From 73fd90eabf898b586a07b8832c574147f1cd420f Mon Sep 17 00:00:00 2001
From: Tomas Oberhuber <tomas.oberhuber@fjfi.cvut.cz>
Date: Fri, 20 May 2011 18:30:53 +0000
Subject: [PATCH] Debuging tnlAdaptiveRgCSRMatrix.
---
src/matrix/tnlAdaptiveRgCSRMatrix.h | 87 +++--
tests/run-sparse-matrix-benchmark | 1 +
tests/sparse-matrix-benchmark.h | 2 +
tests/tnlSpmvBenchmark.h | 1 +
tests/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h | 4 +-
tests/tnlSpmvBenchmarkHybridMatrix.h | 2 +-
tests/unit-tests/Makefile.am | 2 +-
tests/unit-tests/Makefile.in | 2 +-
.../matrix/tnlAdaptiveRgCSRMatrixTester.h | 322 +++++++++++++++---
tests/unit-tests/tnl-unit-tests.cpp | 6 +-
10 files changed, 350 insertions(+), 79 deletions(-)
diff --git a/src/matrix/tnlAdaptiveRgCSRMatrix.h b/src/matrix/tnlAdaptiveRgCSRMatrix.h
index 624c97151e..a2af863172 100644
--- a/src/matrix/tnlAdaptiveRgCSRMatrix.h
+++ b/src/matrix/tnlAdaptiveRgCSRMatrix.h
@@ -276,10 +276,10 @@ void tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: tuneFormat( const Index cu
}
template< typename Real, tnlDevice Device, typename Index >
-bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatrix< Real, tnlHost, Index >& mat )
+bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatrix< Real, tnlHost, Index >& csrMatrix )
{
dbgFunctionName( "tnlAdaptiveRgCSRMatrix< Real, tnlHost >", "copyFrom" );
- if( ! this -> setSize( mat.getSize() ) )
+ if( ! this -> setSize( csrMatrix.getSize() ) )
return false;
Index nonzerosInGroup( 0 );
@@ -304,7 +304,7 @@ bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatr
if( groupEnd > this -> getSize() )
groupEnd = this -> getSize();
- nonzerosInGroup = mat. row_offsets[ groupEnd ] - mat. row_offsets[ groupBegin ];
+ nonzerosInGroup = csrMatrix. row_offsets[ groupEnd ] - csrMatrix. row_offsets[ groupBegin ];
rowsInGroup = groupEnd - groupBegin;
if( nonzerosInGroup < targetNonzeroesPerGroup &&
@@ -326,7 +326,7 @@ bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatr
Index usedThreads = 0;
for( Index i = groupBegin; i < groupEnd; i++ )
{
- double nonzerosInRow = mat. getNonzeroElementsInRow( i );
+ double nonzerosInRow = csrMatrix. getNonzeroElementsInRow( i );
double nonzerosInRowRatio = nonzerosInRow / ( double ) nonzerosInGroup;
usedThreads += threadsPerRow[ i - groupBegin ] = floor( freeThreads * nonzerosInRowRatio );
}
@@ -357,7 +357,7 @@ bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatr
Index rounds( 0 ), roundsFinal( 0 );
for( Index i = groupBegin; i < groupEnd; i++ )
{
- double nonzerosInRow = mat. getNonzeroElementsInRow( i );
+ double nonzerosInRow = csrMatrix. getNonzeroElementsInRow( i );
rounds = ceil( nonzerosInRow / ( double ) threadsPerRow[ i - groupBegin ] );
roundsFinal = Max( rounds, roundsFinal );
}
@@ -392,7 +392,7 @@ bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatr
dbgCout( "Allocating " << numberOfStoredValues << " elements.");
if( ! setNonzeroElements( numberOfStoredValues ) )
return false;
- artificial_zeros = numberOfStoredValues - mat. getNonzeroElements();
+ artificial_zeros = numberOfStoredValues - csrMatrix. getNonzeroElements();
last_nonzero_element = numberOfStoredValues;
@@ -413,7 +413,7 @@ bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatr
*/
for( Index j = 0; j < groupInfo[ i ]. numRows; j++ )
{
- NZperRow[ j ] = mat. getNonzeroElementsInRow( baseRow + j );
+ NZperRow[ j ] = csrMatrix. getNonzeroElementsInRow( baseRow + j );
if( j > 0 )
threadsPerRow[ j ] = threads[ baseRow + j ] - threads[ baseRow + j - 1 ];
else
@@ -429,10 +429,10 @@ bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatr
{
if( counters[ j ] < NZperRow[ j ] )
{
- Index pos = mat. row_offsets[ baseRow + j ] + counters[ j ];
