObj, i )
+ {
+ if ( Gia_ObjIsMuxId(pGia, i) )
+ pObj->Value = Gia_ManAppendMux( pNew, Gia_ObjFanin2Copy(pGia, pObj), Gia_ObjFanin1Copy(pObj), Gia_ObjFanin0Copy(pObj) );
+ else if ( Gia_ObjIsXor(pObj) )
+ pObj->Value = Gia_ManAppendXor( pNew, Gia_ObjFanin0Copy(pObj), Gia_ObjFanin1Copy(pObj) );
+ else
+ pObj->Value = Gia_ManAppendAnd( pNew, Gia_ObjFanin0Copy(pObj), Gia_ObjFanin1Copy(pObj) );
+ if ( !Mf_ObjMapRefNum(p, i) )
+ continue;
+ pCut = Mf_ObjCutBest( p, i );
+ Vec_IntWriteEntry( pNew->vMapping, Abc_Lit2Var(pObj->Value), Vec_IntSize(pNew->vMapping) );
+ Vec_IntPush( pNew->vMapping, Mf_CutSize(pCut));
+ for ( k = 1; k <= Mf_CutSize(pCut); k++ )
+ Vec_IntPush( pNew->vMapping, Abc_Lit2Var(Gia_ManObj(pGia, pCut[k])->Value) );
+ Vec_IntPush( pNew->vMapping, Abc_Lit2Var(pObj->Value) );
+ }
+ Gia_ManForEachCo( pGia, pObj, i )
+ pObj->Value = Gia_ManAppendCo( pNew, Gia_ObjFanin0Copy(pObj) );
+ Gia_ManSetRegNum( pNew, Gia_ManRegNum(pGia) );
+ assert( Vec_IntCap(pNew->vMapping) == 16 || Vec_IntSize(pNew->vMapping) == Vec_IntCap(pNew->vMapping) );
+ return pNew;
+}
+Gia_Man_t * Mf_ManDeriveMappingGia( Mf_Man_t * p )
+{
+ Gia_Man_t * pNew;
+ Gia_Obj_t * pObj;
+ Vec_Int_t * vCopies = Vec_IntStartFull( Gia_ManObjNum(p->pGia) );
+ Vec_Int_t * vMapping = Vec_IntStart( 2 * Gia_ManObjNum(p->pGia) + (int)p->pPars->Edge + 2 * (int)p->pPars->Area );
+ Vec_Int_t * vMapping2 = Vec_IntStart( (int)p->pPars->Edge + 2 * (int)p->pPars->Area + 1000 );
+ Vec_Int_t * vCover = Vec_IntAlloc( 1 << 16 );
+ Vec_Int_t * vLeaves = Vec_IntAlloc( 16 );
+ int i, k, Id, iLit, * pCut;
+ word uTruth = 0, * pTruth = &uTruth;
+ assert( p->pPars->fCutMin );
+ // create new manager
+ pNew = Gia_ManStart( Gia_ManObjNum(p->pGia) );
+ pNew->pName = Abc_UtilStrsav( p->pGia->pName );
+ pNew->pSpec = Abc_UtilStrsav( p->pGia->pSpec );
+ // map primary inputs
+ Vec_IntWriteEntry( vCopies, 0, 0 );
+ Gia_ManForEachCiId( p->pGia, Id, i )
+ Vec_IntWriteEntry( vCopies, Id, Gia_ManAppendCi(pNew) );
+ // iterate through nodes used in the mapping
+ Gia_ManForEachAnd( p->pGia, pObj, i )
+ {
+ if ( !Mf_ObjMapRefNum(p, i) )
+ continue;
+ pCut = Mf_ObjCutBest( p, i );
+ if ( Mf_CutSize(pCut) == 0 )
+ {
+ assert( Abc_Lit2Var(Mf_CutFunc(pCut)) == 0 );
+ Vec_IntWriteEntry( vCopies, i, Mf_CutFunc(pCut) );
+ continue;
+ }
+ if ( Mf_CutSize(pCut) == 1 )
+ {
+ assert( Abc_Lit2Var(Mf_CutFunc(pCut)) == 1 );
+ iLit = Vec_IntEntry( vCopies, pCut[1] );
