3.4.1 to 5.0.0 · libeigen / eigen · GitLab

Commits on Source 1433

1d79c68b Fix ldexp for AVX512 (#2215 )
Apr 20, 2021
2b1dfd1b HasExp added for AVX512 Packet8d
Apr 20, 2021
06c2760b Fix taking address of rvalue compiler issue with TensorFlow (plus other warnings).
Apr 21, 2021
d72c794c Compilation of basicbenchmark fixed
Apr 21, 2021
85a76a16 Make vectorized compute_inverse_size4 compile with AVX.
Apr 22, 2021
d213a0bc DenseStorage safely copy/swap.
Apr 22, 2021
045c0609 Check existence of BSD random before use.
Apr 22, 2021
3804ca0d Fix for issue with static global variables in TensorDeviceGpu.h
Apr 23, 2021
1115f546 Tests for pcmp_lt and pcmp_le added
Apr 23, 2021
d87648a6 Tests added and AVX512 bug fixed for pcmp_lt_or_nan
Apr 25, 2021
2ced0cc2 Added complex matrix unit tests for SelfAdjointEigenSolve
Apr 26, 2021
172db7bf Add missing pcmp_lt_or_nan for NEON Packet4bf.
Apr 27, 2021
1c013be2 Better CUDA complex division.
Apr 29, 2021
e3b7f596 Simplify TensorRandom and remove time-dependence.
May 04, 2021
722ca0b6 Revert addition of unused `paddsub<Packet2cf>`. This fixes #2242
May 06, 2021
90e9a33e Fix numext::arg return type.
May 07, 2021
0eba8a1f Clean up gpu device properties.
May 07, 2021
c0eb5f89 Restore ABI compatibility for conj with 3.3, fix conflict with boost.
May 07, 2021
39ec31c0 Fix for issue where numext::imag and numext::real are used before they are defined.
May 10, 2021
3d9051ea Changing the storage of the SSE complex packets to that of the wrapper. This should fix #2242 .
May 10, 2021
6753f0f1 Fix ambiguity due to argument dependent lookup.
May 11, 2021
7e6a1c12 Device implementation of log for std::complex types.
May 11, 2021
972cf0c2 Fix calls to device functions from host code
May 11, 2021
3e006bfd Ensure all generated matrices for inverse_4x4 testes are invertible, this fix #2248 .
May 13, 2021
8877f8d9 ptranpose test for non-square kernels added
May 19, 2021
391094c5 Use derived object type in conservative_resize_like_impl
May 20, 2021
17200570 Adds macro for checking if C++14 variable templates are supported
May 21, 2021
12471fcb predux_half_dowto4 test extended to all applicable packets
May 21, 2021
ebb300d0 Modify Unary/Binary/TernaryOp evaluators to work for non-class types.
May 23, 2021
dba753a9 Add missing NEON ptranspose implementations.
May 25, 2021
91cd67f0 Remove EIGEN_DEVICE_FUNC from CwiseBinaryOp's default copy constructor.
May 26, 2021
c8d40a7b Removed dead code from GPU float16 unit test.
May 28, 2021
001a5751 Fix parsing of version for nvhpc
Jun 10, 2021
f64b2954 Fix c++20 warnings about using enums in arithmetic expressions.
Jun 10, 2021
fc87e2cb Use bit_cast to create -0.0 for floating point types to avoid compiler...
Jun 11, 2021
6aec8326 Allow custom TENSOR_CONTRACTION_DISPATCH macro.
Jun 11, 2021
514977f3 Add ability to permanently enable HIP/CUDA gpu* defines.
Jun 11, 2021
ad82d20c Fix checking of version number for mingw.
Jun 11, 2021
13fb5ab9 Fix more enum arithmetic.
Jun 15, 2021
95487918 Fix placement of permanent GPU defines.
Jun 15, 2021
9e94c595 Add missing ppc pcmp_lt_or_nan<Packet8bf>
Jun 15, 2021
175f0cc1 changed documentation to make example compile
Jun 16, 2021
ef1fd341 EIGEN_STRONG_INLINE was NOT inlining in some critical needed areas (6.6X...
Jun 16, 2021
12e8d571 Remove pset, replace with ploadu.
Jun 16, 2021
35a367d5 Fix fix<> for gcc-4.9.3.
Jun 18, 2021
e9ab4278 Rewrite balancer to avoid overflows.
Jun 21, 2021
c8a2b4d2 Fix typo in SelfAdjointEigenSolver_eigenvectors.cpp
Jun 21, 2021
ea62c937 Update ComplexEigenSolver_eigenvectors.cpp
Jun 21, 2021
4ad30a73 Use internal::ref_selector to avoid holding a reference to a RHS expression.
Jun 22, 2021
52a5f982 Get rid of code duplication for conj_helper. For packets where LhsType=RhsType...
Jun 24, 2021
bffd267d Small cleanup: Get rid of the macros EIGEN_HAS_SINGLE_INSTRUCTION_CJMADD and...