+ Index pos = csrMatrix. row_offsets[ baseRow + j ] + counters[ j ];
dbgCout( "Inserting data from CSR format at position " << pos << " to AdaptiveRgCSR at " << index );
- nonzeroElements[ index ] = mat. nonzero_elements[ pos ];
- columns[ index ] = mat.columns[ pos ];
+ nonzeroElements[ index ] = csrMatrix. nonzero_elements[ pos ];
+ columns[ index ] = csrMatrix. columns[ pos ];
}
else
{
@@ -442,6 +442,7 @@ bool tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: copyFrom( const tnlCSRMatr
}
counters[ j ] ++;
index ++;
+ tnlAssert( index < numberOfStoredValues, );
}
}
}
@@ -518,7 +519,7 @@ Real tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: getElement( Index row,
if( columns[ pointer ] == column )
return nonzeroElements[ pointer ];
pointer ++;
- }
+ }
}
return 0.0;
}
@@ -547,54 +548,76 @@ void tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: vectorProduct( const tnlLo
const Index TB_SIZE = 256;
- const Index MAX_ROWS = 128;
+ const Index MAX_ROWS = 256;
if( Device == tnlHost )
{
- Index idx[TB_SIZE];
- Real psum[TB_SIZE]; //partial sums for each thread
- Index limits[MAX_ROWS + 1]; //indices of first threads for each row + index of first unused thread
- Real results[MAX_ROWS];
+ Index idx[ TB_SIZE ];
+ Real psum[ TB_SIZE ]; //partial sums for each thread
+ Index limits[ MAX_ROWS + 1 ]; //indices of first threads for each row + index of first unused thread
+ Real results[ MAX_ROWS ];
/****
* Go over all groups ...
*/
- for( Index group = 0; group < this -> numberOfGroups; group ++ )
+ dbgExpr( this -> numberOfGroups );
+ for( Index groupId = 0; groupId < this -> numberOfGroups; groupId ++ )
{
/****
* In each group compute partial sums of each thread
*/
+ dbgExpr( groupId );
+ dbgExpr( this -> usedThreadsInGroup[ groupId ] );
for( Index thread = 0;
- thread < this -> usedThreadsInGroup[ group ];
+ thread < this -> usedThreadsInGroup[ groupId ];
thread ++ )
{
- idx[ thread ] = this -> groupInfo[ group ]. idxFirstValue + thread;
- psum[thread] = 0;
-
+ idx[ thread ] = this -> groupInfo[ groupId ]. idxFirstValue + thread;
+ psum[ thread ] = 0;
for( Index j = 0;
- j < this -> groupInfo[ group ]. numRounds;
+ j < this -> groupInfo[ groupId ]. numRounds;
j ++ )
{
- psum[ thread ] += this -> nonzeroElements[ idx[ thread ] ] * vec[ this -> columns[ idx[ thread ] ] ];
- idx[ j ] += TB_SIZE;
+ if( this -> columns[ idx[ thread ] ] != -1 )
+ psum[ thread ] += this -> nonzeroElements[ idx[ thread ] ] * vec[ this -> columns[ idx[ thread ] ] ];
+ idx[ thread ] += TB_SIZE;
}
+ dbgExpr( psum[ thread ] );
}
+ for( Index row = groupInfo[ groupId ]. idxFirstRow;
+ row < groupInfo[ groupId ]. idxFirstRow + groupInfo[ groupId ]. numRows;
+ row ++ )
+ {
+ Index firstThreadInRow = 0;
+ if( row > 0 )
+ firstThreadInRow = threads[ row - 1 ];
+ Index lastThreadInRow = threads[ row ];
+
+ dbgCout( "Row: " << row << " firstThreadInRow: " << firstThreadInRow << " lastThreadInRow: " << lastThreadInRow );
+ result[ row ] = 0.0;
+ for( Index thread = firstThreadInRow; thread < lastThreadInRow; thread ++ )
+ {
+ result[ row ] += psum[ thread ];
+ dbgCout( "Thread: " << thread << " psum[ thread ]: " << psum[ thread ] << " result[ row ]: " << result[ row ] );
+ }
+ }
+#ifdef UNDEF
/****
* Compute local copy of thread indexes mapped to given row of the group.
* (this is only to simulate copying data to the fast shared memory on GPU)
*/
for( Index thread = 0;
- thread < this -> groupInfo[ group ]. numRows;
+ thread < this -> groupInfo[ groupId ]. numRows;
thread ++ )
- limits[ thread ] = this -> threads[ this -> groupInfo[ group ]. idxFirstRow + thread ];
+ limits[ thread ] = this -> threads[ this -> groupInfo[ groupId ]. idxFirstRow + thread ];
/****
* For convenience, add the index of first unused row.