+ Vec_IntWriteEntry( vCopies, i, Abc_LitNotCond(iLit, Abc_LitIsCompl(Mf_CutFunc(pCut))) );
+ continue;
+ }
+ Vec_IntClear( vLeaves );
+ for ( k = 1; k <= Mf_CutSize(pCut); k++ )
+ Vec_IntPush( vLeaves, Vec_IntEntry(vCopies, pCut[k]) );
+ pTruth = Vec_MemReadEntry( p->vTtMem, Abc_Lit2Var(Mf_CutFunc(pCut)) );
+ iLit = Kit_TruthToGia( pNew, (unsigned *)pTruth, Vec_IntSize(vLeaves), vCover, vLeaves, 0 );
+ Vec_IntWriteEntry( vCopies, i, Abc_LitNotCond(iLit, Abc_LitIsCompl(Mf_CutFunc(pCut))) );
+ // create mapping
+ Vec_IntSetEntry( vMapping, Abc_Lit2Var(iLit), Vec_IntSize(vMapping2) );
+ Vec_IntPush( vMapping2, Vec_IntSize(vLeaves) );
+ Vec_IntForEachEntry( vLeaves, iLit, k )
+ Vec_IntPush( vMapping2, Abc_Lit2Var(iLit) );
+ Vec_IntPush( vMapping2, Abc_Lit2Var(Vec_IntEntry(vCopies, i)) );
+ }
+ Gia_ManForEachCo( p->pGia, pObj, i )
+ {
+ iLit = Vec_IntEntry( vCopies, Gia_ObjFaninId0p(p->pGia, pObj) );
+ iLit = Gia_ManAppendCo( pNew, Abc_LitNotCond(iLit, Gia_ObjFaninC0(pObj)) );
+ }
+ Vec_IntFree( vCopies );
+ Vec_IntFree( vCover );
+ Vec_IntFree( vLeaves );
+ // finish mapping
+ if ( Vec_IntSize(vMapping) > Gia_ManObjNum(pNew) )
+ Vec_IntShrink( vMapping, Gia_ManObjNum(pNew) );
+ else
+ Vec_IntFillExtra( vMapping, Gia_ManObjNum(pNew), 0 );
+ assert( Vec_IntSize(vMapping) == Gia_ManObjNum(pNew) );
+ Vec_IntForEachEntry( vMapping, iLit, i )
+ if ( iLit > 0 )
+ Vec_IntAddToEntry( vMapping, i, Gia_ManObjNum(pNew) );
+ Vec_IntAppend( vMapping, vMapping2 );
+ Vec_IntFree( vMapping2 );
+ // attach mapping and packing
+ assert( pNew->vMapping == NULL );
+ pNew->vMapping = vMapping;
+ Gia_ManSetRegNum( pNew, Gia_ManRegNum(p->pGia) );
+ return pNew;
+}
+
+/**Function*************************************************************
+
+ Synopsis []
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+Mf_Man_t * Mf_ManAlloc( Gia_Man_t * pGia, Jf_Par_t * pPars )
+{
+ Mf_Man_t * p;
+ Vec_Int_t * vFlowRefs;
+ int i, Entry;
+ assert( pPars->nCutNum > 1 && pPars->nCutNum <= MF_CUT_MAX );
+ assert( pPars->nLutSize > 1 && pPars->nLutSize <= MF_LEAF_MAX );
+ ABC_FREE( pGia->pRefs );
+ Vec_IntFreeP( &pGia->vMapping );
+ p = ABC_CALLOC( Mf_Man_t, 1 );
+ p->clkStart = Abc_Clock();
+ p->pGia = pGia;
+ p->pPars = pPars;
+ p->vTtMem = pPars->fCutMin ? Vec_MemAllocForTT( pPars->nLutSize, 0 ) : NULL;
+ p->pLfObjs = ABC_CALLOC( Mf_Obj_t, Gia_ManObjNum(pGia) );
+ p->iCur = 2;
+ Vec_PtrGrow( &p->vPages, 256 );
+ if ( pPars->fGenCnf )
+ {
+ Vec_IntGrow( &p->vCnfSizes, 10000 );
