devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 74e558cfa8 Pulled latest updates from trunk Benoit Steiner 2015-04-01 23:24:11 -07:00
  • 03a0df2010 Fixed some compilation warnings triggered by pre-cxx11 comoilers Benoit Steiner 2015-04-01 22:51:33 -07:00
  • b8b7807269 Fixed some compilation warning triggered by the cxx11 emulation code Benoit Steiner 2015-04-01 21:48:18 -07:00
  • 383b6dfafe Fixed 2 typos Benoit Steiner 2015-04-01 16:44:36 -07:00
  • 5861cfb55e Remove unused GenericSparseBlockInnerIteratorImpl code. Gael Guennebaud 2015-04-01 22:29:29 +02:00
  • 3105986e71 bug #875: remove broken SparseMatrixBase::nonZeros and introduce a nonZerosEstimate() method to sparse evaluators for internal uses. Factorize some code in SparseCompressedBase. Gael Guennebaud 2015-04-01 22:27:34 +02:00
  • 39dcd01b0a bug #973: enable alignment of multiples of half-packet size (e.g., Vector6d with AVX) Gael Guennebaud 2015-04-01 13:55:09 +02:00
  • 8481dc21ea bug #986: add support for coefficient-based product with 0 depth. Gael Guennebaud 2015-04-01 13:15:23 +02:00
  • 79b4e6acaf Fix bug #987: wrong alignement guess in diagonal product. Gael Guennebaud 2015-03-31 23:35:12 +02:00
  • 3c38589984 Remove most of the dynamic memory allocations that occured in D&C SVD. Still remains the calls to JacobiSVD and UpperBidiagonalization. Gael Guennebaud 2015-03-31 22:54:47 +02:00
  • 8313fb7df7 Add row/column-wise reverseInPlace feature. Gael Guennebaud 2015-03-31 21:35:53 +02:00
  • dfb674a25e Make reverseInPlace really work in-place. Gael Guennebaud 2015-03-31 20:17:10 +02:00
  • 20d030f207 Fix vectorization of swap for non trivial expressions Gael Guennebaud 2015-03-31 20:16:02 +02:00
  • 678207e02a Added regression tests for tensor convolutions Benoit Steiner 2015-03-31 09:08:08 -07:00
  • 68d4afe985 Added support for convolution of tensors laid out in RowMajor mode Benoit Steiner 2015-03-31 09:07:09 -07:00
  • f873686602 Added documentation for the convolution operation Benoit Steiner 2015-03-31 08:27:23 -07:00
  • 73cdeae1d3 Only use blocking sizes LUTs for single-thread products for now Benoit Jacob 2015-03-31 11:17:23 -04:00
  • 0cbd5ae3cb Correctly detect Android with ndk_build Benoit Jacob 2015-03-31 11:17:21 -04:00
  • ae01c05e18 Fix computeProductBlockingSizes with m==0, and add respective unit test. Gael Guennebaud 2015-03-31 15:19:57 +02:00
  • bd76d837e6 Fix sign of SuperLU::determinant Gael Guennebaud 2015-03-31 14:57:32 +02:00
  • 35d3053d55 Fix regression introduced in 3b169d792d Gael Guennebaud 2015-03-31 09:23:53 +02:00
  • 731d7b84b4 Sharded a large test Benoit Steiner 2015-03-30 23:26:45 -07:00
  • 7bd578d11d Change CMake warning to simple message for old Metis versions Christoph Hertzberg 2015-03-31 00:50:04 +02:00
  • 3b169d792d Suppress unused variable warning Christoph Hertzberg 2015-03-31 00:49:08 +02:00
  • 3238ca6abc Addendum to last patch: k is Index and not int Christoph Hertzberg 2015-03-31 00:42:14 +02:00
  • 1efae98fee bug #985: RealQZ failed when either matrix had zero rows or columns (report and patch by Ben Goodrich) Also added a regression test Christoph Hertzberg 2015-03-30 23:56:20 +02:00
  • 35722fa022 Made the index type a template parameter of the tensor class instead of encoding it in the options. Benoit Steiner 2015-03-30 14:55:54 -07:00
  • 71950f02e5 Deleted unnecessary semicolons Benoit Steiner 2015-03-30 14:49:10 -07:00
  • 58af8bf90c bug #982: Make sure numext::maxi and numext::mini are called correctly, in case Scalar expressions return expression templates. Christoph Hertzberg 2015-03-30 16:47:22 +02:00
  • 2adbf6b8ca fix stupid warning with old GCC Gael Guennebaud 2015-03-28 22:34:54 +01:00
  • 41e20248f8 merge Gael Guennebaud 2015-03-28 14:43:35 +01:00
  • 09a5361d1b bug #983: Pass Vector3 by const reference and not by value Christoph Hertzberg 2015-03-28 12:36:24 +01:00