*/
- limits[ this -> groupInfo[ group ]. numRows ] = this -> usedThreadsInGroup[ group ];
+ limits[ this -> groupInfo[ groupId ]. numRows ] = this -> usedThreadsInGroup[ groupId ];
//reduction of partial sums and writing to the output
for( Index thread = 0;
- thread < this -> groupInfo[ group ]. numRows;
+ thread < this -> groupInfo[ groupId ]. numRows;
thread ++) //for threads corresponding to rows in group
{
results[ thread ] = 0;
@@ -602,8 +625,9 @@ void tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: vectorProduct( const tnlLo
j < limits[ thread + 1 ];
j++ ) //sum up partial sums belonging to that row
results[ thread ] += psum[ j ];
- result[ this -> groupInfo[ group ]. idxFirstRow + thread ] = results[ thread ];
+ result[ this -> groupInfo[ groupId ]. idxFirstRow + thread ] = results[ thread ];
}
+#endif
}
}
if( Device == tnlCuda )
@@ -640,6 +664,11 @@ template< typename Real, tnlDevice Device, typename Index >
void tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: printOut( ostream& str,
const Index lines ) const
{
+ /*****
+ * THIS IS NOT CORRECT
+ */
+ cerr << "tnlAdaptiveRgCSRMatrix< Real, Device, Index > :: printOut is not correct" << endl;
+ abort();
str << "Structure of tnlAdaptiveRgCSRMatrix" << endl;
str << "Matrix name:" << this -> getName() << endl;
str << "Matrix size:" << this -> getSize() << endl;
diff --git a/tests/run-sparse-matrix-benchmark b/tests/run-sparse-matrix-benchmark
index c6feaa096a..be8869d2ca 100755
--- a/tests/run-sparse-matrix-benchmark
+++ b/tests/run-sparse-matrix-benchmark
@@ -7,6 +7,7 @@ IWD="$PWD"
DEBUG="no"
SPARSE_MATRIX_BENCHMARK="$IWD/sparse-matrix-benchmark"
SPARSE_MATRIX_BENCHMARK_DBG="$IWD/.libs/sparse-matrix-benchmark-dbg"
+#SPARSE_MATRIX_BENCHMARK="$IWD/.libs/sparse-matrix-benchmark-dbg"
STOP_TIME="1"
source ../tnl-env-variables
diff --git a/tests/sparse-matrix-benchmark.h b/tests/sparse-matrix-benchmark.h
index b2a448c692..9bc61a6d56 100644
--- a/tests/sparse-matrix-benchmark.h
+++ b/tests/sparse-matrix-benchmark.h
@@ -221,6 +221,7 @@ bool benchmarkMatrix( const tnlString& input_file,
}
}
+#ifdef UNDEF
/****
* Row-Grouped CSR format
*/
@@ -371,6 +372,7 @@ bool benchmarkMatrix( const tnlString& input_file,
}
+#endif
csrMatrix. vectorProduct( refX, refB );
/****
diff --git a/tests/tnlSpmvBenchmark.h b/tests/tnlSpmvBenchmark.h
index 86c891ec58..b412eff1f0 100644
--- a/tests/tnlSpmvBenchmark.h
+++ b/tests/tnlSpmvBenchmark.h
@@ -238,6 +238,7 @@ void tnlSpmvBenchmark< Real, Device, Index, Matrix > :: runBenchmark( const tnlL
benchmarkWasSuccesful = false;
firstErrorOccurence = j;
//cerr << " b[ " << j << " ] = " << resB[ j ] << " while refB[ " << j << " ] = " << refB[ j ] << endl;
+ //abort();
}
}
diff --git a/tests/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h b/tests/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h
index a27b873188..d1b4773ebe 100644
--- a/tests/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h
+++ b/tests/tnlSpmvBenchmarkAdaptiveRgCSRMatrix.h
@@ -68,7 +68,7 @@ template< typename Real,
typename Index>
bool tnlSpmvBenchmarkAdaptiveRgCSRMatrix< Real, Device, Index > :: setup( const tnlCSRMatrix< Real, tnlHost, Index >& matrix )
{
- tnlAssert( this -> groupSize > 0, cerr << "groupSize = " << this -> groupSize );
+ //tnlAssert( this -> groupSize > 0, cerr << "groupSize = " << this -> groupSize );
if( Device == tnlHost )
{
if( ! this -> matrix. copyFrom( matrix ) )
