Commit message (Collapse) | Author | Age | |
---|---|---|---|
* | Merged eigen/eigen into default | 2016-11-03 | |
|\ | |||
| * | bug #1004: improve accuracy of LinSpaced for abs(low) >> abs(high). | 2016-11-02 | |
| | | |||
| * | Add pinsertfirst function and implement pinsertlast for complex on SSE/AVX. | 2016-11-02 | |
| | | |||
| * | Fix regression in X = (X*X.transpose())/s with X rectangular by deferring ↵ | 2016-10-26 | |
| | | | | | | | | resizing of the destination after the creation of the evaluator of the source expression. | ||
| * | bug #1004: one more rewrite of LinSpaced for floating point numbers to ↵ | 2016-10-25 | |
| | | | | | | | | | | | | | | | | guarantee both interpolation and monotonicity. This version simply does low+i*step plus a branch to return high if i==size-1. Vectorization is accomplished with a branch and the help of pinsertlast. Some quick benchmark revealed that the overhead is really marginal, even when filling small vectors. | ||
| * | Add a pinsertlast function replacing the last entry of a packet by a scalar. | 2016-10-25 | |
| | | | | | | | | (useful to vectorize LinSpaced) | ||
| * | bug #1004: remove the inaccurate "sequential" path for LinSpaced, mark ↵ | 2016-10-24 | |
| | | | | | | | | | | | | respective function as deprecated, and enforce strict interpolation of the higher range using a correction term. Now, even with floating point precision, both the 'low' and 'high' bounds are exactly reproduced at i=0 and i=size-1 respectively. | ||
| * | bug #698: rewrite LinSpaced for integer scalar types to avoid overflow and ↵ | 2016-10-24 | |
| | | | | | | | | | | | | | | guarantee an even spacing when possible. Otherwise, the "high" bound is implicitly lowered to the largest value allowing for an even distribution. This changeset also disable vectorization for this integer path. | ||
| * | Add unit test for overflow in LinSpaced | 2016-10-24 | |
| | | |||
* | | Merged eigen/eigen into default | 2016-10-12 | |
|\| | |||
| * | Fix SPQR for rectangular matrices | 2016-10-12 | |
| | | |||
| * | Merged in rmlarsen/eigen (pull request PR-230) | 2016-10-12 | |
| |\ | | | | | | | | | | Fix a bug in psqrt for SSE and AVX when EIGEN_FAST_MATH=1 | ||
| * | | Remove static qualifier of free-functions (inline is enough and this helps ↵ | 2016-10-07 | |
| | | | | | | | | | | | | ICC to find the right overload) | ||
* | | | Renamed predux_half into predux_downto4 | 2016-10-06 | |
| | | | |||
| * | | Fix compilation of qr.inverse() for column and full pivoting variants. | 2016-10-06 | |
| | | | |||
* | | | Merged latest updates from trunk | 2016-10-05 | |
|\| | | |||
| | * | Fix a bug in the implementation of Carmack's fast sqrt algorithm in Eigen ↵ | 2016-10-04 | |
| |/ | | | | | | | | | | | | | | | | | | | | | | | | | | | (enabled by EIGEN_FAST_MATH), which causes the vectorized parts of the computation to return -0.0 instead of NaN for negative arguments. Benchmark speed in Giga-sqrts/s Intel(R) Xeon(R) CPU E5-1650 v3 @ 3.50GHz ----------------------------------------- SSE AVX Fast=1 2.529G 4.380G Fast=0 1.944G 1.898G Fast=1 fixed 2.214G 3.739G This table illustrates the worst case in terms speed impact: It was measured by repeatedly computing the sqrt of an n=4096 float vector that fits in L1 cache. For large vectors the operation becomes memory bound and the differences between the different versions almost negligible. | ||
| * | Improved support for compiling CUDA code with clang as the host compiler | 2016-10-03 | |
| | | |||
