[rocSolver docs](https://rocm.docs.amd.com/projects/rocSOLVER/en/latest/index.html) [CUSOLVER docs](https://docs.nvidia.com/cuda/cusolver/index.html) CUSOLVER/rocSOLVER supports: - [x] Randomized SVD - [x] CUDA - [ ] Symmetric/Hermitian eigensolving - [x] `syevd` (QR Iteration) - [x] CUDA - [x] AMD - [x] `syevdx` (QR Iteration with truncation) - [ ] CUDA - [x] AMD - [x] `syev` - [x] AMD - [x] `syevx` - [x] AMD - [x] `syevj` (Jacobi) - [ ] CUDA - [x] AMD - [ ] `syevdj` (Jacobi/divide and conquer) - [ ] CUDA - [x] AMD - [ ] `syevBatched` (Batched symmetric/Hermitian eigensolve) - [ ] AMD - [ ] CUDA - [x] General eigensolving - [x] `geev` (simple balancing) - [x] CUDA (https://github.com/QuantumKitHub/MatrixAlgebraKit.jl/pull/47) If anything else would be useful to have, please comment and I can add it to the tracker
rocSolver docs
CUSOLVER docs
CUSOLVER/rocSOLVER supports:
syevd(QR Iteration)syevdx(QR Iteration with truncation)syevsyevxsyevj(Jacobi)syevdj(Jacobi/divide and conquer)syevBatched(Batched symmetric/Hermitian eigensolve)geev(simple balancing)If anything else would be useful to have, please comment and I can add it to the tracker