mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-01-06 14:14:46 +08:00
Code Issues Packages Projects Releases Wiki Activity
7,490 Commits 11 Branches 89 Tags 168 MiB
dfa80b2060
Commit Graph

4430 Commits

Author SHA1 Message Date
Gael Guennebaud
fe630c9873 Improve numerical accuracy in LLT and triangular solve by using true scalar divisions (instead of x * (1/y)) 2015-10-18 22:15:01 +02:00
Gael Guennebaud
c0adf6e38d Fix perm*sparse return type and nesting, and add several sanity checks for perm*sparse 2015-10-14 10:16:48 +02:00
Gael Guennebaud
527fc4bc86 Fix ambiguous instantiation issues of product_evaluator. 2015-10-14 10:14:47 +02:00
Gael Guennebaud
2598f3987e Add a plain_object_eval<> helper returning a plain object type based on evaluator's Flags, 
and base nested_eval on it.
 2015-10-14 10:12:58 +02:00
Gael Guennebaud
b4c79ee1d3 Update custom setFromTripplets API to allow passing a functor object, and add a collapseDuplicates method to cleanup the API. Also add respective unit test 2015-10-13 11:30:41 +02:00
Calixte Denizet
b9d81c9150 Add a functor to setFromTriplets to handle duplicated entries 2015-10-06 13:29:41 +02:00
Gael Guennebaud
9acfc7c4f3 remove reference to internal method 2015-10-13 10:55:58 +02:00
Gael Guennebaud
a44d91a0b2 extend unit test for SparseMatrix::prune 2015-10-13 10:53:38 +02:00
Gael Guennebaud
252e89b11b bug #1086: replace deprecated UF_long by SuiteSparse_long 2015-10-12 16:20:12 +02:00
Gael Guennebaud
6407e367ee Add missing epxlicit keyword, and fix regression in DynamicSparseMatrix 2015-10-12 09:49:05 +02:00
Gael Guennebaud
63e29e7765 Workaround ICC issue with first_aligned 2015-10-11 22:47:28 +02:00
Gael Guennebaud
6163db814c bug #1085: workaround gcc default ABI issue 2015-10-10 22:38:55 +02:00
Gael Guennebaud
6536b4bad7 Implement temporary-free path for "D.nolias() ?= C + A*B". (I thought it was already implemented) 2015-10-09 15:28:09 +02:00
Gael Guennebaud
a4cc4c1e5e Clarify note in nested_eval for evaluator creating temporaries. 2015-10-09 14:57:51 +02:00
Gael Guennebaud
ae38910693 The evalautor of Solve was missing the EvalBeforeNestingBit flag. 2015-10-09 14:57:19 +02:00
Gael Guennebaud
2632b3446c Improve documentation of TriangularView. 2015-10-09 12:10:58 +02:00
Gael Guennebaud
1429daf850 Add lvalue check for TriangularView::swap, and fix deprecated TriangularView::lazyAssign 2015-10-09 12:10:48 +02:00
Gael Guennebaud
72bd05b6d8 Cleaning in Redux.h 2015-10-09 12:07:42 +02:00
Gael Guennebaud
041e038fef Remove dead code in selfadjoint_matrix_vector_product 2015-10-09 10:42:14 +02:00
Gael Guennebaud
c2d68b984f Optimize a bit complex selfadjoint * vector product. 2015-10-09 10:34:58 +02:00
Gael Guennebaud
1932a24760 Simplify EIGEN_DENSE_PUBLIC_INTERFACE 2015-10-09 10:21:54 +02:00
Gael Guennebaud
186ec1437c Cleanup EIGEN_SPARSE_PUBLIC_INTERFACE, it is now a simple alias to EIGEN_GENERIC_PUBLIC_INTERFACE 2015-10-08 22:06:49 +02:00
Gael Guennebaud
c9718514f5 Fix nesting sub-expression in outer-products 2015-10-08 21:41:53 +02:00
Gael Guennebaud
4140ee039d Fix propagation of AssumeAliasing for expression as: "scalar * (A*B)" 2015-10-08 21:41:27 +02:00
Gael Guennebaud
d866279364 Clean a bit the implementation of inverse permutations 2015-10-08 18:36:39 +02:00
Gael Guennebaud
8d00a953af Fix a nesting issue in some matrix-vector cases. 2015-10-08 17:36:57 +02:00
Gael Guennebaud
dd934ad057 Re-enable vectorization of LinSpaced, plus some cleaning 2015-10-08 17:27:01 +02:00
Gael Guennebaud
f6f6f50272 Clean evaluator<EvalToTemp> 2015-10-08 16:34:33 +02:00
Gael Guennebaud
67bfba07fd Fix some CUDA issues 2015-10-08 16:30:28 +02:00
Gael Guennebaud
aa6b1aebf3 Properly implement PartialReduxExpr on top of evaluators, and fix multiple evaluation of nested expression 2015-10-08 15:57:05 +02:00
Gael Guennebaud
5cc7251188 Some cleaning in evaluators 2015-10-08 15:22:04 +02:00
Gael Guennebaud
e30bc89190 Add missing include of std vector 2015-10-08 15:20:50 +02:00
Gael Guennebaud
1b148d9e2e Move IncompleteCholesky to official modules 2015-10-08 11:32:46 +02:00
Gael Guennebaud
64242b8bf3 Doc: add link to doc of sparse solver concept 2015-10-08 10:50:39 +02:00
Gael Guennebaud
f047ecc36a _mm_hadd_epi32 is for SSSE3 only (and not SSE3) 2015-10-07 15:48:35 +02:00
Gael Guennebaud
aba1eda71e Help clang to inline some functions, thus fixing some regressions 2015-10-07 15:44:12 +02:00
Gael Guennebaud
41cc1f9033 Remove debuging prod() and lazyprod() function, plus some cleaning in noalias assignment 2015-10-07 15:41:22 +02:00
Gael Guennebaud
69a7897e72 Fix storage index type in empty permutations 2015-10-06 17:21:24 +02:00
Gael Guennebaud
26cde4db3c Define Permutation*<>::Scalar to 'void', re-enable scalar type compatibility check in assignment while relaxing this test for void types. 2015-10-06 17:18:06 +02:00
Gael Guennebaud
fb51bab272 Some cleaning 2015-10-06 17:14:56 +02:00
Gael Guennebaud
2c676ddb40 Handle various TODOs in SSE vectorization (remove splitted storeu, enable SSE3 integer vectorization, plus minor tweaks) 2015-10-06 15:43:27 +02:00
Gael Guennebaud
2d287a4898 Fix Ref<SparseMatrix> for Transpose<SparseVector> 2015-10-06 15:09:04 +02:00
Gael Guennebaud
f25bdc707f Optimise assignment into a Block<SparseMatrix> by using Ref and avoiding useless updates in non-compressed mode. This make row-by-row filling of a row-major sparse matrix very efficient. 2015-10-06 11:59:08 +02:00
Gael Guennebaud
945b80c83e Optimize Ref<SparseMatrix> by removing useless default initialisation of SparseMapBase and SparseMatrix 2015-10-06 11:57:03 +02:00
Gael Guennebaud
9a070638de Enable to view a SparseVector as a Ref<SparseMatrix> 2015-10-06 11:53:19 +02:00
Gael Guennebaud
1b43860bc1 Make SparseVector derive from SparseCompressedBase, thus improving compatibility between sparse vectors and matrices 2015-10-06 11:41:03 +02:00
Gael Guennebaud
