From 7dd81aad74462ba5d3e51e2efdd27e7469b16b90 Mon Sep 17 00:00:00 2001 From: Gael Guennebaud Date: Wed, 3 Mar 2010 18:47:58 +0100 Subject: factorize default performance related settings to a single file included after the architecture specific files such that they can be adapted by each platform. --- Eigen/Core | 4 +-- Eigen/src/Core/DenseStorageBase.h | 2 +- Eigen/src/Core/arch/Default/Settings.h | 65 ++++++++++++++++++++++++++++++++++ Eigen/src/Core/util/Macros.h | 24 ------------- 4 files changed, 67 insertions(+), 28 deletions(-) create mode 100644 Eigen/src/Core/arch/Default/Settings.h (limited to 'Eigen') diff --git a/Eigen/Core b/Eigen/Core index 2908fef2e..01456fe7e 100644 --- a/Eigen/Core +++ b/Eigen/Core @@ -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" diff --git a/Eigen/src/Core/DenseStorageBase.h b/Eigen/src/Core/DenseStorageBase.h index c7f903c7a..dac2142a4 100644 --- a/Eigen/src/Core/DenseStorageBase.h +++ b/Eigen/src/Core/DenseStorageBase.h @@ -601,7 +601,7 @@ struct ei_conservative_resize_like_impl 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); } diff --git a/Eigen/src/Core/arch/Default/Settings.h b/Eigen/src/Core/arch/Default/Settings.h new file mode 100644 index 000000000..1e7cebdba --- /dev/null +++ b/Eigen/src/Core/arch/Default/Settings.h @@ -0,0 +1,65 @@ +// This file is part of Eigen, a lightweight C++ template library +// for linear algebra. +// +// Copyright (C) 2008-2010 Gael Guennebaud +// Copyright (C) 2006-2008 Benoit Jacob +// +// 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 . + + +/* 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 diff --git a/Eigen/src/Core/util/Macros.h b/Eigen/src/Core/util/Macros.h index 7970b3bb0..19420f610 100644 --- a/Eigen/src/Core/util/Macros.h +++ b/Eigen/src/Core/util/Macros.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: -- cgit v1.2.3