+ Vec_IntPush( &p->vCnfSizes, 1 );
+ Vec_IntPush( &p->vCnfSizes, 2 );
+ Vec_IntGrow( &p->vCnfMem, 10000 );
+ }
+ vFlowRefs = Vec_IntAlloc(0);
+ Mf_ManSetFlowRefs( pGia, vFlowRefs );
+ Vec_IntForEachEntry( vFlowRefs, Entry, i )
+ p->pLfObjs[i].nFlowRefs = Entry;
+ Vec_IntFree(vFlowRefs);
+ return p;
+}
+void Mf_ManFree( Mf_Man_t * p )
+{
+ assert( !p->pPars->fGenCnf || Vec_IntSize(&p->vCnfSizes) == Vec_MemEntryNum(p->vTtMem) );
+ if ( p->pPars->fCutMin )
+ Vec_MemHashFree( p->vTtMem );
+ if ( p->pPars->fCutMin )
+ Vec_MemFree( p->vTtMem );
+ Vec_PtrFreeData( &p->vPages );
+ ABC_FREE( p->vCnfSizes.pArray );
+ ABC_FREE( p->vCnfMem.pArray );
+ ABC_FREE( p->vPages.pArray );
+ ABC_FREE( p->pLfObjs );
+ ABC_FREE( p );
+}
/**Function*************************************************************
@@ -46,11 +1352,279 @@ ABC_NAMESPACE_IMPL_START
***********************************************************************/
void Mf_ManSetDefaultPars( Jf_Par_t * pPars )
{
- Jf_ManSetDefaultPars( pPars );
+ memset( pPars, 0, sizeof(Jf_Par_t) );
+ pPars->nLutSize = 6;
+ pPars->nCutNum = 8;
+ pPars->nProcNum = 0;
+ pPars->nRounds = 2;
+ pPars->nRoundsEla = 1;
+ pPars->nRelaxRatio = 0;
+ pPars->nCoarseLimit = 3;
+ pPars->nVerbLimit = 5;
+ pPars->DelayTarget = -1;
+ pPars->fAreaOnly = 0;
+ pPars->fOptEdge = 1;
+ pPars->fCoarsen = 1;
+ pPars->fCutMin = 0;
+ pPars->fGenCnf = 0;
+ pPars->fPureAig = 0;
+ pPars->fVerbose = 0;
+ pPars->fVeryVerbose = 0;
+ pPars->nLutSizeMax = MF_LEAF_MAX;
+ pPars->nCutNumMax = MF_CUT_MAX;
+}
+void Mf_ManPrintStats( Mf_Man_t * p, char * pTitle )
+{
+ if ( !p->pPars->fVerbose )
+ return;
+ printf( "%s : ", pTitle );
+ printf( "Level =%6lu ", p->pPars->Delay );
+ printf( "Area =%9lu ", p->pPars->Area );
+ printf( "Edge =%9lu ", p->pPars->Edge );
+ if ( p->pPars->fGenCnf )
+ printf( "CNF =%9lu ", p->pPars->Clause );
+ Abc_PrintTime( 1, "Time", Abc_Clock() - p->clkStart );
+ fflush( stdout );
+}
+void Mf_ManPrintInit( Mf_Man_t * p )
+{
+ if ( !p->pPars->fVerbose )
+ return;
+ printf( "LutSize = %d ", p->pPars->nLutSize );
+ printf( "CutNum = %d ", p->pPars->nCutNum );
+ printf( "Iter = %d ", p->pPars->nRounds + p->pPars->nRoundsEla );
+ printf( "Edge = %d ", p->pPars->fOptEdge );
+ printf( "CutMin = %d ", p->pPars->fCutMin );
+ printf( "Coarse = %d ", p->pPars->fCoarsen );
+ printf( "CNF = %d ", p->pPars->fGenCnf );
+ printf( "\n" );
+ printf( "Computing cuts...\r" );
+ fflush( stdout );