@@ -102,7 +102,7 @@ template< typename Real,
typename Index >
void tnlSpmvBenchmarkAdaptiveRgCSRMatrix< Real, Device, Index > :: writeProgress() const
{
- cout << left << setw( this -> formatColumnWidth - 15 ) << "Row-grouped CSR ";
+ cout << left << setw( this -> formatColumnWidth - 15 ) << "Adaptive Row-grouped CSR ";
if( Device == tnlCuda )
cout << setw( 5 ) << this -> groupSize
<< setw( 10 ) << this -> cudaBlockSize;
diff --git a/tests/tnlSpmvBenchmarkHybridMatrix.h b/tests/tnlSpmvBenchmarkHybridMatrix.h
index cdda4122d4..7505b04ec1 100644
--- a/tests/tnlSpmvBenchmarkHybridMatrix.h
+++ b/tests/tnlSpmvBenchmarkHybridMatrix.h
@@ -59,7 +59,7 @@ void tnlSpmvBenchmarkHybridMatrix< Real, Index > :: setFileName( const tnlString
template< typename Real, typename Index>
bool tnlSpmvBenchmarkHybridMatrix< Real, Index > :: setup( const tnlCSRMatrix< Real, tnlHost, Index >& matrix )
{
-
+ return true;
}
template< typename Real,
diff --git a/tests/unit-tests/Makefile.am b/tests/unit-tests/Makefile.am
index b35b63fc99..b6db4a4361 100644
--- a/tests/unit-tests/Makefile.am
+++ b/tests/unit-tests/Makefile.am
@@ -67,7 +67,7 @@ endif
# ../../src/debug/libtnldebug-dbg-0.1.la
#endif
-TESTS = tnl-unit-tests
+TESTS = tnl-unit-tests-dbg
#if BUILD_CUDA
#TESTS += tnl-cuda-unit-tests
diff --git a/tests/unit-tests/Makefile.in b/tests/unit-tests/Makefile.in
index d3a66eed46..c4de5fbc90 100644
--- a/tests/unit-tests/Makefile.in
+++ b/tests/unit-tests/Makefile.in
@@ -42,7 +42,7 @@ check_PROGRAMS = tnl-unit-tests$(EXEEXT) $(am__EXEEXT_1) \
$(am__EXEEXT_2)
@BUILD_CUDA_TRUE@am__append_1 = tnl-cuda-unit-tests
@BUILD_DBG_TRUE@am__append_2 = tnl-unit-tests-dbg
-TESTS = tnl-unit-tests$(EXEEXT)
+TESTS = tnl-unit-tests-dbg$(EXEEXT)
subdir = tests/unit-tests
DIST_COMMON = $(srcdir)/Makefile.am $(srcdir)/Makefile.in
ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
diff --git a/tests/unit-tests/matrix/tnlAdaptiveRgCSRMatrixTester.h b/tests/unit-tests/matrix/tnlAdaptiveRgCSRMatrixTester.h
index ae7b79e8c9..f92e28ea3e 100644
--- a/tests/unit-tests/matrix/tnlAdaptiveRgCSRMatrixTester.h
+++ b/tests/unit-tests/matrix/tnlAdaptiveRgCSRMatrixTester.h
@@ -38,59 +38,219 @@ template< class T > class tnlAdaptiveRgCSRMatrixTester : public CppUnit :: TestC
CppUnit :: TestSuite* suiteOfTests = new CppUnit :: TestSuite( "tnlAdaptiveRgCSRMatrixTester" );
CppUnit :: TestResult result;
suiteOfTests -> addTest( new CppUnit :: TestCaller< tnlAdaptiveRgCSRMatrixTester< T > >(
- "diagonalMatrixTest",
- & tnlAdaptiveRgCSRMatrixTester< T > :: diagonalMatrixTest )
+ "ifDiagonalMatrixIsStoredProperly",
+ & tnlAdaptiveRgCSRMatrixTester< T > :: ifDiagonalMatrixIsStoredProperly )
);
suiteOfTests -> addTest( new CppUnit :: TestCaller< tnlAdaptiveRgCSRMatrixTester< T > >(
- "triDiagonalMatrixTest",
- & tnlAdaptiveRgCSRMatrixTester< T > :: triDiagonalMatrixTest )
+ "ifTriDiagonalMatrixIsStoredProperly",
+ & tnlAdaptiveRgCSRMatrixTester< T > :: ifTriDiagonalMatrixIsStoredProperly )
);
suiteOfTests -> addTest( new CppUnit :: TestCaller< tnlAdaptiveRgCSRMatrixTester< T > >(
- "upperTriangularMatrixTest",
- & tnlAdaptiveRgCSRMatrixTester< T > :: upperTriangularMatrixTest )
+ "ifUpperTriangularMatrixIsStoredProperly",
+ & tnlAdaptiveRgCSRMatrixTester< T > :: ifUpperTriangularMatrixIsStoredProperly )
);
+ suiteOfTests -> addTest( new CppUnit :: TestCaller< tnlAdaptiveRgCSRMatrixTester< T > >(
