mirror of
https://gitlab.com/libeigen/eigen.git
synced 2025-03-01 18:26:24 +08:00
factorize default performance related settings to a single file
included after the architecture specific files such that they can be adapted by each platform.
This commit is contained in:
Gael Guennebaud
parent
112c550b4a
commit
7dd81aad74
@ -214,9 +214,7 @@ struct Dense {};
|
||||
#include "src/Core/arch/NEON/PacketMath.h"
|
||||
#endif
|
||||
|
||||
#ifndef EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD
|
||||
#define EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD 8
|
||||
#endif
|
||||
#include "src/Core/arch/Default/Settings.h"
|
||||
|
||||
#include "src/Core/Functors.h"
|
||||
#include "src/Core/DenseBase.h"
|
||||
|
||||
@ -601,7 +601,7 @@ struct ei_conservative_resize_like_impl<Derived,OtherDerived,true>
|
||||
const int new_rows = Derived::RowsAtCompileTime==1 ? 1 : other.rows();
|
||||
const int new_cols = Derived::RowsAtCompileTime==1 ? other.cols() : 1;
|
||||
_this.derived().m_storage.conservativeResize(other.size(),new_rows,new_cols);
|
||||
|
||||
|
||||
if (num_new_elements > 0)
|
||||
_this.tail(num_new_elements) = other.tail(num_new_elements);
|
||||
}
|
||||
|
||||
65
Eigen/src/Core/arch/Default/Settings.h
Normal file
65
Eigen/src/Core/arch/Default/Settings.h
Normal file
@ -0,0 +1,65 @@
|
||||
// This file is part of Eigen, a lightweight C++ template library
|
||||
// for linear algebra.
|
||||
//
|
||||
// Copyright (C) 2008-2010 Gael Guennebaud <g.gael@free.fr>
|
||||
// Copyright (C) 2006-2008 Benoit Jacob <jacob.benoit.1@gmail.com>
|
||||
//
|
||||
// Eigen is free software; you can redistribute it and/or
|
||||
// modify it under the terms of the GNU Lesser General Public
|
||||
// License as published by the Free Software Foundation; either
|
||||
// version 3 of the License, or (at your option) any later version.
|
||||
//
|
||||
// Alternatively, you can redistribute it and/or
|
||||
// modify it under the terms of the GNU General Public License as
|
||||
// published by the Free Software Foundation; either version 2 of
|
||||
// the License, or (at your option) any later version.
|
||||
//
|
||||
// Eigen is distributed in the hope that it will be useful, but WITHOUT ANY
|
||||
// WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
|
||||
// FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License or the
|
||||
// GNU General Public License for more details.
|
||||
//
|
||||
// You should have received a copy of the GNU Lesser General Public
|
||||
// License and a copy of the GNU General Public License along with
|
||||
// Eigen. If not, see <http://www.gnu.org/licenses/>.
|
||||
|
||||
|
||||
/* All the parameters defined in this file can be specialized in the
|
||||
* architecture specific files, and/or by the user.
|
||||
* More to come... */
|
||||
|
||||
#ifndef EIGEN_DEFAULT_SETTINGS_H
|
||||
#define EIGEN_DEFAULT_SETTINGS_H
|
||||
|
||||
/** Defines the maximal loop size to enable meta unrolling of loops.
|
||||
* Note that the value here is expressed in Eigen's own notion of "number of FLOPS",
|
||||
* it does not correspond to the number of iterations or the number of instructions
|
||||
*/
|
||||
#ifndef EIGEN_UNROLLING_LIMIT
|
||||
#define EIGEN_UNROLLING_LIMIT 100
|
||||
#endif
|
||||
|
||||
/** Defines the threshold between a "small" and a "large" matrix.
|
||||
* This threshold is mainly used to select the proper product implementation.
|
||||
*/
|
||||
#ifndef EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD
|
||||
#define EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD 8
|
||||
#endif
|
||||
|
||||
/** Defines the maximal size in Bytes of blocks fitting in CPU cache.
|
||||
* The current value is set to generate blocks of 256x256 for float
|
||||
*
|
||||
* Typically for a single-threaded application you would set that to 25% of the size of your CPU caches in bytes
|
||||
*/
|
||||
#ifndef EIGEN_TUNE_FOR_CPU_CACHE_SIZE
|
||||
#define EIGEN_TUNE_FOR_CPU_CACHE_SIZE (sizeof(float)*256*256)
|
||||
#endif
|
||||
|
||||
/** Defines the maximal width of the blocks used in the triangular product and solver
|
||||
* for vectors (level 2 blas xTRMV and xTRSV). The default is 8.