6100d1ae64 Improve counting of sparse temporaries 2015-10-06 11:32:02 +02:00
Gael Guennebaud
deb261f64b Make abs2 compatible with custom complex types 2015-10-02 10:33:25 +02:00
Doug Kwan
5c9ee73eb9 Implement plog and pexp for AltiVec. 2015-07-30 11:12:42 -07:00
Gael Guennebaud
5a1cc5d24c bug #1053: fix SuplerLU::solve with EIGEN_DEFAULT_TO_ROW_MAJOR 2015-09-03 11:25:36 +02:00
Gael Guennebaud
2795ffd6a0 Fix Index vs StorageIndex naming convention 2015-09-03 11:18:27 +02:00
Gael Guennebaud
ef2b54f422 Fix AMD ordering when a column has only one off-diagonal non-zero (also fix bug #1045) 2015-09-03 11:04:06 +02:00
Christoph Hertzberg
5ad7981f73 Use full packet size for Dynamic-sized objects (otherwise, the unalignedcount unit test fails with AVX enabled) 2015-09-02 22:51:43 +02:00
Gael Guennebaud
aa768add0b Since there is no reason for evaluators to be nested by reference, let's remove the evaluator<>::nestedType indirection. 2015-09-02 22:10:39 +02:00
Gael Guennebaud
f8976fdbe0 Make evaluators non-copyable. This guarantee that evaluators storing temporaries do not introduce unwanted copy overhead. 2015-09-02 21:39:49 +02:00
Gael Guennebaud
92b9f0e102 Cleaning pass on evaluators: remove the useless and error prone evaluator<>::type indirection. 2015-09-02 21:38:40 +02:00
Gael Guennebaud
6b99afa5ae Fix LSCG::solve with a sparse destination. 2015-09-02 15:34:03 +02:00
Gael Guennebaud
b5ad3d2cf7 Remove deprecated Flagged expression. 2015-09-02 14:53:50 +02:00
Gael Guennebaud
be5e2ecc21 bug #505: add more examples of bad and correct usages of auto and eval(). 2015-09-02 13:04:30 +02:00
Gael Guennebaud
a75616887e bug #1057: fix a declaration missmatch with MSVC 2015-09-02 09:31:32 +02:00
Gael Guennebaud
280f93ff65 Fix FullPivLU::image documentation 2015-09-02 09:19:27 +02:00
Gael Guennebaud
6059188f9d Simplify implementation of the evaluation's iterator of Sparse*Diagonal products to help the compiler to generate better code. 2015-09-01 22:34:30 +02:00
Gael Guennebaud
0b2412df50 Remove duplicated temporary in Sparse to Sparse assignment 2015-09-01 22:31:30 +02:00
Gael Guennebaud
9001f4a46b Add missing specialization of evaluator of sub-sparse-matrices that can be seen as a SparseCompressedBase. This changeset enable faster iterator for such expressions. 2015-09-01 22:29:17 +02:00
Benoit Steiner
2ed1495eec nvcc doesn't support std::min or std::max on GPU. Use our own custom implementation instead 2015-08-27 16:59:55 -07:00
Sergiu Dotenco
d4c24eb016 fixed Quaternion identity initialization for non-implicitly convertible types 2015-08-20 20:55:37 +02:00
Christoph Hertzberg
78358a7241 Fixed broken commit a09cfe650f 
. Missing } and unprotected min/max calls and definitions.
 2015-08-22 15:03:16 +02:00
Benoit Steiner
a09cfe650f std::numeric_limits doesn't work reliably on CUDA devices. Use our own definition of numeric_limit<T>::max() and numeric_limit<T>::min() instead of the stl ones. 2015-08-21 16:01:40 -07:00
Christoph Hertzberg
e5c78d85c8 bug #1043: Avoid integer conversion sign warning 2015-08-19 21:50:21 +02:00
Christoph Hertzberg
8097d8d028 surpress some warnings 2015-08-17 21:50:52 +02:00
Christoph Hertzberg
d2e0927127 Define EIGEN_MAX_STATIC_ALIGN_BYTES to 0 for architectures that don't require stack alignment 2015-08-17 16:44:52 +02:00
Gael Guennebaud
dc2c103b3b merge 2015-08-16 14:22:02 +02:00
Christoph Hertzberg
d6a4805fdf Protect further isnan/isfinite/isinf calls 2015-08-16 14:00:02 +02:00
Christoph Hertzberg
61e0977e10 Protect all calls to isnan, isinf and isfinite with parentheses. 2015-08-14 17:32:34 +02:00
Christoph Hertzberg
712e2fed17 bug #829: Introduce macro EIGEN_HAS_CXX11_CONTAINERS and do not specialize std-containers if it is enabled. 2015-08-14 16:09:48 +02:00
Gael Guennebaud
23aab82c0c merge 2015-08-09 21:24:20 +02:00
Christoph Hertzberg
cac6b23033 bug #1053: SparseLU failed with EIGEN_DEFAULT_TO_ROW_MAJOR 2015-08-07 23:10:56 +02:00
Gael Guennebaud
febcce34f1 Enable vectorization with half-packets 2015-08-07 20:05:31 +02:00
Gael Guennebaud
6245591349 Fix prototype of plset and generalize linspace functor. 2015-08-07 19:27:59 +02:00
Gael Guennebaud
60e4260d0d Some functors were not generic wrt packet-type. 2015-08-07 17:41:39 +02:00
Gael Guennebaud
e68c7b8368 Include SSE packetmath when AVX is enabled, and enable AVX's sine function only in fast-math mode (as SSE) 2015-08-07 17:40:39 +02:00
Gael Guennebaud
65bfa5fce7 Allow to use arbitrary packet-types during evaluation. 
This is implemented by adding a PacketType template parameter to packet and writePacket members of evaluator<>.
 2015-08-07 12:01:39 +02:00
Gael Guennebaud
3602926ed5 Mark ALignedBit as deprecated. 2015-08-07 10:45:02 +02:00
Gael Guennebaud
ce57dbd937 Let unpacket_traits<> exposes the required alignment and make use of it everywhere 2015-08-07 10:44:01 +02:00
Gael Guennebaud
2afdef6a54 Generalize first_aligned to take the requested alignment as a template parameter, and add a first_default_aligned variante calling first_aligned with the requirement of the largest packet for the given scalar type. 2015-08-06 17:52:01 +02:00
Gael Guennebaud
1f5024332e First part of a big refactoring of alignment control to enable the handling of arbitrarily aligned buffers. It includes: 
- AlignedBit flag is deprecated. Alignment is now specified by the evaluator through the 'Alignment' enum, e.g., evaluator<Xpr>::Alignment. Its value is in Bytes.
 - Add several enums to specify alignment: Aligned8, Aligned16, Aligned32, Aligned64, Aligned128. AlignedMax corresponds to EIGEN_MAX_ALIGN_BYTES. Such enums are used to define the above Alignment value, and as the 'Options' template parameter of Map<> and Ref<>.
 - The Aligned enum is now deprecated. It is now an alias for Aligned16.
 - Currently, traits<Matrix<>>, traits<Array<>>, traits<Ref<>>, traits<Map<>>, and traits<Block<>> also expose the Alignment enum.