  • 266a84558f Optionally build the documentation when building unit tests. Christoph Hertzberg 2015-03-27 16:36:59 +01:00
  • 1b4bb20cf1 Merged in d_hood/eigen/sparse-tutorial-doc-fix (pull request PR-107) Christoph Hertzberg 2015-03-27 16:22:16 +01:00
  • eb7e4c2b9c Pass Vector3 type by reference Gael Guennebaud 2015-03-27 12:11:24 +01:00
  • ad044008da Fix transpose versus adjoint. Gael Guennebaud 2015-03-27 12:07:14 +01:00
  • 79cb875249 merge Gael Guennebaud 2015-03-27 10:56:04 +01:00
  • 7e225b6fa4 Suppress some false negatives in SVD unit test Gael Guennebaud 2015-03-27 10:55:53 +01:00
  • 1b8cc9af43 Slight numerical stability improvement in 2x2 svd Gael Guennebaud 2015-03-27 10:55:00 +01:00
  • 3d59ae0203 Fix hypot(0,0). Gael Guennebaud 2015-03-27 09:59:24 +01:00
  • 4df8b5a75e Avoid making an unecessary copy of the tensor expression when evaluating it on a GPU device Benoit Steiner 2015-03-25 14:36:07 -07:00
  • b3343bfdae Fixed the vectorized implementation of the Tensor select() method Benoit Steiner 2015-03-25 13:25:53 -07:00
  • ccf290a65c Cleaned up the TensorDevice code a little bit. Benoit Steiner 2015-03-25 12:37:38 -07:00
  • d3f7915aeb Pulled latest update from the eigen main codebase Benoit Steiner 2015-03-24 13:12:14 -07:00
  • abdbe8562e Fixed the CUDA packet primitives Benoit Steiner 2015-03-24 10:45:46 -07:00
  • 29eaa2b0f1 Make MatrixBase::is* methods aware of nested_eval. Gael Guennebaud 2015-03-24 13:42:42 +01:00
  • f42b105f73 Add the possibility to make VERIFY* checks to output a warning instead of abording. Gael Guennebaud 2015-03-24 13:39:14 +01:00
  • d27968eb7e D&C SVD: directly falls back to JacobiSVD for very small problems (by-pass upper-bidiagonalization) Gael Guennebaud 2015-03-24 13:38:07 +01:00
  • 4472f3e578 Avoid SVD: consider denormalized small numbers as zero when computing the rank of the matrix Gael Guennebaud 2015-03-23 09:40:21 +01:00
  • 83e5b7656b Use M_PI instead of acos(-1) for pi Deanna Hood 2015-03-22 06:04:31 +10:00
  • 4bab4790c0 Add \sa tags of isFinite/isInf for each other Deanna Hood 2015-03-22 05:39:08 +10:00
  • 4e2b18d909 Update approx. minimum ordering method to push and keep structural empty diagonal elements to the bottom-right part of the matrix Gael Guennebaud 2015-03-20 16:33:48 +01:00
  • 8d9bfb3a7b fix loadMarket wrt Index versus int Gael Guennebaud 2015-03-20 16:00:10 +01:00
  • a6a628ca6b Added the -= operator to the device classes Benoit Steiner 2015-03-19 23:22:19 -07:00
  • e134226a03 Fixed a bug in the handling of packets by the MeanReducer Benoit Steiner 2015-03-19 23:11:42 -07:00
  • 9ee62fdcd5 Fix random unit test for 32bits systems. Gael Guennebaud 2015-03-19 21:39:37 +01:00
  • d6b2f300db Fix MSVC compilation: aligned type must be passed by reference Gael Guennebaud 2015-03-19 17:28:32 +01:00
  • 61c45d7cfd Fix comparison warning Gael Guennebaud 2015-03-19 17:13:22 +01:00
  • d7698c18b7 Split sparse_basic unit test Gael Guennebaud 2015-03-19 15:11:05 +01:00
  • f329d0908a Improve random number generation for integer and add unit test Gael Guennebaud 2015-03-19 15:10:36 +01:00
  • 2ab4922431 Make html directory before generating output image there Deanna Hood 2015-03-18 07:24:13 +10:00
  • 41b717de25 More extensive unit tests for recent array-wise functors Deanna Hood 2015-03-18 03:11:03 +10:00
  • cc0f89eb3b Changed the way lvalue operations are declared in TensorBase: this fixes constness isses that prevented some expressions mixing lvalues and rvalues from compiling. Benoit Steiner 2015-03-17 09:57:20 -07:00
  • dc04f12967 use unsigned short instead of uint16_t which doesn't exist in c++98 Benoit Jacob 2015-03-17 10:31:45 -04:00
  • 8878e1c1de Remove ambiguity with recent numext methods isNaN and isInf Deanna Hood 2015-03-17 22:39:51 +10:00