+ "ifSpmvWithDiagonalMatrixWorks",
+ & tnlAdaptiveRgCSRMatrixTester< T > :: ifSpmvWithDiagonalMatrixWorks )
+ );
+ suiteOfTests -> addTest( new CppUnit :: TestCaller< tnlAdaptiveRgCSRMatrixTester< T > >(
+ "ifSpmvWithTriDiagonalMatrixWorks",
+ & tnlAdaptiveRgCSRMatrixTester< T > :: ifSpmvWithTriDiagonalMatrixWorks )
+ );
+
+ suiteOfTests -> addTest( new CppUnit :: TestCaller< tnlAdaptiveRgCSRMatrixTester< T > >(
+ "ifSpmvWithUpperTriangularMatrixWorks",
+ & tnlAdaptiveRgCSRMatrixTester< T > :: ifSpmvWithUpperTriangularMatrixWorks )
+ );
+ suiteOfTests -> addTest( new CppUnit :: TestCaller< tnlAdaptiveRgCSRMatrixTester< T > >(
+ "ifSpmvWithBcsstk20MatrixWorks",
+ & tnlAdaptiveRgCSRMatrixTester< T > :: ifSpmvWithBcsstk20MatrixWorks )
+ );
+
- return suiteOfTests;
+ return suiteOfTests;
}
- void diagonalMatrixTest()
+ void setDiagonalMatrix( tnlCSRMatrix< T >& csrMatrix,
+ tnlAdaptiveRgCSRMatrix< T >& argcsrMatrix,
+ const int size )
{
- tnlCSRMatrix< T > csr_matrix( "test-matrix:Diagonal" );
- tnlAdaptiveRgCSRMatrix< T > argcsrMatrix( "test-matrix:Diagonal" );
- csr_matrix. setSize( 12 );
- csr_matrix. setNonzeroElements( 12 );
- for( int i = 0; i < 12; i ++ )
- csr_matrix. setElement( i, i, T( i + 1 ) );
+ csrMatrix. setSize( size );
+ csrMatrix. setNonzeroElements( size );
+ for( int i = 0; i < size; i ++ )
+ csrMatrix. setElement( i, i, T( i + 1 ) );
//cerr << "Copying data to coalesced CSR matrix." << endl;
- argcsrMatrix. copyFrom( csr_matrix );
+ argcsrMatrix. copyFrom( csrMatrix );
//argcsrMatrix. printOut( cout );
+ }
+
+ void setTridiagonalMatrix( tnlCSRMatrix< T >& csrMatrix,
+ tnlAdaptiveRgCSRMatrix< T >& argcsrMatrix,
+ const int size )
+ {
+ csrMatrix. setSize( size );
+ csrMatrix. setNonzeroElements( size * 3 - 2 );
+ T data[] = { -1.0, 2.0, -1.0 };
+ int offsets[] = { -1, 0, 1 };
+ for( int i = 0; i < size; i ++ )
+ {
+ csrMatrix. insertRow( i, // row
+ 3, // elements
+ data, // data
+ i, // first column
+ offsets );
+ }
+ argcsrMatrix. copyFrom( csrMatrix );
+ //argcsr_matrix. printOut( cout );
+ }
+
+ void setUpperTriangularMatrix( tnlCSRMatrix< T >& csrMatrix,
+ tnlAdaptiveRgCSRMatrix< T >& argcsrMatrix,
+ const int size )
+ {
+ csrMatrix. setSize( size );
+ csrMatrix. setNonzeroElements( size * size );
+ for( int i = 0; i < size; i ++ )
+ for( int j = i; j < size; j ++ )
+ csrMatrix. setElement( i, j, 1.0 );
+
+ argcsrMatrix. copyFrom( csrMatrix );
+ //argcsr_matrix. printOut( cout );
+ }
+
+ void setBcsstk20Matrix( tnlCSRMatrix< T >& csrMatrix,
+ tnlAdaptiveRgCSRMatrix< T >& argcsrMatrix )
+ {
+ /****
+ * Sets a real matrix from a matrix market
+ */
+ csrMatrix. setSize( 50 );
+ csrMatrix. setNonzeroElements( 100 );
+
+ csrMatrix. setElement( 1, 1, 3.4651842369950e+08 );
+ csrMatrix. setElement( 3, 1, 3.0695529658430e+10 );
+ csrMatrix. setElement( 4, 1, -3.4651842369950e+08 );
+ csrMatrix. setElement( 6, 1, 3.0695529658430e+10 );
+ csrMatrix. setElement( 2, 2, 5.4150467497750e+08 );
+ csrMatrix. setElement( 5, 2, -5.4150467497750e+08 );
+ csrMatrix. setElement( 3, 3, 3.6254562588700e+12 );
+ csrMatrix. setElement( 4, 3, -3.0695529658430e+10 );
+ csrMatrix. setElement( 6, 3, 1.8127281294350e+12 );
+ csrMatrix. setElement( 4, 4, 5.8200610488580e+09 );
+ csrMatrix. setElement( 6, 4, 1.6324889406770e+11 );
+ csrMatrix. setElement( 7, 4, -5.4735426251580e+09 );
+ csrMatrix. setElement( 9, 4, 1.9394442372610e+11 );