 2015-08-06 15:31:07 +02:00
Gael Guennebaud
65186ef18d Fix logic in compute_default_alignment, extend it to Dynamic size, and move it to XprHelper.h file. 2015-08-06 14:07:59 +02:00
Gael Guennebaud
becd89df29 Enable runtime stack alignment in gemm_blocking_space. 2015-08-06 14:00:26 +02:00
Gael Guennebaud
d4f5efc51a Add a EIGEN_DEFAULT_ALIGN_BYTES macro defining default alignment for alloca and aligned_malloc. 
It is defined as the max of EIGEN_IDEAL_MAX_ALIGN_BYTES and EIGEN_MAX_ALIGN_BYTES
 2015-08-06 13:56:53 +02:00
Gael Guennebaud
506964fc29 Propagate precondition info to the iterative solver. 2015-08-04 16:13:34 +02:00
Gael Guennebaud
f7d5b9323d typo 2015-07-29 11:08:49 +02:00
Gael Guennebaud
175ed636ea bug #973: update macro-level control of alignement by introducing user-controllable EIGEN_MAX_ALIGN_BYTES and EIGEN_MAX_STATIC_ALIGN_BYTES macros. This changeset also removes EIGEN_ALIGN (replaced by EIGEN_MAX_ALIGN_BYTES>0), EIGEN_ALIGN_STATICALLY (replaced by EIGEN_MAX_STATIC_ALIGN_BYTES>0), EIGEN_USER_ALIGN*, EIGEN_ALIGN_DEFAULT (replaced by EIGEN_ALIGN_MAX). 2015-07-29 10:22:25 +02:00
Gael Guennebaud
76874b128e bug #1047: document the structure layout of class Matrix 2015-07-29 10:21:28 +02:00
Gael Guennebaud
41e1f3498c bug #1048: fix unused variable warning 2015-07-28 22:59:50 +02:00
Christoph Hertzberg
a44d022caf bug #792: SparseLU::factorize failed for structurally rank deficient matrices 2015-07-26 20:30:30 +02:00
Benoit Steiner
6d6e6d0b88 Define EIGEN_VECTORIZE_AVX2 and EIGEN_VECTORIZE_FMA when the corresponding instructions can be used by the compiler 2015-07-22 18:22:16 -07:00
Gael Guennebaud
3b0ad02c10 Remove wrongly pushed debugging statements 2015-07-22 14:33:57 +02:00
Jonas Adler
815fa0dbf6 Fixed some compiler bugs in NVCC, now compiles with CUDA. 
(chtz: Manually joined sevaral commits to keep the history clean)
 2015-07-22 12:29:18 +02:00
Benoit Steiner
d259b719d1 Made sure that the use const expressions are not enabled when compiling with nvcc even when gcc 4.9 is used as the host compiler. 2015-07-21 17:35:58 -07:00
Gael Guennebaud
586d10f7e0 Fix compilation of tri(sparse) * dense with OpenMP 2015-07-21 22:52:21 +02:00
Gael Guennebaud
6544b49e59 Generalize pow(x,e) such that x and e can be a different expression type or a scalar for either x or e. Add x.pow(e) with e an array expression. 2015-07-20 13:57:55 +02:00
Gael Guennebaud
c11971de37 Fix compilation of isnan(complex) 2015-07-20 12:56:01 +02:00
Gael Guennebaud
88e352adac Add support for replicate in CUDA 2015-07-20 10:53:03 +02:00
Benoit Steiner
513e357b48 Added support for prefetching on cuda devices 2015-07-17 15:35:16 -07:00
Benoit Steiner
06a22ca5bd Added support for sigmoid function to the tensor module 2015-07-17 09:29:00 -07:00
Benoit Steiner
979b73cebf Fixed a typo in Macro.h 2015-07-16 14:17:50 -07:00
Gael Guennebaud
f5aa640862 Clean some previous changes and more cuda fixes 2015-07-15 10:57:55 +02:00
Nicolas Mellado
7cecd39a84 Merged eigen/eigen into default 2015-07-15 10:15:54 +02:00
Nicolas Mellado
592ee2a4b4 Add missing EIGEN_DEVICE_FUNC 2015-07-15 10:14:52 +02:00
Christoph Hertzberg
ea87561564 bug #1039: Redefining EIGEN_DEFAULT_DENSE_INDEX_TYPE may lead to errors 2015-07-13 16:08:25 +02:00
Nicolas Mellado
dbb3e2cf8a Cleaning 2015-07-11 18:15:31 +00:00
Nicolas Mellado
0d09845562 Revert files to remove EIGEN_USING_NUMEXT_MATH 2015-07-11 20:11:36 +02:00
Nicolas Mellado
20b96025fd Replace double constants by Scalar constants 2015-07-11 20:02:30 +02:00
Nicolas Mellado
1dd6a329e8 Cuda compatibility: remove explicit call to std math functions 2015-07-11 19:40:15 +02:00
Nicolas Mellado
bc40eb745d Merged eigen/eigen into default 2015-07-11 19:33:43 +02:00
Benoit Steiner
6de6fa9483 Use NumTraits<T>::RequireInitialization instead of internal::is_arithmetic<T>::value to check whether it's possible to bypass the type constructor in the tensor code. 2015-07-07 15:23:56 -07:00
Benoit Steiner
7b7df7b6b8 Updated internal::is_arithmetic::value to be true for complex numbers 2015-07-07 12:57:35 -07:00
Gael Guennebaud
7fa6fe8d8c typo 2015-07-07 17:47:24 +02:00
Gael Guennebaud
fa17358c4b Rotation2D: fix slerp to take the shortest path, and add convenient method to get the angle in [-pi,pi] or [0,pi] 2015-07-07 17:27:12 +02:00
Nicolas Mellado
5359e5cdb2 Protect against compilation errors with nvcc and numext/complex. 
Disable functions explicitely involving std::complex when compiling with nvcc.
Improve code compatilibity using the new macro EIGEN_USING_NUMEXT_MATH (same spirit than EIGEN_USING_STD_MATH but for numext functions)
 2015-07-06 20:55:01 +02:00
Gael Guennebaud
c2019dfeb3 Merged in Emie/eigen (pull request PR-121) 
typo correction in mathFunction
 2015-07-06 16:48:54 +02:00
Emilie Guy
ea7113dd0c typo correction in mathFunction 2015-07-06 14:31:08 +02:00
Nicolas Mellado
9115896590 Merged eigen/eigen into default 2015-07-03 00:41:11 +02:00
Benoit Steiner
95ef94f1ee Fixed a typo in the patch 2015-07-02 07:06:55 +00:00
Benoit Steiner
44eedd8915 Marked the cast functions as EIGEN_DEVICE_FUNC to ensure that we can run casting on GPUs 2015-06-30 15:48:55 -07:00
Gael Guennebaud
c911fc8dee split compiler intensive bdcsvd_1 unit test 2015-06-26 16:14:23 +02:00
Gael Guennebaud
98ff17eb9e Add special path for matrix<complex>/real. 
This also fixes underflow issues when scaling complex matrices through complex/complex operator.
