hipSOLVER
rocm-5.2.1
1. hipSOLVER User Guide
1.1. Introduction
1.1.1. Library overview
1.1.2. Currently implemented functionality
LAPACK auxiliary functions
LAPACK main functions
1.1.3. Compatibility-only functions
Iterative Jacobi functions
1.2. Installation
1.2.1. Install pre-built packages
1.2.2. Build & install library using script (Ubuntu only)
1.2.3. Build & install library manually
1.2.4. Build library + tests + benchmarks + samples manually
Dependencies (only necessary for hipSOLVER clients)
Library and clients
1.3. Using hipSOLVER
1.3.1. Some considerations when using the regular hipSOLVER API
Different signatures and additional API methods
Unsupported methods
Arguments not referenced by rocSOLVER
Using rocSOLVER’s memory model
Using rocSOLVER’s in-place functions
1.3.2. Porting cuSOLVER applications to hipSOLVER
Possible performance implications of the compatibility API
2. hipSOLVER API
2.1. Types
2.1.1. hipSOLVER regular API types
hipsolverHandle_t
hipsolverStatus_t
hipsolverOperation_t
hipsolverFillMode_t
hipsolverSideMode_t
hipsolverEigMode_t
hipsolverEigType_t
2.2. Helper Functions
2.2.1. Handle set-up and tear-down
hipsolverCreate()
hipsolverDestroy()
2.2.2. Stream manipulation
hipsolverSetStream()
hipsolverGetStream()
2.3. LAPACK Auxiliary Functions
2.3.1. Orthonormal matrices
hipsolver<type>orgbr_bufferSize()
hipsolver<type>orgbr()
hipsolver<type>orgqr_bufferSize()
hipsolver<type>orgqr()
hipsolver<type>orgtr_bufferSize()
hipsolver<type>orgtr()
hipsolver<type>ormqr_bufferSize()
hipsolver<type>ormqr()
hipsolver<type>ormtr_bufferSize()
hipsolver<type>ormtr()
2.3.2. Unitary matrices
hipsolver<type>ungbr_bufferSize()
hipsolver<type>ungbr()
hipsolver<type>ungqr_bufferSize()
hipsolver<type>ungqr()
hipsolver<type>ungtr_bufferSize()
hipsolver<type>ungtr()
hipsolver<type>unmqr_bufferSize()
hipsolver<type>unmqr()
hipsolver<type>unmtr_bufferSize()
hipsolver<type>unmtr()
2.4. LAPACK Functions
2.4.1. Triangular factorizations
hipsolver<type>potrf_bufferSize()
hipsolver<type>potrfBatched_bufferSize()
hipsolver<type>potrf()
hipsolver<type>potrfBatched()
hipsolver<type>getrf_bufferSize()
hipsolver<type>getrf()
hipsolver<type>sytrf_bufferSize()
hipsolver<type>sytrf()
2.4.2. Orthogonal factorizations
hipsolver<type>geqrf_bufferSize()
hipsolver<type>geqrf()
2.4.3. Problem and matrix reductions
hipsolver<type>gebrd_bufferSize()
hipsolver<type>gebrd()
hipsolver<type>sytrd_bufferSize()
hipsolver<type>hetrd_bufferSize()
hipsolver<type>sytrd()
hipsolver<type>hetrd()
2.4.4. Linear-systems solvers
hipsolver<type>potri_bufferSize()
hipsolver<type>potri()
hipsolver<type>potrs_bufferSize()
hipsolver<type>potrsBatched_bufferSize()
hipsolver<type>potrs()
hipsolver<type>potrsBatched()
hipsolver<type>getrs_bufferSize()
hipsolver<type>getrs()
hipsolver<type><type>gesv_bufferSize()
hipsolver<type><type>gesv()
2.4.5. Least-squares solvers
hipsolver<type><type>gels_bufferSize()
hipsolver<type><type>gels()
2.4.6. Symmetric eigensolvers
hipsolver<type>syevd_bufferSize()
hipsolver<type>heevd_bufferSize()
hipsolver<type>syevd()
hipsolver<type>heevd()
hipsolver<type>sygvd_bufferSize()
hipsolver<type>hegvd_bufferSize()
hipsolver<type>sygvd()
hipsolver<type>hegvd()
2.4.7. Singular value decomposition
hipsolver<type>gesvd_bufferSize()
hipsolver<type>gesvd()
3. hipSOLVER Compatibility API
3.1. Types
3.1.1. hipSOLVER compatibility API types
hipsolverDnHandle_t
hipsolverGesvdjInfo_t
hipsolverSyevjInfo_t
hipsolverStatus_t
hipsolverOperation_t
hipsolverFillMode_t
hipsolverSideMode_t
hipsolverEigMode_t
hipsolverEigType_t
3.2. Helper Functions
3.2.1. Handle set-up and tear-down
hipsolverDnCreate()
hipsolverDnDestroy()
3.2.2. Stream manipulation
hipsolverDnSetStream()
hipsolverDnGetStream()
3.2.3. Gesvdj parameter manipulation
hipsolverDnCreateGesvdjInfo()
hipsolverDnDestroyGesvdjInfo()
hipsolverDnXgesvdjSetMaxSweeps()
hipsolverDnXgesvdjSetSortEig()
hipsolverDnXgesvdjSetTolerance()
hipsolverDnXgesvdjGetResidual()