+ csrMatrix. setElement( 5, 5, 1.7185252532270e+09 );
+ csrMatrix. setElement( 8, 5, -1.1770205782500e+09 );
+ csrMatrix. setElement( 6, 6, 1.2788184938840e+13 );
+ csrMatrix. setElement( 7, 6, -1.9394442372610e+11 );
+ csrMatrix. setElement( 9, 6, 4.5813643399870e+12 );
+ csrMatrix. setElement( 7, 7, 1.1034132683920e+10 );
+ csrMatrix. setElement( 9, 7, 3.0843578836850e+09 );
+ csrMatrix. setElement( 10, 7, -5.5605900587650e+09 );
+ csrMatrix. setElement( 12, 7, 1.9702878160980e+11 );
+ csrMatrix. setElement( 8, 8, 2.3681428407440e+09 );
+ csrMatrix. setElement( 11, 8, -1.1911222624940e+09 );
+ csrMatrix. setElement( 9, 9, 1.8471175055240e+13 );
+ csrMatrix. setElement( 10, 9, -1.9702878160980e+11 );
+ csrMatrix. setElement( 12, 9, 4.6542231876330e+12 );
+ csrMatrix. setElement( 10, 10, 1.1211709448480e+10 );
+ csrMatrix. setElement( 12, 10, 3.2077321990330e+09 );
+ csrMatrix. setElement( 13, 10, -5.6511193897150e+09 );
+ csrMatrix. setElement( 15, 10, 2.0023651380880e+11 );
+ csrMatrix. setElement( 11, 11, 2.3954060969490e+09 );
+ csrMatrix. setElement( 14, 11, -1.2042838344550e+09 );
+ csrMatrix. setElement( 12, 12, 1.8768439153640e+13 );
+ csrMatrix. setElement( 13, 12, -2.0023651380880e+11 );
+ csrMatrix. setElement( 15, 12, 4.7299963891850e+12 );
+ csrMatrix. setElement( 13, 13, 1.1392768110380e+10 );
+ csrMatrix. setElement( 15, 13, 3.2077321990300e+09 );
+ csrMatrix. setElement( 16, 13, -5.7416487206660e+09 );
+ csrMatrix. setElement( 18, 13, 2.0344424600780e+11 );
+ csrMatrix. setElement( 14, 14, 2.4226693531550e+09 );
+ csrMatrix. setElement( 17, 14, -1.2183855187000e+09 );
+ csrMatrix. setElement( 15, 15, 1.9071531959840e+13 );
+ csrMatrix. setElement( 16, 15, -2.0344424600780e+11 );
+ csrMatrix. setElement( 18, 15, 4.8057695907370e+12 );
+ csrMatrix. setElement( 16, 16, 1.1570344874940e+10 );
+ csrMatrix. setElement( 18, 16, 3.0843578836840e+09 );
+ csrMatrix. setElement( 19, 16, -5.8286961542720e+09 );
+ csrMatrix. setElement( 21, 16, 2.0652860389150e+11 );
+ csrMatrix. setElement( 17, 17, 2.4499326093600e+09 );
+ csrMatrix. setElement( 20, 17, -1.2315470906610e+09 );
+ csrMatrix. setElement( 18, 18, 1.9368796058240e+13 );
+ csrMatrix. setElement( 19, 18, -2.0652860389150e+11 );
+ csrMatrix. setElement( 21, 18, 4.8786284383830e+12 );
+ csrMatrix. setElement( 19, 19, 1.1747921639490e+10 );
+ csrMatrix. setElement( 21, 19, 3.2077321990320e+09 );
+ csrMatrix. setElement( 22, 19, -5.9192254852220e+09 );
+ csrMatrix. setElement( 24, 19, 2.0973633609060e+11 );
+ csrMatrix. setElement( 20, 20, 2.4771958655660e+09 );
+ csrMatrix. setElement( 23, 20, -1.2456487749050e+09 );
+ csrMatrix. setElement( 21, 21, 1.9666060156640e+13 );
+ csrMatrix. setElement( 22, 21, -2.0973633609060e+11 );
+ csrMatrix. setElement( 24, 21, 4.9544016399350e+12 );
+ csrMatrix. setElement( 22, 22, 1.1928980301390e+10 );
+ csrMatrix. setElement( 24, 22, 3.2077321990300e+09 );
+ csrMatrix. setElement( 25, 22, -6.0097548161730e+09 );
+ csrMatrix. setElement( 27, 22, 2.1294406828960e+11 );
+ csrMatrix. setElement( 23, 23, 2.5044591217710e+09 );
+ csrMatrix. setElement( 26, 23, -1.2588103468660e+09 );
+ csrMatrix. setElement( 24, 24, 1.9969152962840e+13 );
+ csrMatrix. setElement( 25, 24, -2.1294406828960e+11 );
+ csrMatrix. setElement( 27, 24, 5.0301748414870e+12 );
+ csrMatrix. setElement( 25, 25, 1.2106557065950e+10 );
+ csrMatrix. setElement( 27, 25, 3.0843578836860e+09 );