+}
+void Mf_ManPrintQuit( Mf_Man_t * p, Gia_Man_t * pNew )
+{
+ float MemGia = Gia_ManMemory(p->pGia) / (1<<20);
+ float MemMan = 1.0 * sizeof(Mf_Obj_t) * Gia_ManObjNum(p->pGia) / (1<<20);
+ float MemCuts = 1.0 * sizeof(int) * (1 << 16) * Vec_PtrSize(&p->vPages) / (1<<20);
+ float MemTt = p->vTtMem ? Vec_MemMemory(p->vTtMem) / (1<<20) : 0;
+ float MemMap = Vec_IntMemory(pNew->vMapping) / (1<<20);
+ if ( p->CutCount[0] == 0 )
+ p->CutCount[0] = 1;
+ if ( !p->pPars->fVerbose )
+ return;
+ printf( "CutPair = %.0f ", p->CutCount[0] );
+ printf( "Merge = %.0f (%.2f %%) ", p->CutCount[1], 100.0*p->CutCount[1]/p->CutCount[0] );
+ printf( "Eval = %.0f (%.2f %%) ", p->CutCount[2], 100.0*p->CutCount[2]/p->CutCount[0] );
+ printf( "Cut = %.0f (%.2f %%) ", p->CutCount[3], 100.0*p->CutCount[3]/p->CutCount[0] );
+ printf( "\n" );
+ printf( "Gia = %.2f MB ", MemGia );
+ printf( "Man = %.2f MB ", MemMan );
+ printf( "Cut = %.2f MB ", MemCuts );
+ printf( "Map = %.2f MB ", MemMap );
+ printf( "TT = %.2f MB ", MemTt );
+ printf( "Total = %.2f MB", MemGia + MemMan + MemCuts + MemMap + MemTt );
+ printf( "\n" );
+ if ( 1 )
+ {
+ int i;
+ for ( i = 0; i <= p->pPars->nLutSize; i++ )
+ printf( "%d = %d ", i, p->nCutCounts[i] );
+ if ( p->vTtMem )
+ printf( "TT = %d (%.2f %%) ", Vec_MemEntryNum(p->vTtMem), 100.0 * Vec_MemEntryNum(p->vTtMem) / p->CutCount[2] );
+ Abc_PrintTime( 1, "Time", Abc_Clock() - p->clkStart );
+ }
+ fflush( stdout );
+}
+void Mf_ManComputeCuts( Mf_Man_t * p )
+{
+ int i;
+ Gia_ManForEachAndId( p->pGia, i )
+ Mf_ObjMergeOrder( p, i );
+ Mf_ManSetMapRefs( p );
+ Mf_ManPrintStats( p, p->fUseEla ? "Ela " : (p->Iter ? "Area " : "Delay") );
+}
+
+/**Function*************************************************************
+
+ Synopsis [Flow and area.]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+int Mf_CutRef_rec( Mf_Man_t * p, int * pCut )
+{
+ int i, Count = Mf_CutArea(p, Mf_CutSize(pCut), Mf_CutFunc(pCut));
+ for ( i = 1; i <= Mf_CutSize(pCut); i++ )
+ if ( !Mf_ObjMapRefInc(p, pCut[i]) && Mf_ManObj(p, pCut[i])->iCutSet )
+ Count += Mf_CutRef_rec( p, Mf_ObjCutBest(p, pCut[i]) );
+ return Count;
+}
+int Mf_CutDeref_rec( Mf_Man_t * p, int * pCut )
+{
+ int i, Count = Mf_CutArea(p, Mf_CutSize(pCut), Mf_CutFunc(pCut));
+ for ( i = 1; i <= Mf_CutSize(pCut); i++ )
+ if ( !Mf_ObjMapRefDec(p, pCut[i]) && Mf_ManObj(p, pCut[i])->iCutSet )
+ Count += Mf_CutDeref_rec( p, Mf_ObjCutBest(p, pCut[i]) );
+ return Count;