hipsolverDnXgesvdjGetSweeps()
3.2.4. Syevj parameter manipulation
hipsolverDnCreateSyevjInfo()
hipsolverDnDestroySyevjInfo()
hipsolverDnXsyevjSetMaxSweeps()
hipsolverDnXsyevjSetSortEig()
hipsolverDnXsyevjSetTolerance()
hipsolverDnXsyevjGetResidual()
hipsolverDnXsyevjGetSweeps()
3.3. LAPACK Auxiliary Functions
3.3.1. Orthonormal matrices
hipsolverDn<type>orgbr_bufferSize()
hipsolverDn<type>orgbr()
hipsolverDn<type>orgqr_bufferSize()
hipsolverDn<type>orgqr()
hipsolverDn<type>orgtr_bufferSize()
hipsolverDn<type>orgtr()
hipsolverDn<type>ormqr_bufferSize()
hipsolverDn<type>ormqr()
hipsolverDn<type>ormtr_bufferSize()
hipsolverDn<type>ormtr()
3.3.2. Unitary matrices
hipsolverDn<type>ungbr_bufferSize()
hipsolverDn<type>ungbr()
hipsolverDn<type>ungqr_bufferSize()
hipsolverDn<type>ungqr()
hipsolverDn<type>ungtr_bufferSize()
hipsolverDn<type>ungtr()
hipsolverDn<type>unmqr_bufferSize()
hipsolverDn<type>unmqr()
hipsolverDn<type>unmtr_bufferSize()
hipsolverDn<type>unmtr()
3.4. LAPACK Functions
3.4.1. Triangular factorizations
hipsolverDn<type>potrf_bufferSize()
hipsolverDn<type>potrf()
hipsolverDn<type>potrfBatched()
hipsolverDn<type>getrf_bufferSize()
hipsolverDn<type>getrf()
hipsolverDn<type>sytrf_bufferSize()
hipsolverDn<type>sytrf()
3.4.2. Orthogonal factorizations
hipsolverDn<type>geqrf_bufferSize()
hipsolverDn<type>geqrf()
3.4.3. Problem and matrix reductions
hipsolverDn<type>gebrd_bufferSize()
hipsolverDn<type>gebrd()
hipsolverDn<type>sytrd_bufferSize()
hipsolverDn<type>hetrd_bufferSize()
hipsolverDn<type>sytrd()
hipsolverDn<type>hetrd()
3.4.4. Linear-systems solvers
hipsolverDn<type>potri_bufferSize()
hipsolverDn<type>potri()
hipsolverDn<type>potrs()
hipsolverDn<type>potrsBatched()
hipsolverDn<type>getrs()
hipsolverDn<type><type>gesv_bufferSize()
hipsolverDn<type><type>gesv()
3.4.5. Least-squares solvers
hipsolverDn<type><type>gels_bufferSize()
hipsolverDn<type><type>gels()
3.4.6. Symmetric eigensolvers
hipsolverDn<type>syevd_bufferSize()
hipsolverDn<type>heevd_bufferSize()
hipsolverDn<type>syevd()
hipsolverDn<type>heevd()
hipsolverDn<type>syevj_bufferSize()
hipsolverDn<type>heevj_bufferSize()
hipsolverDn<type>syevjBatched_bufferSize()
hipsolverDn<type>heevjBatched_bufferSize()
hipsolverDn<type>syevj()
hipsolverDn<type>heevj()
hipsolverDn<type>syevjBatched()
hipsolverDn<type>heevjBatched()
hipsolverDn<type>sygvd_bufferSize()
hipsolverDn<type>hegvd_bufferSize()
hipsolverDn<type>sygvd()
hipsolverDn<type>hegvd()
hipsolverDn<type>sygvj_bufferSize()
hipsolverDn<type>hegvj_bufferSize()
hipsolverDn<type>sygvj()
hipsolverDn<type>hegvj()
3.4.7. Singular value decomposition
hipsolverDn<type>gesvd_bufferSize()
hipsolverDn<type>gesvd()
hipsolverDn<type>gesvdj_bufferSize()
hipsolverDn<type>gesvdjBatched_bufferSize()
hipsolverDn<type>gesvdj()
hipsolverDn<type>gesvdjBatched()
hipSOLVER
»
1.
hipSOLVER User Guide
Edit on GitHub
Next
Previous
1.
hipSOLVER User Guide
¶
1.1. Introduction
1.1.1. Library overview
1.1.2. Currently implemented functionality
LAPACK auxiliary functions
LAPACK main functions
1.1.3. Compatibility-only functions
Iterative Jacobi functions
1.2. Installation
1.2.1. Install pre-built packages
1.2.2. Build & install library using script (Ubuntu only)
1.2.3. Build & install library manually
1.2.4. Build library + tests + benchmarks + samples manually
Dependencies (only necessary for hipSOLVER clients)
Library and clients
1.3. Using hipSOLVER
1.3.1. Some considerations when using the regular hipSOLVER API
Different signatures and additional API methods
Unsupported methods
Arguments not referenced by rocSOLVER
Using rocSOLVER’s memory model
Using rocSOLVER’s in-place functions
1.3.2. Porting cuSOLVER applications to hipSOLVER
Possible performance implications of the compatibility API
Read the Docs
v: rocm-5.2.1
Versions
master
latest
rocm-5.2.1
rocm-5.2.0
rocm-5.1.3
rocm-5.1.1
rocm-5.1.0
rocm-5.0.2
rocm-5.0.1
rocm-5.0.0
rocm-4.5.2
rocm-4.5.0
Downloads
On Read the Docs
Project Home
Builds