 2015-06-26 16:08:15 +02:00
Gael Guennebaud
e102ddbf1f bug #1026: fix infinite loop for an empty input 2015-06-26 14:02:52 +02:00
Gael Guennebaud
555b9c6843 Doc: explain perf and multithreading issues in sparse iterative solvers 2015-06-26 10:49:40 +02:00
Gael Guennebaud
53b930887d Enable OpenMP parallelization of row-major-sparse * dense products. 
I observed significant speed-up of the CG solver.
 2015-06-26 10:32:34 +02:00
Gael Guennebaud
7f824dd613 Optimize CG to enable faster spare row-major * dense vector products when the input matrix is complete (Upper|Lower) but column major. 2015-06-25 17:17:38 +02:00
Gael Guennebaud
33e699c9fe Remove redundant accessors in Reverse 2015-06-25 14:14:59 +02:00
Gael Guennebaud
973b0a90db Clarify documentation of the tolerance and error returned in iterative solvers 2015-06-25 13:51:13 +02:00
Gael Guennebaud
b4ab72678c bug #1000: MSVC 2013 does need the operator= workaround 2015-06-25 09:45:22 +02:00
Gael Guennebaud
788941d3b1 Workaround MSVC ambiguous instanciation 2015-06-24 23:35:17 +02:00
Gael Guennebaud
4c8cd13b35 Add explicit ctor for diagonal to sparse conversion 2015-06-24 18:11:06 +02:00
Gael Guennebaud
c38c195321 Document how cross behaves on complex numbers 2015-06-24 18:02:33 +02:00
Gael Guennebaud
62f21e2d11 Add support for sparse = diagonal 2015-06-24 17:55:00 +02:00
Gael Guennebaud
763c833637 Make SparseSelfAdjointView, twists, and SparseQR more evaluator friendly 2015-06-24 17:54:09 +02:00
Gael Guennebaud
36643eec0c Add a call_assignment_no_alias_no_transpose shortcut 2015-06-24 17:50:43 +02:00
Gael Guennebaud
02db7c9bc6 Inherit operator+= and -= with 'using' kkeyword 2015-06-24 17:49:20 +02:00
Gael Guennebaud
95e19be381 Fix compilation of MKL Pardiso support 2015-06-24 14:53:43 +02:00
Gael Guennebaud
b0d08869a9 Fix underflow in 3x3 tridiagonalization 2015-06-23 14:54:31 +02:00
Gael Guennebaud
71523a2e25 Fix a warning with icc 2015-06-23 14:20:20 +02:00
Gael Guennebaud
d9778f3391 Enable VML's pow wrapper on windows (the previous wrapper used the Fortran interface) 2015-06-23 14:04:50 +02:00
Gael Guennebaud
5f9630d7f9 bug #923: update support for Intel's VML wrt new evaluation mechanisms 2015-06-23 14:03:25 +02:00
Gael Guennebaud
793e4c6d77 bug #923: fix EIGEN_USE_BLAS mode 2015-06-23 11:13:24 +02:00
Gael Guennebaud
307c4fc292 Workaround missalignment produced by first_aligned for PacketSize==1 and size==1 2015-06-23 10:10:17 +02:00
Gael Guennebaud
bb3a9b4941 Use Ref<> to bypass forceAlignmentIf 2015-06-22 17:48:28 +02:00
Gael Guennebaud
476beed7f8 bug #1017: apply Christoph's patch preventing underflows in makeHouseholder 2015-06-22 16:51:45 +02:00
Gael Guennebaud
0848ba0a6e Fix return nullary return types: it must be based on the PlainObject type instead of the expression type. 2015-06-22 10:52:08 +02:00
Nicolas Mellado
ad5fdc7ddd Fix double to Scalar unwanted promotions 2015-06-21 20:21:23 +02:00
Gael Guennebaud
40821876ea Fix regression on CompressedStorage::operator= 2015-06-20 13:59:13 +02:00
Gael Guennebaud
84aaef93ba Merged in vanhoucke/eigen_vanhoucke (pull request PR-118) 
Fix two small undefined behaviors caught by static analysis.
 2015-06-20 13:56:48 +02:00
Gael Guennebaud
6b33b29f00 Get rid of must_nest_by_value 2015-06-19 18:12:40 +02:00
Gael Guennebaud
846b227bb7 Get rid of class internal::nested<> (still have to updated Tensor module) 2015-06-19 17:56:39 +02:00
vanhoucke
368ea23406 Fix undefined behavior. When resizing a default-constructed SparseArray, we end up calling memcpy(ptr, 0, 0), which is technically UB and gets caught by static analysis. 2015-06-19 15:53:30 +00:00
Gael Guennebaud
386d9e5ebd Fix usage of nested versus nested_eval 2015-06-19 17:42:27 +02:00
Gael Guennebaud
a5a7b68b76 Fix ambiguous instanciation using clean class-level SFINAE in product_evaluator 2015-06-19 17:25:13 +02:00
Gael Guennebaud
6fc5438205 Remove a few deprecated internal expressions 2015-06-19 17:06:12 +02:00
Gael Guennebaud
5c84dd5665 Fix permutation/transposiitons products wrt nested_eval 2015-06-19 16:37:04 +02:00
Gael Guennebaud
0c8b0e007b Introduce a AliasFreeProduct option for Permutations and Transpositions 2015-06-19 15:38:19 +02:00
Gael Guennebaud
3f6aa4cd5d Remove useless specializations of evaluator_traits 2015-06-19 14:18:29 +02:00
Gael Guennebaud
4a8888dfbc Improbe compatibility of Transpositions and evaluators 2015-06-19 14:10:44 +02:00
Gael Guennebaud
3af4c6c1c9 Make Transpositions use evaluators 2015-06-19 11:50:24 +02:00
Gael Guennebaud
82b6ac0864 Enforce eigenvectors to be column-major (for performance reasons) 2015-06-19 11:25:46 +02:00
Gael Guennebaud
fad36cc814 Clean implementation of permutation * matrix products. 2015-06-19 10:51:57 +02:00
Gael Guennebaud
06036d8bb1 Fix compilation of BDCSVD with DEFAULT_TO_ROWMAJOR 2015-06-19 10:37:25 +02:00
Gael Guennebaud
d2db15016b Fix storage order computation in traits<Product> 2015-06-19 10:36:38 +02:00
Gael Guennebaud
7baa1ba03e Remove the usage of result_of for DenseBase::redux as discussed in bug #1006 2015-06-15 22:40:18 +02:00
Gael Guennebaud
97cbe28829 Remove support of std::binder* in C++11 2015-06-15 15:34:05 +02:00
Gael Guennebaud
972a535288 Remove aligned-on-scalar assert and fallback to non vectorized path at runtime (first_aligned already had this runtime guard) 2015-06-14 15:04:07 +02:00
Gael Guennebaud
a546be56e0 typo 2015-06-15 15:08:51 +02:00
Gael Guennebaud
321a2cbe3d Add missing forward declaration of AlignedBox 2015-06-15 15:01:20 +02:00
Gael Guennebaud
91b64a9c65 Relax aligned-on-scalar assert as in the 3.2 branch 2015-06-12 11:25:57 +02:00
Gael Guennebaud
84d103bee8 Enable C++11 math function in a more conservative manner. 2015-06-11 21:45:02 +02:00
Gael Guennebaud
d93ba137f2 Introduce EIGEN_PI, get rid of M_PI and acos(-1.0) 2015-06-10 17:12:10 +02:00
Gael Guennebaud
9756c7fb4d Make more use of EIGEN_HAS_C99_MATH 2015-06-10 16:26:55 +02:00
Gael Guennebaud
93a62265dc fix isinf(complex(inf,NaN)) to return false. 2015-06-10 16:19:10 +02:00
Gael Guennebaud
b0d5aaafcc Rename free functions isFinite, isInf, isNaN to be compatible with c++11 2015-06-10 16:17:09 +02:00
Gael Guennebaud
25a98be948 bug #80: merge with d_hood branch on adding more coefficient-wise unary array functors 2015-06-10 15:52:05 +02:00
Gael Guennebaud
192bce2795 bug #890, add a more general routine to check that two dense object reference to the same data 2015-06-10 10:09:04 +02:00
Gael Guennebaud
0b2cbb2bdc bug #897: make umfpack support use Ref<> 2015-06-09 23:30:06 +02:00
Gael Guennebaud
feaf76c001 bug #910: add a StandardCompressedFormat option to Ref<SparseMatrix> to enforce standard compressed storage format. 
