After more than three years of efforts, Eigen 3.3 has been released today, on November 10, 2016.
Those include, a novel evaluation mechanism [2] of expressions, support for AVX, FMA, AVX512, VSX and ZVector vector instructions, unaligned vectorization, nvcc/CUDA, more OpenMP parallelism [3], a fast divide and
conquer SVD algorithm [4], a CompleteOrthogonalDecomposition class for fast minimal norm solving [5], a LS-CG solver [6], a fast reciprocal condition number estimators in LU and Cholesky factorizations, LU::transpose()/adjoint() API [7], support for inplace
decompositions [8], support for matrix-free iterative solvers [9], new array functions [10], support for any BLAS/LAPACK libraries as backend [11], improved support for mixing scalar types [12], eigenvectors in GeneralizedEigenSolver, a complete rewrite of
LinSpaced [13], a non officially supported but massively used Tensor module with CUDA and OpenCL support [14], and more [1].