+ csrMatrix. setElement( 28, 25, -6.0968022497790e+09 );
+ csrMatrix. setElement( 30, 25, 2.1602842617330e+11 );
+ csrMatrix. setElement( 26, 26, 2.5317223779770e+09 );
+ csrMatrix. setElement( 29, 26, -1.2729120311100e+09 );
+ csrMatrix. setElement( 27, 27, 2.0266417061240e+13 );
+ csrMatrix. setElement( 28, 27, -2.1602842617330e+11 );
+ csrMatrix. setElement( 30, 27, 5.1030336891330e+12 );
+ csrMatrix. setElement( 28, 28, 1.2284133830500e+10 );
+ csrMatrix. setElement( 30, 28, 3.2077321988540e+09 );
+ csrMatrix. setElement( 31, 28, -6.1873315807220e+09 );
+ csrMatrix. setElement( 33, 28, 2.1923615837210e+11 );
+ csrMatrix. setElement( 29, 29, 2.5589856341820e+09 );
+ csrMatrix. setElement( 32, 29, -1.2860736030710e+09 );
+ csrMatrix. setElement( 30, 30, 2.0563681159630e+13 );
+ csrMatrix. setElement( 31, 30, -2.1923615837210e+11 );
+ csrMatrix. setElement( 33, 30, 5.1788068906830e+12 );
+ csrMatrix. setElement( 31, 31, 1.2465192492400e+10 );
+ csrMatrix. setElement( 33, 31, 3.2077321992110e+09 );
+ csrMatrix. setElement( 34, 31, -6.2778609116800e+09 );
+ csrMatrix. setElement( 36, 31, 2.2244389057130e+11 );
+ csrMatrix. setElement( 32, 32, 2.5862488903870e+09 );
+ csrMatrix. setElement( 35, 32, -1.3001752873160e+09 );
+ csrMatrix. setElement( 33, 33, 2.0866773965840e+13 );
+
+
+ argcsrMatrix. copyFrom( csrMatrix );
+
+ }
+
+ void ifDiagonalMatrixIsStoredProperly()
+ {
+ const int size = 12;
+ tnlCSRMatrix< T > csrMatrix( "test-matrix:Diagonal" );
+ tnlAdaptiveRgCSRMatrix< T > argcsrMatrix( "test-matrix:Diagonal" );
+ setDiagonalMatrix( csrMatrix, argcsrMatrix, size );
+
bool error( false );
- for( int i = 0; i < 12; i ++ )
+ for( int i = 0; i < size; i ++ )
if( argcsrMatrix. getElement( i, i ) != T( i + 1 ) )
error = true;
CPPUNIT_ASSERT( ! error );
};
- void triDiagonalMatrixTest()
+ void ifTriDiagonalMatrixIsStoredProperly()
{
tnlCSRMatrix< T > csr_matrix( "test-matrix:Tridiagonal" );
tnlAdaptiveRgCSRMatrix< T > argcsr_matrix( "test-matrix:Tridiagonal" );
int size = 12;
- csr_matrix. setSize( size );
- csr_matrix. setNonzeroElements( size * 3 - 2 );
- T data[] = { -1.0, 2.0, -1.0 };
- int offsets[] = { -1, 0, 1 };
- for( int i = 0; i < size; i ++ )
- {
- csr_matrix. insertRow( i, // row
- 3, // elements
- data, // data
- i, // first column
- offsets );
- }
- argcsr_matrix. copyFrom( csr_matrix );
- //argcsr_matrix. printOut( cout );
+ setTridiagonalMatrix( csr_matrix, argcsr_matrix, size );
bool error( false );
for( int i = 0; i < size; i ++ )
for( int j = 0; j < size; j ++ )
@@ -99,20 +259,12 @@ template< class T > class tnlAdaptiveRgCSRMatrixTester : public CppUnit :: TestC
CPPUNIT_ASSERT( ! error );
}
- void upperTriangularMatrixTest()
+ void ifUpperTriangularMatrixIsStoredProperly()
{
tnlCSRMatrix< T > csr_matrix( "test-matrix:upperTriangular" );
tnlAdaptiveRgCSRMatrix< T > argcsr_matrix( "test-matrix:upperTriangular" );
- int size = 12;
- csr_matrix. setSize( size );
- csr_matrix. setNonzeroElements( size * size );
- for( int i = 0; i < size; i ++ )
- for( int j = i; j < size; j ++ )
- csr_matrix. setElement( i, j, 1.0 );
-
- argcsr_matrix. copyFrom( csr_matrix );
- //cerr << "----------------" << endl;
- //argcsr_matrix. printOut( cout );
+ const int size = 12;
+ setUpperTriangularMatrix( csr_matrix, argcsr_matrix, size );
bool error( false );
for( int i = 0; i < size; i ++ )
for( int j = 0; j < size; j ++ )