If the input is not compressed, then this trigger a copy for a const Ref, and a runtime assert for non-const Ref.
 2015-06-09 23:11:24 +02:00
Gael Guennebaud
f899aeb301 bug #650: fix sparse * dense wrt noalias and compound assignment 2015-06-09 18:33:24 +02:00
Gael Guennebaud
785b9c0127 bug #1003: assert in MapBase if the provided pointer is not aligned on scalar while it is expected to be. Also add a EIGEN_ALIGN8 macro. 2015-06-09 17:42:09 +02:00
Gael Guennebaud
4aba24a1b2 Clean argument names of some functions 2015-06-09 13:32:12 +02:00
Gael Guennebaud
302cf8ffe2 Add missing documentation for TriangularViewImpl<MatrixType,Mode,Sparse> 2015-06-09 11:40:07 +02:00
Gael Guennebaud
3a4299b245 bug #872: remove usage of deprecated bind1st. 2015-06-09 10:52:04 +02:00
Gael Guennebaud
9a2447b0c9 Fix shadow warnings triggered by clang 2015-06-09 09:11:12 +02:00
Gael Guennebaud
cd8b996f99 Extend unit test and documentation of SelfAdjointEigenSolver::computeDirect 2015-06-08 16:16:42 +02:00
Gael Guennebaud
8f031a3cee bug #997: add missing evaluators for m.lazyProduct(v.homogeneous()) 2015-06-08 15:43:41 +02:00
Gael Guennebaud
274b1f5d7e Fix homogeneous() for 1x1 matrix: in this case, homogeneous follows the storage order guaranteeing that v.transpose().homogeneous() == v.homogeneous().transpose() 2015-06-08 15:40:51 +02:00
Gael Guennebaud
cbe3a1a83e Add missing accessors for 1D index based access to Replicate<> expressions. 2015-06-08 15:39:09 +02:00
Gael Guennebaud
a7ae628c9f bug #1005: fix regression regarding sparse coeff-wise binary operator that did not trigger a static assertion for mismatched storage 2015-06-08 10:14:08 +02:00
Gael Guennebaud
0a9b5d1396 bug #705: fix handling of Lapack potrf return code 2015-06-05 15:59:13 +02:00
Gael Guennebaud
d0b7b5cb55 minor documentation fixes 2015-06-05 14:40:07 +02:00
Gael Guennebaud
56d4ef7ad6 BiCGSTAB: set default guess to 0, and improve restart mechanism by recomputing the accurate residual. 2015-06-05 14:37:57 +02:00
Gael Guennebaud
d457734a19 Avoid calling smart_copy with null pointers. 2015-05-25 22:30:56 +02:00
Benoit Jacob
051d5325cc Abandon blocking size lookup table approach. Not performing as well in real world as in microbenchmark. 2015-05-19 11:03:59 -04:00
Christoph Hertzberg
ebea530782 bug #1014: More stable direct computation of eigenvalues and -vectors for 3x3 matrices 2015-05-17 21:54:32 +02:00
Benoit Jacob
c88e1abaf3 also uninitialized here, see previous cset 2015-05-15 11:34:57 -04:00
Benoit Jacob
807793ec3b Fix uninitialized var warning. The compiler was clearing the register anyway, so this does not change resulting code 2015-05-15 11:15:53 -04:00
Pete Warden
140f85bb99 Check for the macro __ARM_NEON__ (with two underscores at the end) as well as __ARM_NEON. The second macro is correct according to the ARM language extensions specification, but historically the first one has been more common. Some older compilers (e.g. gcc v4.6 on a Beaglebone Black) only define the first, so without this patch NEON isn't enabled. 2015-05-12 16:03:43 -07:00
Gael Guennebaud
ef81730625 Ignore denormal numbers in selfadjoint eigensolver. 2015-05-12 18:38:43 +02:00
Christoph Hertzberg
494fa991c3 bug #872: Avoid deprecated binder1st/binder2nd usage by providing custom functors for comparison operators 2015-05-07 17:28:40 +02:00
Gael Guennebaud
4a936974a5 bug #1013: fix 2x2 direct eigensolver for identical eiegnvalues 2015-05-07 15:55:12 +02:00
Gael Guennebaud
ebf8ca4fa8 Fix bug #1010: m_isInitialized was improperly updated 2015-05-07 14:20:42 +02:00
Konstantinos Margaritis
dd698e6680 Merged in doug_kwan/eigen (pull request PR-103) 
Fix bug in pdiv<Packet1cd> which swaps 32-bit halves of a pair of
 2015-05-05 20:50:14 +03:00
Benoit Steiner
1dded10cb7 Added a double-precision implementation of the exp() function for AVX. 2015-05-04 10:42:51 -07:00
Christoph Hertzberg
28a4c92cbf bug #998: Started fixing doxygen warnings 2015-05-01 22:10:41 +02:00
Christoph Hertzberg
173b34e9ab bug #999: clarify that behavior of empty AlignedBoxes is undefined, and further improvements in documentation 2015-04-30 19:30:36 +02:00
Gael Guennebaud
de18cd413d Disable posix_memalign on Solaris and SunOS, and allows to by-pass built-in posix_memalign detection rules. 2015-04-24 11:26:51 +02:00
Gael Guennebaud
40258078c6 bug #360: add value_type typedef to DenseBase/SparseMatrixBase 2015-04-24 09:44:24 +02:00
Christoph Hertzberg
c460af414e Fix bug #1000: Manually inherit assignment operators for MSVC 2013 and later (as required by the standard). 2015-04-23 13:39:03 +02:00
Gael Guennebaud
dbd12b4cda Make sure that BlockImpl<const SparseMatrix> ctor is called with the right type 2015-04-21 10:15:36 +02:00
Gael Guennebaud
d6a8b43b39 Fix typo in the definition of EIGEN_COMP_GNUC_STRICT 2015-04-21 10:12:38 +02:00
Deanna Hood
e5048b5501 Use std::isfinite when available 2015-04-20 14:59:57 -04:00
Deanna Hood
249c48ba00 Incorporate C++11 check into EIGEN_HAS_C99_MATH macro 2015-04-20 14:57:04 -04:00
Deanna Hood
0250f4a9f2 Merged default into unary-array-cwise-functors 2015-04-20 14:01:35 -04:00
Deanna Hood
0339502a4f Only use std::isnan and std::isinf if they are available 2015-04-20 13:14:06 -04:00
Gael Guennebaud
fc2d5b86ce simplify previous changeset: sub-expressions are nested by value 2015-04-18 22:50:16 +02:00
Gael Guennebaud
5a3c48e3c6 bug #942: fix dangling references in evaluator of diagonal * sparse products. 2015-04-18 22:43:27 +02:00
Christoph Hertzberg
4f126b862d Add internal assertions to purely fixed-size DenseStorage, mark optional variables always as unused 2015-04-17 11:36:21 +02:00
Christoph Hertzberg
9d7843d0d0 Add internal assertions to DenseStorage constructor 2015-04-16 15:47:06 +02:00
Christoph Hertzberg
3be9f5c4d7 Constructing a Matrix/Array with implicit transpose could lead to memory leaks. 