  • 596be3cd86 Use std::arg for real numbers when c++11 is used, custom implementation otherwise Deanna Hood 2015-03-17 15:28:12 +10:00
  • e26134ed62 Use std::round when c++11 is used, custom implementation otherwise Deanna Hood 2015-03-17 14:55:14 +10:00
  • e21e29a088 Update cost of arg call to depend on if the scalar is complex or not Deanna Hood 2015-03-17 14:04:33 +10:00
  • 447a5a6b01 Fix VML declarations to only be for real/complex as appropriate Deanna Hood 2015-03-17 13:33:31 +10:00
  • f52b78491c Remove packet isNaN, isInf, isFinite Deanna Hood 2015-03-17 09:26:24 +10:00
  • 1c78d6f2a6 Add boolean not operator (!) array support Deanna Hood 2015-03-17 08:29:57 +10:00
  • 85da0c2281 Remove test of now-missing floor, ceil, round complex implementations Deanna Hood 2015-03-17 06:56:47 +10:00
  • 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 Benoit Jacob 2015-03-16 16:28:44 -04:00
  • 25664afacd Pulled latest updates from trunk Benoit Steiner 2015-03-16 13:25:45 -07:00
  • e1d6e6c972 Make cube, inverse and abs2 free-functions Deanna Hood 2015-03-17 06:25:24 +10:00
  • 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. Benoit Jacob 2015-03-16 16:21:50 -04:00
  • 5144f66728 Fixed compilation warning Benoit Steiner 2015-03-16 13:17:52 -07:00
  • 0fd6d52724 Fixed compilation error with clang Benoit Steiner 2015-03-16 13:16:12 -07:00
  • eb6929cb19 fix bug in maxsize calculation, which would cause products of size > 2048 to address the lookup table out of bounds Benoit Jacob 2015-03-16 16:15:47 -04:00
  • f218c0181d Fixes the Lvalue computation by actually setting the LvalueBit properly when instantiating tensors of const T. Added a test to check the fix. Benoit Steiner 2015-03-16 13:05:00 -07:00
  • fef4e071d7 Rename isinf to isInf Deanna Hood 2015-03-17 05:58:47 +10:00
  • 46cf9cda32 Add isfinite array support as isFinite Deanna Hood 2015-03-17 04:33:12 +10:00
  • 7b829940d1 Add code snippets for new methods Deanna Hood 2015-03-17 03:40:28 +10:00
  • 1d76ceab55 Remove floor, ceil, round for complex numbers Deanna Hood 2015-03-17 02:36:07 +10:00
  • 717b7954ce Update cost of coeff-wise arg call Deanna Hood 2015-03-17 02:11:57 +10:00
  • fb68b149cb Rename isnan to isNaN Deanna Hood 2015-03-17 02:04:42 +10:00
  • 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. Benoit Jacob 2015-03-16 11:05:51 -04:00
  • e274607d7f fix compilation with GCC 4.8 Benoit Jacob 2015-03-16 10:48:27 -04:00
  • 151b8b95c6 Fix bug in case where EIGEN_TEST_SPECIFIC_BLOCKING_SIZE is defined but false Benoit Jacob 2015-03-15 19:10:51 -04:00
  • 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. Benoit Jacob 2015-03-15 18:13:12 -04:00
  • 3589a9c115 actual_panel_rows computation should always be resilient to parameters not consistent with the known L1 cache size, see comment Benoit Jacob 2015-03-15 18:12:18 -04:00
  • 1dd3d89818 Fix a unused-var warning Benoit Jacob 2015-03-15 18:07:19 -04:00
  • ca5c12587b Polish lookup tables generation Benoit Jacob 2015-03-15 18:05:53 -04:00
  • e56aabf205 Refactor computeProductBlockingSizes to make room for the possibility of using lookup tables Benoit Jacob 2015-03-15 18:05:12 -04:00
  • b6b88c0808 update perf_monitoring/gemm/changesets.txt Benoit Jacob 2015-03-13 14:57:05 -07:00
  • 488c15615a organize a little our default cache sizes, and use a saner default L1 outside of x86 (10% faster on Nexus 5) Benoit Jacob 2015-03-13 14:51:26 -07:00
  • 9f58524cbd merge Gael Guennebaud 2015-03-13 21:16:39 +01:00
  • 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. Gael Guennebaud 2015-03-13 21:15:50 +01:00
  • 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. Gael Guennebaud 2015-03-13 21:12:46 +01:00
  • 8580eb6808 bug #949: add static assertion for incompatible scalar types in dense end-user decompositions. Gael Guennebaud 2015-03-13 21:06:20 +01:00