@@ -120,9 +272,95 @@ template< class T > class tnlAdaptiveRgCSRMatrixTester : public CppUnit :: TestC
error = true;
CPPUNIT_ASSERT( ! error );
+ }
+
+ void ifSpmvWithDiagonalMatrixWorks()
+ {
+ const int size = 12;
+ tnlCSRMatrix< T > csrMatrix( "test-matrix:Diagonal" );
+ tnlAdaptiveRgCSRMatrix< T > argcsrMatrix( "test-matrix:Diagonal" );
+ setDiagonalMatrix( csrMatrix, argcsrMatrix, size );
+
+ tnlLongVector< T > x( "x" ), b1( "b1" ), b2( "b2" );
+ x. setSize( size );
+ b1. setSize( size );
+ b2. setSize( size );
+ x. setValue( 1.0 );
+ csrMatrix. vectorProduct( x, b1 );
+ argcsrMatrix. vectorProduct( x, b2 );
+
+ cout << b1 << endl;
+ cout << b2 << endl;
+ CPPUNIT_ASSERT( b1 == b2 );
}
+ void ifSpmvWithTriDiagonalMatrixWorks()
+ {
+ const int size = 12;
+ tnlCSRMatrix< T > csrMatrix( "test-matrix:TriDiagonal" );
+ tnlAdaptiveRgCSRMatrix< T > argcsrMatrix( "test-matrix:TriDiagonal" );
+ setTridiagonalMatrix( csrMatrix, argcsrMatrix, size );
+
+ tnlLongVector< T > x( "x" ), b1( "b1" ), b2( "b2" );
+ x. setSize( size );
+ b1. setSize( size );
+ b2. setSize( size );
+ x. setValue( 1.0 );
+ csrMatrix. vectorProduct( x, b1 );
+ argcsrMatrix. vectorProduct( x, b2 );
+
+ cout << b1 << endl;
+ cout << b2 << endl;
+
+ CPPUNIT_ASSERT( b1 == b2 );
+ }
+
+ void ifSpmvWithUpperTriangularMatrixWorks()
+ {
+ const int size = 12;
+ tnlCSRMatrix< T > csrMatrix( "test-matrix:TriDiagonal" );
+ tnlAdaptiveRgCSRMatrix< T > argcsrMatrix( "test-matrix:TriDiagonal" );
+ setUpperTriangularMatrix( csrMatrix, argcsrMatrix, size );
+
+ tnlLongVector< T > x( "x" ), b1( "b1" ), b2( "b2" );
+ x. setSize( size );
+ b1. setSize( size );
+ b2. setSize( size );
+ x. setValue( 1.0 );
+ csrMatrix. vectorProduct( x, b1 );
+ argcsrMatrix. vectorProduct( x, b2 );
+
+ cout << b1 << endl;
+ cout << b2 << endl;
+
+ CPPUNIT_ASSERT( b1 == b2 );
+ }
+
+ void ifSpmvWithBcsstk20MatrixWorks()
+ {
+ tnlCSRMatrix< T > csrMatrix( "test-matrix:TriDiagonal" );
+ tnlAdaptiveRgCSRMatrix< T > argcsrMatrix( "test-matrix:TriDiagonal" );
+ setBcsstk20Matrix( csrMatrix, argcsrMatrix );
+ const int size = csrMatrix. getSize();
+
+ tnlLongVector< T > x( "x" ), b1( "b1" ), b2( "b2" );
+ x. setSize( size );
+ b1. setSize( size );
+ b2. setSize( size );
+ x. setValue( 1.0 );
+ csrMatrix. vectorProduct( x, b1 );
+ argcsrMatrix. vectorProduct( x, b2 );
+
+ cout << b1 << endl;
+ cout << b2 << endl;
+
+ CPPUNIT_ASSERT( b1 == b2 );
+ }
+
+
+
+
};
#endif /* TNLAdaptiveRgCSRMATRIXTESTER_H_ */
diff --git a/tests/unit-tests/tnl-unit-tests.cpp b/tests/unit-tests/tnl-unit-tests.cpp
index e1e84498f2..77af728501 100644
--- a/tests/unit-tests/tnl-unit-tests.cpp
+++ b/tests/unit-tests/tnl-unit-tests.cpp
@@ -76,12 +76,12 @@ int main( int argc, char* argv[] )
runner. addTest( tnlRgCSRMatrixTester< float > :: suite() );
runner. addTest( tnlRgCSRMatrixTester< double > :: suite() );
- runner. addTest( tnlAdaptiveRgCSRMatrixTester< float > :: suite() );
- runner. addTest( tnlAdaptiveRgCSRMatrixTester< double > :: suite() );
-
runner. addTest( tnlEllpackMatrixTester< float > :: suite() );
runner. addTest( tnlEllpackMatrixTester< double > :: suite() );
+ runner. addTest( tnlAdaptiveRgCSRMatrixTester< float > :: suite() );
+ runner. addTest( tnlAdaptiveRgCSRMatrixTester< double > :: suite() );
+
runner. addTest( tnlMPIMeshTester< float > :: suite() );
//runner. addTest( tnlSharedMemoryTester< tnlHost > :: suite() );
--
GitLab