+}
+static inline int Mf_CutAreaDerefed( Mf_Man_t * p, int * pCut )
+{
+ int Ela1 = Mf_CutRef_rec( p, pCut );
+ int Ela2 = Mf_CutDeref_rec( p, pCut );
+ assert( Ela1 == Ela2 );
+ return Ela1;
+}
+static inline float Mf_CutFlow( Mf_Man_t * p, int * pCut, int * pTime )
+{
+ Mf_Obj_t * pObj;
+ float Flow = 0;
+ int i, Time = 0;
+ for ( i = 1; i <= Mf_CutSize(pCut); i++ )
+ {
+ pObj = Mf_ManObj( p, pCut[i] );
+ Time = Abc_MaxInt( Time, pObj->Delay );
+ Flow += pObj->Flow;
+ }
+ *pTime = Time + 1;
+ return Flow + Mf_CutArea(p, Mf_CutSize(pCut), Mf_CutFunc(pCut));
+}
+static inline void Mf_ObjComputeBestCut( Mf_Man_t * p, int iObj )
+{
+ Mf_Obj_t * pBest = Mf_ManObj(p, iObj);
+ int * pCutSet = Mf_ObjCutSet( p, iObj );
+ int * pCut, * pCutBest = NULL;
+ int Value1 = -1, Value2 = -1;
+ int i, Time = 0, TimeBest = ABC_INFINITY;
+ float Flow, FlowBest = ABC_INFINITY;
+ if ( p->fUseEla && pBest->nMapRefs )
+ Value1 = Mf_CutDeref_rec( p, Mf_ObjCutBest(p, iObj) );
+ Mf_SetForEachCut( pCutSet, pCut, i )
+ {
+ assert( !Mf_CutIsTriv(pCut, iObj) );
+ assert( Mf_CutSize(pCut) <= p->pPars->nLutSize );
+ Flow = p->fUseEla ? Mf_CutAreaDerefed(p, pCut) : Mf_CutFlow(p, pCut, &Time);
+ if ( pCutBest == NULL || FlowBest > Flow || (FlowBest == Flow && TimeBest > Time) )
+ pCutBest = pCut, FlowBest = Flow, TimeBest = Time;
+ }
+ assert( pCutBest != NULL );
+ if ( p->fUseEla && pBest->nMapRefs )
+ Value1 = Mf_CutRef_rec( p, pCutBest );
+ else
+ pBest->nMapRefs = 0;
+ assert( Value1 >= Value2 );
+ if ( p->fUseEla )
+ Mf_CutFlow( p, pCutBest, &TimeBest );
+ pBest->Delay = TimeBest;
+ pBest->Flow = FlowBest / Mf_ManObj(p, iObj)->nFlowRefs;
+ Mf_ObjSetBestCut( pCutSet, pCutBest );
+// Mf_CutPrint( Mf_ObjCutBest(p, iObj) ); printf( "\n" );
+}
+
+
+/**Function*************************************************************
+
+ Synopsis [Technology mappping.]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+void Mf_ManComputeMapping( Mf_Man_t * p )
+{
+ int i;
+ Gia_ManForEachAndId( p->pGia, i )
+ Mf_ObjComputeBestCut( p, i );
+ Mf_ManSetMapRefs( p );
+ Mf_ManPrintStats( p, p->fUseEla ? "Ela " : (p->Iter ? "Area " : "Delay") );
}
Gia_Man_t * Mf_ManPerformMapping( Gia_Man_t * pGia, Jf_Par_t * pPars )
{
- return Jf_ManPerformMapping( pGia, pPars );
+ Mf_Man_t * p;
+ Gia_Man_t * pNew, * pCls;
+ if ( pPars->fGenCnf )
+ pPars->fCutMin = 1;
+ pCls = pPars->fCoarsen ? Gia_ManDupMuxes(pGia, pPars->nCoarseLimit) : pGia;
+ p = Mf_ManAlloc( pCls, pPars );