Jun 24, 2021
2d132d17 Commit 52a5f982 broke conjhelper functionality for HIP GPUs.
Jun 25, 2021
3a087ccb Modify tensor argmin/argmax to always return first occurence.
Jun 29, 2021
5aebbe90 Get rid of redundant `pabs` instruction in complex square root.
Jun 29, 2021
c81da59a Correct declarations for aarch64-pc-windows-msvc
Jun 30, 2021
60400334 Make DenseStorage<> trivially_copyable
Jun 30, 2021
91e99ec1 Create the ability to disable the specialized gemm_pack_rhs in Eigen (only PPC) for TensorFlow
Jun 30, 2021
eb047759 Fix duplicate definitions on Mac
Jul 01, 2021
154f00e9 Fix inverse nullptr/asan errors for LU.
Jul 01, 2021
6035da52 Fix compile issues for gcc 4.8.
Jul 01, 2021
9312a5bf Implement a generic vectorized version of Smith's algorithms for complex division.
Jul 01, 2021
bbfc4d54 Use `padd` instead of `+`.
Jul 02, 2021
aab74702 Don't crash when attempting to shuffle an empty tensor.
Jul 02, 2021
7b35638d Fix breakage of conj_helper in conjunction with custom types introduced in !537 .
Jul 02, 2021
f5a9873b Fix Tensor documentation page.
Jul 03, 2021
f44f0553 Fix CMake directory issues.
Jul 07, 2021
a59cf78c Add Doxygen-style documentation to main.h.
Jul 07, 2021
4bcd42c2 Make a copy of input matrix when try to do the inverse in place, this fixes #2285 .
Jul 08, 2021
e9c9a313 Removed superfluous boolean `degenerate` in TensorMorphing.h.
Jul 08, 2021
1e6c6c15 Replace memset with fill to work for non-trivial scalars.
Jul 08, 2021
0c361c48 Defer to std::fill_n when filling a dense object with a constant value.
Jul 09, 2021
beea14a1 Enable extract et. al. for HIP GPU.
Jul 09, 2021
9c22795d Put attach/detach buffer back in for TensorDeviceSycl.
Jul 09, 2021
1fd5ce10 For GpuDevice::fill, use a single memset if all bytes are equal.
Jul 10, 2021
297f0f56 Fix explicit default cache size typo.
Jul 20, 2021
a7763838 Fixes #1387 for compilation error in JacobiSVD with...
Jul 20, 2021
66ca41bd Add support for vectorizing logical comparisons.
Jul 23, 2021
de2e62c6 Disable vectorization of comparisons except for bool.
Jul 25, 2021
5b81764c Fixed typo in TutorialSparse.dox
Jul 26, 2021
f357283d _DerType -> DerivativeType as underscore-followed-by-caps is a reserved identifier
Jul 29, 2021
9816fe59 Fix assignment operator issue for latest MSVC+NVCC.
Aug 03, 2021
02a0e79c fix:typo in dox (has->have)
Aug 03, 2021
c86ac71b Put code in monospace (typewriter) style.
Aug 03, 2021
7880f105 Enable equality comparisons on GPU.
Aug 03, 2021
3d98a6ef Modify scalar pzero, ptrue, pselect, and p<binary> operations to avoid memset.
Aug 03, 2021
8cf6cb27 Fix TriSycl CMake files.
Aug 03, 2021
1cdec386 Fix cmake warnings, FindPASTIX/FindPTSCOTCH.
Aug 03, 2021
31f796eb Fix MPReal detection and support.
Aug 03, 2021
5ad8b9bf Make inverse 3x3 faster and avoid gcc bug.
Aug 04, 2021
4ba872bd Avoid leading underscore followed by cap in template identifiers
Aug 04, 2021
4d870c49 updated documentation for middleCol and middleRow
Aug 05, 2021
ae2abe1f ci: ppc64le: disable MMA and remove 'allow_failure' for gcc-10 CXX11=off
Aug 05, 2021
a5a7faeb Avoid memory allocation in tridiagonalization_inplace_selector::run.
Aug 06, 2021
09d71224 Do not build shared libs if not supported
Aug 06, 2021
e6a5a594 remove denormal flushing in fp32tobf16 for avx & avx512
Aug 09, 2021
e3e74001 added includes for unordered_map
Aug 10, 2021
413bc491 Fix errors on older compilers (gcc 7.5 - lack of vec_neg, clang10 - can not...
Aug 10, 2021
f1b899ee Do not set AnnoyingScalar::dont_throw if not defined EIGEN_TEST_ANNOYING_SCALAR_DONT_THROW.
Aug 11, 2021
8ce341ca * revise the meta_least_common_multiple function template, add a bool variable...
Aug 11, 2021
fb1718ad Update code snippet for tridiagonalize_inplace.
Aug 12, 2021

1,333 additional commits have been omitted to prevent performance issues.