Enum cusolverIRSRefinement_t 

#[repr(u32)]
pub enum cusolverIRSRefinement_t {
    CUSOLVER_IRS_REFINE_NOT_SET = 1_100,
    CUSOLVER_IRS_REFINE_NONE = 1_101,
    CUSOLVER_IRS_REFINE_CLASSICAL = 1_102,
    CUSOLVER_IRS_REFINE_CLASSICAL_GMRES = 1_103,
    CUSOLVER_IRS_REFINE_GMRES = 1_104,
    CUSOLVER_IRS_REFINE_GMRES_GMRES = 1_105,
    CUSOLVER_IRS_REFINE_GMRES_NOPCOND = 1_106,
    CUSOLVER_PREC_DD = 1_150,
    CUSOLVER_PREC_SS = 1_151,
    CUSOLVER_PREC_SHT = 1_152,
}

The cusolverIRSRefinement_t type indicates which solver type would be used for the specific cusolver function. Most of our experimentation shows that cusolverIRSRefinement_t::CUSOLVER_IRS_REFINE_GMRES is the best option.

More details about the refinement process can be found in Azzam Haidar, Stanimire Tomov, Jack Dongarra, and Nicholas J. Higham. 2018. Harnessing GPU tensor cores for fast FP16 arithmetic to speed up mixed-precision iterative refinement solvers. In Proceedings of the International Conference for High Performance Computing, Networking, Storage, and Analysis (SC ‘18). IEEE Press, Piscataway, NJ, USA, Article 47, 11 pages..

Variants

CUSOLVER_IRS_REFINE_NOT_SET = 1_100

Solver is not set; this value is what is set when creating the params structure. IRS solver will return an error.

CUSOLVER_IRS_REFINE_NONE = 1_101

No refinement solver, the IRS solver performs a factorization followed by a solve without any refinement. For example if the IRS solver was cusolverDnIRSXgesv, this is equivalent to a Xgesv routine without refinement and where the factorization is carried out in the lowest precision. If for example the main precision was CUSOLVER_R_64F and the lowest was CUSOLVER_R_64F as well, then this is equivalent to a call to cusolverDnDgesv().

CUSOLVER_IRS_REFINE_CLASSICAL = 1_102

Classical iterative refinement solver. Similar to the one used in LAPACK routines.

CUSOLVER_IRS_REFINE_CLASSICAL_GMRES = 1_103

Classical iterative refinement solver that uses the GMRES (Generalized Minimal Residual) internally to solve the correction equation at each iteration. We call the classical refinement iteration the outer iteration while the GMRES is called inner iteration. Note that if the tolerance of the inner GMRES is set very low, lets say to machine precision, then the outer classical refinement iteration will performs only one iteration and thus this option will behave like cusolverIRSRefinement_t::CUSOLVER_IRS_REFINE_GMRES.

CUSOLVER_IRS_REFINE_GMRES = 1_104

GMRES (Generalized Minimal Residual) based iterative refinement solver. In recent study, the GMRES method has drawn the scientific community attention for its ability to be used as refinement solver that outperforms the classical iterative refinement method. Based on our experimentation, we recommend this setting.

CUSOLVER_IRS_REFINE_GMRES_GMRES = 1_105

Similar to cusolverIRSRefinement_t::CUSOLVER_IRS_REFINE_CLASSICAL_GMRES which consists of classical refinement process that uses GMRES to solve the inner correction system; here it is a GMRES (Generalized Minimal Residual) based iterative refinement solver that uses another GMRES internally to solve the preconditioned system.

CUSOLVER_IRS_REFINE_GMRES_NOPCOND = 1_106

CUSOLVER_PREC_DD = 1_150

CUSOLVER_PREC_SS = 1_151

CUSOLVER_PREC_SHT = 1_152

Trait Implementations

impl Clone for cusolverIRSRefinement_t

fn clone(&self) -> cusolverIRSRefinement_t

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for cusolverIRSRefinement_t

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl From<cusolverIRSRefinement_t> for u32

fn from(enum_value: cusolverIRSRefinement_t) -> Self

Converts to this type from the input type.

impl Hash for cusolverIRSRefinement_t

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl Ord for cusolverIRSRefinement_t

fn cmp(&self, other: &cusolverIRSRefinement_t) -> Ordering

This method returns an Ordering between self and other.

fn max(self, other: Self) -> Self

where Self: Sized,

Compares and returns the maximum of two values.

fn min(self, other: Self) -> Self

where Self: Sized,

Compares and returns the minimum of two values.

fn clamp(self, min: Self, max: Self) -> Self

where Self: Sized,

Restrict a value to a certain interval.

impl PartialEq for cusolverIRSRefinement_t

fn eq(&self, other: &cusolverIRSRefinement_t) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl PartialOrd for cusolverIRSRefinement_t

fn partial_cmp(&self, other: &cusolverIRSRefinement_t) -> Option<Ordering>

This method returns an ordering between self and other values if one exists.

fn lt(&self, other: &Rhs) -> bool

Tests less than (for self and other) and is used by the < operator.

fn le(&self, other: &Rhs) -> bool

Tests less than or equal to (for self and other) and is used by the <= operator.

fn gt(&self, other: &Rhs) -> bool

Tests greater than (for self and other) and is used by the > operator.

fn ge(&self, other: &Rhs) -> bool

Tests greater than or equal to (for self and other) and is used by the >= operator.

impl TryFrom<u32> for cusolverIRSRefinement_t

type Error = TryFromPrimitiveError<cusolverIRSRefinement_t>

The type returned in the event of a conversion error.

fn try_from(number: u32) -> Result<Self, TryFromPrimitiveError<Self>>

Performs the conversion.

impl TryFromPrimitive for cusolverIRSRefinement_t

const NAME: &'static str = "cusolverIRSRefinement_t"

type Primitive = u32

type Error = TryFromPrimitiveError<cusolverIRSRefinement_t>

fn try_from_primitive( number: Self::Primitive, ) -> Result<Self, TryFromPrimitiveError<Self>>

impl Copy for cusolverIRSRefinement_t

impl Eq for cusolverIRSRefinement_t

impl StructuralPartialEq for cusolverIRSRefinement_t