2023 08

The SLATE project made the 2023.08.25 release of its distributed, GPU-accelerated dense linear algebra library, with all major functionality to replace ScaLAPACK, and ports to all three target GPU platforms: AMD ROCm, Intel oneAPI, and NVIDIA CUDA. Major new features include eigenvectors in the Hermitian eigenvalue decomposition, used in many scientific applications such as chemistry electronic structure calculations, and singular vectors in the Singular Value Decomposition (SVD), used across a plethora of applications such as data fitting and compressed data representations. This release also added oneAPI support to target the upcoming Aurora supercomputer. SLATE is available at https://github.com/icl-utk-edu/slate/releases