Abstract: Integrated Singular Value Decomposition (iSVD) is a randomized method that　was proposed recently to compute low-rank SVD of large matrices arising in　data analysis and scientific computing. The iSVD randomly projects the　coefficient matrix to several low-dimensional subspaces and performs　low-rank SVD on each of them. The method then integrates these　low-dimensional approximate SVDs by solving a minimization problem. We will　discuss the recent developments of iSVD in various forms of parallelism and　its extension to the principal component analysis matrices and high order　SVD of tensors. Collaborations in iSVD in terms of high-performance　computing, auto-tuning, applications, etc. are sincerely welcome.