| * | bug #1310: workaround a compilation regression from 3.2 regarding triangular ↵ | 2016-09-30 | |
| | | | | | | | | * homogeneous | ||
| * | bug #1308: fix compilation of some small products involving nullary-expressions. | 2016-09-29 | |
| | | |||
| * | bug #1311: fix alignment logic in some cases of ↵ | 2016-09-26 | |
| | | | | | | | | (scalar*small).lazyProduct(small) | ||
| * | bug #1308: fix compilation of vector * rowvector::nullary. | 2016-09-25 | |
| | | |||
| * | bug #1304: fix Projective * scaling and Projective *= scaling | 2016-09-23 | |
| | | |||
| * | Add a note regarding gcc bug #72867 | 2016-09-22 | |
| | | |||
| * | fix typo | 2016-09-21 | |
| | | |||
| * | Disable a failure test if extended double precision is in use (x87) | 2016-09-21 | |
| | | |||
| * | Fix compilation if ICC. | 2016-09-21 | |
| | | |||
| * | Silent warning. | 2016-09-17 | |
| | | |||
| * | fix compilation with boost::multiprec | 2016-09-17 | |
| | | |||
| * | Fix alignement of statically allocated temporaries in gemv. | 2016-09-17 | |
| | | |||
| * | Fix undeflow | 2016-09-16 | |
| | | |||
| * | bug #1195: move NumTraits::Div<>::Cost to internal::scalar_div_cost (with ↵ | 2016-09-08 | |
| | | | | | | | | some specializations in arch/SSE and arch/AVX) | ||
| * | Workaround MSVC instantiation faillure of has_*ary_operator at the level of ↵ | 2016-09-06 | |
| | | | | | | | | triats<Ref>::match so that the has_*ary_operator are really properly instantiated throughout the compilation unit. | ||
| * | bug #1266: remove CUDA guards on MatrixBase::<decomposition> definitions. ↵ | 2016-09-06 | |
| | | | | | | | | (those used to break old nvcc versions that we propably don't care anymore) | ||
| * | Fix compilation when cuda_fp16.h does not exist. | 2016-09-05 | |
| | | |||
| * | Workaround a weird msvc 2012 compilation error. | 2016-09-05 | |
| | | |||
| * | bug #1286: automatically detect the available prototypes of functors passed ↵ | 2016-08-31 | |
| | | | | | | | | | | | | | | | | | | to CwiseNullaryExpr such that functors have only to implement the operators that matters among: operator()() operator()(i) operator()(i,j) Linear access is also automatically detected based on the availability of operator()(i,j). | ||
| * | bug #1283: add regression unit test | 2016-08-31 | |
| | | |||
| * | Fix 4x4 inverse with non-linear destination | 2016-08-30 | |
| | | |||
| * | Fix unit test for 1x1 matrices | 2016-08-30 | |
| | | |||
| * | bug #1271: add SparseMatrix::coeffs() methods returning a 1D view of the non ↵ | 2016-08-29 | |
| | | | | | | | | zero coefficients. | ||
| * | Add some pre-allocation unit tests (not working yet) | 2016-08-29 | |
| | | |||
| * | Fix compilation with cuda 8 | 2016-08-29 | |
| | | |||
| * | Add support for non trivial scalar factor in sparse selfadjoint * dense ↵ | 2016-08-24 | |
| | | | | | | | | | | | | products, and enable +=/-= assignement for such products. This changeset also improves the performance by working on column of the result at once. | ||
| * | bug #1268: detect faillure in LDLT and report them through info() | 2016-08-23 | |
| | | |||
| * | Fix compilation in non C++11 mode. | 2016-08-23 | |
| | | |||
| * | bug #645: patch from Tobias Wood implementing the extraction of eigenvectors ↵ | 2016-08-23 | |
| | | | | | | | | in GeneralizedEigenSolver | ||
| * | Optimize expression matching "d?=a-b*c" as "d?=a; d?=b*c;" | 2016-08-23 | |
| | | |||
| * | Fix compilation in check_for_aliasing due to ambiguous specializations | 2016-08-23 | |
| | | |||
| * | Fix possible overflow and biais in integer random generator | 2016-08-23 | |
| | |