+ p->pGia0 = pGia;
+ if ( pPars->fVerbose && pPars->fCoarsen )
+ {
+ printf( "Initial " ); Gia_ManPrintMuxStats( pGia ); printf( "\n" );
+ printf( "Derived " ); Gia_ManPrintMuxStats( pCls ); printf( "\n" );
+ }
+ Mf_ManPrintInit( p );
+ Mf_ManComputeCuts( p );
+ for ( p->Iter = 1; p->Iter < p->pPars->nRounds; p->Iter++ )
+ Mf_ManComputeMapping( p );
+ p->fUseEla = 1;
+ for ( ; p->Iter < p->pPars->nRounds + pPars->nRoundsEla; p->Iter++ )
+ Mf_ManComputeMapping( p );
+ if ( pPars->fVeryVerbose && pPars->fCutMin )
+ Vec_MemDumpTruthTables( p->vTtMem, Gia_ManName(p->pGia), pPars->nLutSize );
+ if ( pPars->fCutMin )
+ pNew = Mf_ManDeriveMappingGia( p );
+ else if ( pPars->fCoarsen )
+ pNew = Mf_ManDeriveMappingCoarse( p );
+ else
+ pNew = Mf_ManDeriveMapping( p );
+ if ( p->pPars->fGenCnf )
+ pGia->pData = Mf_ManDeriveCnf( p, p->pPars->fCnfObjIds, p->pPars->fAddOrCla );
+// if ( p->pPars->fGenCnf )
+// Mf_ManProfileTruths( p );
+ Gia_ManMappingVerify( pNew );
+ Mf_ManPrintQuit( p, pNew );
+ Mf_ManFree( p );
+ if ( pCls != pGia )
+ Gia_ManStop( pCls );
+ return pNew;
+}
+
+/**Function*************************************************************
+
+ Synopsis [CNF generation]
+
+ Description []
+
+ SideEffects []
+
+ SeeAlso []
+
+***********************************************************************/
+Cnf_Dat_t * Mf_ManGenerateCnf( Gia_Man_t * pGia, int nLutSize, int fCnfObjIds, int fAddOrCla, int fVerbose )
+{
+ Gia_Man_t * pNew;
+ Jf_Par_t Pars, * pPars = &Pars;
+ assert( nLutSize >= 3 && nLutSize <= 8 );
+ Mf_ManSetDefaultPars( pPars );
+ pPars->fGenCnf = 1;
+ pPars->fCoarsen = !fCnfObjIds;
+ pPars->nLutSize = nLutSize;
+ pPars->fCnfObjIds = fCnfObjIds;
+ pPars->fAddOrCla = fAddOrCla;
+ pPars->fVerbose = fVerbose;
+ pNew = Mf_ManPerformMapping( pGia, pPars );
+ Gia_ManStopP( &pNew );
+// Cnf_DataPrint( (Cnf_Dat_t *)pGia->pData, 1 );
+ return pGia->pData;
+}
+void Mf_ManDumpCnf( Gia_Man_t * p, char * pFileName, int nLutSize, int fVerbose )
+{
+ abctime clk = Abc_Clock();
+ Cnf_Dat_t * pCnf;
+ pCnf = Mf_ManGenerateCnf( p, nLutSize, 0, 1, fVerbose );
+ Cnf_DataWriteIntoFile( pCnf, pFileName, 0, NULL, NULL );
+// if ( fVerbose )
+ {
+ printf( "CNF stats: Vars = %6d. Clauses = %7d. Literals = %8d. ", pCnf->nVars, pCnf->nClauses, pCnf->nLiterals );
+ Abc_PrintTime( 1, "Time", Abc_Clock() - clk );
+ }
+ Cnf_DataFree(pCnf);
}
////////////////////////////////////////////////////////////////////////