Also reduced code duplication for Matrix/Array constructors
 2015-04-16 13:25:20 +02:00
Gael Guennebaud
e0cff9ae0d Fix bug #996: fix comparisons to 0 instead of Scalar(0) 2015-04-15 14:48:53 +02:00
Gael Guennebaud
5dbe758dc3 Backed out changeset 04c8c5d9ef 2015-04-15 14:47:08 +02:00
Gael Guennebaud
04c8c5d9ef Fix bug #996: fix comparisons to 0 instead of Scalar(0) 2015-04-15 14:44:57 +02:00
Benoit Steiner
0f82399fe9 Pulled latest changes from trunk 2015-04-14 19:13:34 -07:00
Christoph Hertzberg
761691f18d Make conversion from 0 to Scalar explicit (issue reported by Brad Bell) 2015-04-13 17:15:00 +02:00
Benoit Steiner
5401fbcc50 Improved the blocking strategy to speedup multithreaded tensor contractions. 2015-04-09 16:44:10 -07:00
Deanna Hood
085aa8e601 Don't use M_PI since it's only guaranteed to be defined in Eigen/Geometry 2015-04-08 13:59:18 -05:00
Gael Guennebaud
0eb220c00d add a note on bug #992 2015-04-08 09:25:34 +02:00
Benoit Jacob
d7f51feb07 bug #992: don't select a 3p GEMM path with non-vectorizable scalar types, this hits unsupported paths in symm/triangular products code 2015-04-07 15:13:55 -04:00
Benoit Steiner
7c18ab921c Pulled latest updates from trunk 2015-04-04 20:07:04 -07:00
Gael Guennebaud
15b5adb327 Fix regression in DynamicSparseMatrix and SuperLUSupport wrt recent change on nonZeros/nonZerosEstimate 2015-04-02 22:21:41 +02:00
Benoit Steiner
74e558cfa8 Pulled latest updates from trunk 2015-04-01 23:24:11 -07:00
Gael Guennebaud
5861cfb55e Remove unused GenericSparseBlockInnerIteratorImpl code. 2015-04-01 22:29:29 +02:00
Gael Guennebaud
3105986e71 bug #875: remove broken SparseMatrixBase::nonZeros and introduce a nonZerosEstimate() method to sparse evaluators for internal uses. 
Factorize some code in SparseCompressedBase.
 2015-04-01 22:27:34 +02:00
Gael Guennebaud
39dcd01b0a bug #973: enable alignment of multiples of half-packet size (e.g., Vector6d with AVX) 2015-04-01 13:55:09 +02:00
Gael Guennebaud
8481dc21ea bug #986: add support for coefficient-based product with 0 depth. 2015-04-01 13:15:23 +02:00
Gael Guennebaud
79b4e6acaf Fix bug #987: wrong alignement guess in diagonal product. 2015-03-31 23:35:12 +02:00
Gael Guennebaud
3c38589984 Remove most of the dynamic memory allocations that occured in D&C SVD. Still remains the calls to JacobiSVD and UpperBidiagonalization. 2015-03-31 22:54:47 +02:00
Gael Guennebaud
8313fb7df7 Add row/column-wise reverseInPlace feature. 2015-03-31 21:35:53 +02:00
Gael Guennebaud
dfb674a25e Make reverseInPlace really work in-place. 2015-03-31 20:17:10 +02:00
Gael Guennebaud
20d030f207 Fix vectorization of swap for non trivial expressions 2015-03-31 20:16:02 +02:00
Benoit Jacob
73cdeae1d3 Only use blocking sizes LUTs for single-thread products for now 2015-03-31 11:17:23 -04:00
Benoit Jacob
0cbd5ae3cb Correctly detect Android with ndk_build 2015-03-31 11:17:21 -04:00
Gael Guennebaud
ae01c05e18 Fix computeProductBlockingSizes with m==0, and add respective unit test. 2015-03-31 15:19:57 +02:00
Gael Guennebaud
bd76d837e6 Fix sign of SuperLU::determinant 2015-03-31 14:57:32 +02:00
Gael Guennebaud
35d3053d55 Fix regression introduced in 3b169d792d 2015-03-31 09:23:53 +02:00
Christoph Hertzberg
3b169d792d Suppress unused variable warning 2015-03-31 00:49:08 +02:00
Christoph Hertzberg
1efae98fee bug #985: RealQZ failed when either matrix had zero rows or columns (report and patch by Ben Goodrich) 
Also added a regression test
 2015-03-30 23:56:20 +02:00
Benoit Steiner
35722fa022 Made the index type a template parameter of the tensor class instead of encoding it in the options. 2015-03-30 14:55:54 -07:00
Christoph Hertzberg
58af8bf90c bug #982: Make sure numext::maxi and numext::mini are called correctly, in case Scalar expressions return expression templates. 2015-03-30 16:47:22 +02:00
Gael Guennebaud
2adbf6b8ca fix stupid warning with old GCC 2015-03-28 22:34:54 +01:00
Gael Guennebaud
41e20248f8 merge 2015-03-28 14:43:35 +01:00
Christoph Hertzberg
09a5361d1b bug #983: Pass Vector3 by const reference and not by value 2015-03-28 12:36:24 +01:00
Gael Guennebaud
eb7e4c2b9c Pass Vector3 type by reference 2015-03-27 12:11:24 +01:00
Gael Guennebaud
79cb875249 merge 2015-03-27 10:56:04 +01:00
Gael Guennebaud
1b8cc9af43 Slight numerical stability improvement in 2x2 svd 2015-03-27 10:55:00 +01:00
Gael Guennebaud
3d59ae0203 Fix hypot(0,0). 2015-03-27 09:59:24 +01:00
Benoit Steiner
d3f7915aeb Pulled latest update from the eigen main codebase 2015-03-24 13:12:14 -07:00
Benoit Steiner
abdbe8562e Fixed the CUDA packet primitives 2015-03-24 10:45:46 -07:00
Gael Guennebaud
29eaa2b0f1 Make MatrixBase::is* methods aware of nested_eval. 2015-03-24 13:42:42 +01:00
Gael Guennebaud
d27968eb7e D&C SVD: directly falls back to JacobiSVD for very small problems (by-pass upper-bidiagonalization) 2015-03-24 13:38:07 +01:00
Gael Guennebaud
4472f3e578 Avoid SVD: consider denormalized small numbers as zero when computing the rank of the matrix 2015-03-23 09:40:21 +01:00
Deanna Hood
83e5b7656b Use M_PI instead of acos(-1) for pi 2015-03-22 06:04:31 +10:00
Deanna Hood
4bab4790c0 Add \sa tags of isFinite/isInf for each other 2015-03-22 05:39:08 +10:00
Gael Guennebaud
4e2b18d909 Update approx. minimum ordering method to push and keep structural empty diagonal elements to the bottom-right part of the matrix 2015-03-20 16:33:48 +01:00
Gael Guennebaud
d6b2f300db Fix MSVC compilation: aligned type must be passed by reference 2015-03-19 17:28:32 +01:00
Gael Guennebaud
61c45d7cfd Fix comparison warning 2015-03-19 17:13:22 +01:00
Gael Guennebaud
f329d0908a Improve random number generation for integer and add unit test 2015-03-19 15:10:36 +01:00
Benoit Jacob
dc04f12967 use unsigned short instead of uint16_t which doesn't exist in c++98 2015-03-17 10:31:45 -04:00
Deanna Hood