|
||||
*/
|
||||
#ifndef EIGEN_TUNE_TRIANGULAR_PANEL_WIDTH
|
||||
#define EIGEN_TUNE_TRIANGULAR_PANEL_WIDTH 8
|
||||
#endif
|
||||
|
||||
#endif // EIGEN_DEFAULT_SETTINGS_H
|
||||
@ -78,30 +78,6 @@
|
||||
#define EIGEN_DEFAULT_MATRIX_STORAGE_ORDER_OPTION ColMajor
|
||||
#endif
|
||||
|
||||
/** Defines the maximal loop size to enable meta unrolling of loops.
|
||||
* Note that the value here is expressed in Eigen's own notion of "number of FLOPS",
|
||||
* it does not correspond to the number of iterations or the number of instructions
|
||||
*/
|
||||
#ifndef EIGEN_UNROLLING_LIMIT
|
||||
#define EIGEN_UNROLLING_LIMIT 100
|
||||
#endif
|
||||
|
||||
/** Defines the maximal size in Bytes of blocks fitting in CPU cache.
|
||||
* The current value is set to generate blocks of 256x256 for float
|
||||
*
|
||||
* Typically for a single-threaded application you would set that to 25% of the size of your CPU caches in bytes
|
||||
*/
|
||||
#ifndef EIGEN_TUNE_FOR_CPU_CACHE_SIZE
|
||||
#define EIGEN_TUNE_FOR_CPU_CACHE_SIZE (sizeof(float)*256*256)
|
||||
#endif
|
||||
|
||||
/** Defines the maximal width of the blocks used in the triangular product and solver
|
||||
* for vectors (level 2 blas xTRMV and xTRSV). The default is 8.
|
||||
*/
|
||||
#ifndef EIGEN_TUNE_TRIANGULAR_PANEL_WIDTH
|
||||
#define EIGEN_TUNE_TRIANGULAR_PANEL_WIDTH 8
|
||||
#endif
|
||||
|
||||
/** Allows to disable some optimizations which might affect the accuracy of the result.
|
||||
* Such optimization are enabled by default, and set EIGEN_FAST_MATH to 0 to disable them.
|
||||
* They currently include:
|
||||
|
||||
14
test/lu.cpp
14
test/lu.cpp
@ -67,7 +67,7 @@ template<typename MatrixType> void lu_non_invertible()
|
||||
|
||||
// The image of the zero matrix should consist of a single (zero) column vector
|
||||
VERIFY((MatrixType::Zero(rows,cols).fullPivLu().image(MatrixType::Zero(rows,cols)).cols() == 1));
|
||||
|
||||
|
||||
MatrixType m1(rows, cols), m3(rows, cols2);
|
||||
CMatrixType m2(cols, cols2);
|
||||
createRandomPIMatrixOfRank(rank, rows, cols, m1);
|
||||
@ -85,9 +85,9 @@ template<typename MatrixType> void lu_non_invertible()
|
||||
RMatrixType l = RMatrixType::Identity(rows,rows);
|
||||
l.block(0,0,rows,std::min(rows,cols)).template triangularView<StrictlyLower>()
|
||||
= lu.matrixLU().block(0,0,rows,std::min(rows,cols));
|
||||
|
||||
|
||||
VERIFY_IS_APPROX(lu.permutationP() * m1 * lu.permutationQ(), l*u);
|
||||
|
||||
|
||||
KernelMatrixType m1kernel = lu.kernel();
|
||||
ImageMatrixType m1image = lu.image(m1);
|
||||
|
||||
@ -192,20 +192,20 @@ void test_lu()
|
||||
|
||||
CALL_SUBTEST_2( (lu_non_invertible<Matrix<double, 4, 6> >()) );
|
||||
CALL_SUBTEST_2( (lu_verify_assert<Matrix<double, 4, 6> >()) );
|
||||
|
||||
|
||||
CALL_SUBTEST_3( lu_non_invertible<MatrixXf>() );
|
||||
CALL_SUBTEST_3( lu_invertible<MatrixXf>() );
|
||||
CALL_SUBTEST_3( lu_verify_assert<MatrixXf>() );
|
||||
|
||||
|
||||
CALL_SUBTEST_4( lu_non_invertible<MatrixXd>() );
|
||||
CALL_SUBTEST_4( lu_invertible<MatrixXd>() );
|
||||
CALL_SUBTEST_4( lu_partial_piv<MatrixXd>() );
|
||||
CALL_SUBTEST_4( lu_verify_assert<MatrixXd>() );
|
||||
|
||||
|
||||
CALL_SUBTEST_5( lu_non_invertible<MatrixXcf>() );
|
||||
CALL_SUBTEST_5( lu_invertible<MatrixXcf>() );
|
||||
CALL_SUBTEST_5( lu_verify_assert<MatrixXcf>() );
|
||||
|
||||
|
||||
CALL_SUBTEST_6( lu_non_invertible<MatrixXcd>() );
|
||||
CALL_SUBTEST_6( lu_invertible<MatrixXcd>() );
|
||||
CALL_SUBTEST_6( lu_partial_piv<MatrixXcd>() );
|
||||
|
||||
Loading…
Reference in New Issue
Block a user