596be3cd86 Use std::arg for real numbers when c++11 is used, custom implementation otherwise 2015-03-17 15:28:12 +10:00
Deanna Hood
e26134ed62 Use std::round when c++11 is used, custom implementation otherwise 2015-03-17 14:55:14 +10:00
Deanna Hood
e21e29a088 Update cost of arg call to depend on if the scalar is complex or not 2015-03-17 14:04:33 +10:00
Deanna Hood
447a5a6b01 Fix VML declarations to only be for real/complex as appropriate 2015-03-17 13:33:31 +10:00
Deanna Hood
f52b78491c Remove packet isNaN, isInf, isFinite 2015-03-17 09:26:24 +10:00
Deanna Hood
1c78d6f2a6 Add boolean not operator (!) array support 2015-03-17 08:29:57 +10:00
Benoit Jacob
364cfd529d Similar to cset 3589a9c115 
, also in 2px4 kernel: actual_panel_rows computation should always be resilient to parameters not consistent with the known L1 cache size, see comment
 2015-03-16 16:28:44 -04:00
Deanna Hood
e1d6e6c972 Make cube, inverse and abs2 free-functions 2015-03-17 06:25:24 +10:00
Benoit Jacob
577056aa94 Include stdint.h. Not going for cstdint because it is a C++11 addition. Needed for uint16_t at least, in lookup-table code. 2015-03-16 16:21:50 -04:00
Benoit Jacob
eb6929cb19 fix bug in maxsize calculation, which would cause products of size > 2048 to address the lookup table out of bounds 2015-03-16 16:15:47 -04:00
Deanna Hood
fef4e071d7 Rename isinf to isInf 2015-03-17 05:58:47 +10:00
Deanna Hood
46cf9cda32 Add isfinite array support as isFinite 2015-03-17 04:33:12 +10:00
Deanna Hood
1d76ceab55 Remove floor, ceil, round for complex numbers 2015-03-17 02:36:07 +10:00
Deanna Hood
717b7954ce Update cost of coeff-wise arg call 2015-03-17 02:11:57 +10:00
Deanna Hood
fb68b149cb Rename isnan to isNaN 2015-03-17 02:04:42 +10:00
Benoit Jacob
35c3a8bb84 Update Nexus 5 lookup table from combining now 2 runs of the benchmark, using the analyze-blocking-sizes partition tool. Gives better worst-case performance. 2015-03-16 11:05:51 -04:00
Benoit Jacob
e274607d7f fix compilation with GCC 4.8 2015-03-16 10:48:27 -04:00
Benoit Jacob
151b8b95c6 Fix bug in case where EIGEN_TEST_SPECIFIC_BLOCKING_SIZE is defined but false 2015-03-15 19:10:51 -04:00
Benoit Jacob
02babb9c0f Provide a empirical lookup table for blocking sizes measured on a Nexus 5. Only for float, only for Android on ARM 32bit for now. 2015-03-15 18:13:12 -04:00
Benoit Jacob
3589a9c115 actual_panel_rows computation should always be resilient to parameters not consistent with the known L1 cache size, see comment 2015-03-15 18:12:18 -04:00
Benoit Jacob
1dd3d89818 Fix a unused-var warning 2015-03-15 18:07:19 -04:00
Benoit Jacob
e56aabf205 Refactor computeProductBlockingSizes to make room for the possibility of using lookup tables 2015-03-15 18:05:12 -04:00
Benoit Jacob
488c15615a organize a little our default cache sizes, and use a saner default L1 outside of x86 (10% faster on Nexus 5) 2015-03-13 14:51:26 -07:00
Gael Guennebaud
1330f8bbd1 bug #973, improve AVX support by enabling vectorization of Vector4i-like types, and enforcing alignement of Vector4f/Vector2d-like types to preserve compatibility with SSE and future Eigen versions that will vectorize them with AVX enabled. 2015-03-13 21:15:50 +01:00
Gael Guennebaud
d99ab35f9e Fix internal::random(x,y) for integer types. The previous implementation could return y+1. The new implementation uses rejection sampling to get an unbiased behabior. 2015-03-13 21:12:46 +01:00
Gael Guennebaud
8580eb6808 bug #949: add static assertion for incompatible scalar types in dense end-user decompositions. 2015-03-13 21:06:20 +01:00
Gael Guennebaud
a9df28c95b SparseMatrix::insert: switch to a fully uncompressed mode if sequential insertion is not possible (otherwise an arbitrary large amount of memory was preallocated in some cases) 2015-03-13 21:00:21 +01:00
Gael Guennebaud
5ffe29cb9f Bound pre-allocation to the maximal size representable by StorageIndex and throw bad_alloc if that's not possible. 2015-03-13 20:57:33 +01:00
Gael Guennebaud
2f6f8bf31c Add missing coeff/coeffRef members to Block<sparse>, and extend unit tests. 2015-03-13 16:24:40 +01:00
Doug Kwan
657407227e Fix bug in pdiv<Packet1cd> which swaps 32-bit halves of a pair of 
doubles instead of swapping the doubles.
 2015-03-11 15:13:37 -07:00
Deanna Hood
f89fcefa79 Add hyperbolic trigonometric functions from std array support 2015-03-11 13:13:30 +10:00
Deanna Hood
a5e49976f5 Add log10 array support 2015-03-11 08:56:42 +10:00
Deanna Hood
19a71056ae Allow calling of square(array) in addition to array.square() 2015-03-11 06:59:28 +10:00
Deanna Hood
31fdd67756 Additional unary coeff-wise functors (isnan, round, arg, e.g.) 2015-03-11 06:39:23 +10:00
Gael Guennebaud
fd78874888 Fix compilation of iterative solvers with dense matrices 2015-03-09 21:31:03 +01:00
Gael Guennebaud
d4317a85e8 Add typedefs for return types of SparseMatrixBase::selfadjointView 2015-03-09 21:29:46 +01:00
Gael Guennebaud
9e885fb766 Add unit tests for CG and sparse-LLT for long int as storage-index 2015-03-09 14:33:15 +01:00
Gael Guennebaud
224a1fe4c6 bug #963: make IncompleteLUT compatible with non-default storage index types. 2015-03-09 13:55:20 +01:00
Gael Guennebaud
0ee391863e Avoid undeflow when blocking size are tuned manually. 2015-03-06 21:51:09 +01:00
Gael Guennebaud
14a5f135a3 bug #969: workaround abiguous calls to Ref using enable_if. 2015-03-06 17:51:31 +01:00
Gael Guennebaud
87681e508f bug #978: early return for vanishing products 2015-03-06 16:11:22 +01:00
Gael Guennebaud
cd3bbffa73 Improve blocking heuristic: if the lhs fit within L1, then block on the rhs in L1 (allows to keep packed rhs in L1) 2015-03-06 14:31:39 +01:00
Gael Guennebaud
58740ce4c6 Improve product kernel: replace the previous dynamic loop swaping strategy by a more general one: 
It consists in increasing the actual number of rows of lhs's micro horizontal panel for small depth such that L1 cache is fully exploited.
 2015-03-06 10:30:35 +01:00
Gael Guennebaud
4c8b95d5c5 Rename LSCG to LeastSquaresConjugateGradient 2015-03-05 10:16:32 +01:00
Gael Guennebaud
7550107028 Product optimization: implement a dynamic loop-swapping startegy to improve memory accesses to the destination matrix in the case of K-rank-update like products, i.e., for products of the kind: "large x small" * "small x large" 2015-03-05 10:03:46 +01:00
Gael Guennebaud
2dc968e453 bug #824: improve accuracy of Quaternion::angularDistance using atan2 instead of acos. 2015-03-04 17:03:13 +01:00
Benoit Steiner
0196141938 Fixed the optimized AVX implementation of the fast rsqrt function 2015-03-02 13:49:39 -08:00
Benoit Steiner
4fd7f47692 Added an optimized version of rsqrt for SSE and AVX that is used when EIGEN_FAST_MATH is defined. 2015-03-02 09:38:47 -08:00
Benoit Steiner
fb53384b0f Improved the default implementation of prsqrt 2015-02-28 01:51:26 -08:00
Benoit Steiner
306fceccbe Pulled latest updates from trunk 2015-02-27 13:05:26 -08:00
Benoit Steiner
2386fc8528 Added support for 32bit index on a per tensor/tensor expression. This enables us to use 32bit indices to evaluate expressions on GPU faster while keeping the ability to use 64 bit indices to manipulate large tensors on CPU in the same binary. 2015-02-27 12:57:13 -08:00
Benoit Jacob
6466fa63be Reimplement the selection between rotating and non-rotating kernels 
using templates instead of macros and if()'s.
That was needed to fix the build of unit tests on ARM, which I had
broken. My bad for not testing earlier.
 2015-02-27 15:30:10 -05:00
Benoit Steiner
05089aba75 Switch to truncated casting when converting floating point types to integer. This ensures that vectorized casts are consistent with scalar casts 2015-02-27 09:27:30 -08:00
Benoit Steiner
573b377110 Added support for vectorized type casting of tensors 2015-02-27 08:46:04 -08:00
Benoit Jacob
2fc3b484d7 remove trailing comma 2015-02-27 11:37:45 -05:00
Benoit Jacob
33669348c4 Disable Packet2f/2i halfpacket support in NEON. 
I believe that it was erroneously turned on, since Packet2f/2i intrinsics are unimplemented,
and code trying to use halfpackets just fails to compile on NEON, as it tries to use the
default implementation of pload/pstore and the types don't match.
 2015-02-27 11:35:37 -05:00
Benoit Jacob
b7fc8746e0 Replace a static assert by a runtime one, fixes the build of unit tests on ARM 
Also safely assert in the non-implemented path that should never be taken in practice,
and would return wrong results.
 2015-02-27 10:01:59 -05:00
Benoit Steiner
f41b1f1666 Added support for fast reciprocal square root computation. 2015-02-26 09:42:41 -08:00
Gael Guennebaud
bcf9bb5c1f Avoid packing rhs multiple-times when blocking on the lhs only. 2015-02-26 17:01:33 +01:00
Gael Guennebaud
4ec3f04b3a Make sure that the block size computation is tested by our unit test. 2015-02-26 17:00:36 +01:00
Gael Guennebaud
a8ad8887bf Implement a more generic blocking-size selection algorithm. See explanations inlines. 
It performs extremely well on Haswell. The main issue is to reliably and quickly find the
actual cache size to be used for our 2nd level of blocking, that is: max(l2,l3/nb_core_sharing_l3)
 2015-02-26 16:04:35 +01:00
Gael Guennebaud
400becc591 Fix typos in block-size testing code, and set peeling on k to 8. 2015-02-26 15:57:06 +01:00
Benoit Jacob
692136350b So I extensively measured the impact of the offset in this prefetch. I tried offset values from 0 to 128 (on this float* pointer, so implicitly times 4 bytes). 
On x86, I tested a Sandy Bridge with AVX with 12M cache and a Haswell with AVX+FMA with 6M cache on MatrixXf sizes up to 2400.

I could not see any significant impact of this offset.

On Nexus 5, the offset has a slight effect: values around 32 (times sizeof float) are worst. Anything else is the same: the current 64 (8*pk), or... 0.

So let's just go with 0!

Note that we needed a fix anyway for not accounting for the value of RhsProgress. 0 nicely avoids the issue altogether!
 2015-02-25 12:37:14 -05:00
Christoph Hertzberg
531fa9de77 bug #970: Add EIGEN_DEVICE_FUNC to RValue functions, in case Cuda supports RValue-references. 2015-02-24 21:03:28 +01:00
Benoit Jacob
26275b250a Fix my recent prefetch changes: 
- the first prefetch is actually harmful on Haswell with FMA,
   but it is the most beneficial on ARM.
 - the second prefetch... I was very stupid and multiplied by sizeof(scalar)
   and offset of a scalar* pointer. The old offset was 64; pk = 8, so 64=pk*8.
   So this effectively restores the older offset. Actually, there were
   two prefetches here, one with offset 48 and one with offset 64. I could not
   confirm any benefit from this strange 48 offset on either the haswell or
   my ARM device.
 2015-02-23 16:55:17 -05:00
Christoph Hertzberg
052b6b40f1 Fix two trivial warnings 2015-02-22 12:40:51 +01:00
Christoph Hertzberg
ecbf2a6656 log1p is defined only for real Scalars in C++11 2015-02-21 19:58:24 +01:00
Gael Guennebaud
3cf642baa3 Fix compilation of unit tests disabling assertion cheking 2015-02-21 14:13:48 +01:00
Gael Guennebaud
2da1594750 Fix doc of Ref<> 2015-02-20 11:52:22 +01:00
Gael Guennebaud
b192e29eae In C++11 destructors do not throw by default (fix CommaInitializer unit test) 2015-02-20 09:28:34 +01:00
Benoit Steiner
ab41652d81 Pulled latest changes from trunk 2015-02-19 21:23:37 -08:00
Benoit Steiner
7765039f1c Marked the CUDA packet primitives as EIGEN_DEVICE_FUNC since they'll end up being executed on the GPU device. 2015-02-19 21:22:51 -08:00
Gael Guennebaud
a66f5fc2fd Fix regression with C++11 support of lambda: now internal::result_of falls back to std::result_of in C++11. 2015-02-19 23:32:12 +01:00
Gael Guennebaud
1b7e12847d Fix some calls to result_of on binary functors as unary ones. 2015-02-19 23:30:41 +01:00
Gael Guennebaud
0f4dd15dfc Declare const some const variables 2015-02-19 23:28:57 +01:00