Eigen-Contrib  5.0.1-dev+master
 
Loading...
Searching...
No Matches
Eigen::gpu::SparseSolverConfig Struct Reference

#include <contrib/Eigen/src/GPU/GpuSparseSolverBase.h>

Detailed Description

Pass-through configuration for the cuDSS-backed sparse direct solvers (SparseLLT, SparseLDLT, SparseLU). Fields left at their defaults keep the corresponding cuDSS default, which is tuned for performance rather than maximum robustness. Non-default fields require cuDSS >= 0.8, whose cudssReorderingAlg_t etc. name the algorithms with stable values; earlier versions declare no algorithm enumerator and accept only a default config, which EIGEN_HAS_CUDSS_SOLVER_CONFIG reports.

Public Member Functions

bool isDefault () const
 

Public Attributes

bool hybridExecute
 
bool hybridMemory
 
int64_t hybridMemoryDeviceLimit
 
double pivotEpsilon
 
double pivotThreshold
 
int refinementSteps
 
double refinementTolerance
 

Member Function Documentation

◆ isDefault()

bool Eigen::gpu::SparseSolverConfig::isDefault ( ) const
inline

True when every field keeps the cuDSS default.

Member Data Documentation

◆ hybridExecute

bool Eigen::gpu::SparseSolverConfig::hybridExecute

Split factorization/solve work between host and device. cuDSS requires this before analysis, which it changes — set it before analyzePattern().

◆ hybridMemory

bool Eigen::gpu::SparseSolverConfig::hybridMemory

Let factor data spill to host memory when device memory is insufficient. Changes what analyzePattern() builds — set it before that phase.

◆ hybridMemoryDeviceLimit

int64_t Eigen::gpu::SparseSolverConfig::hybridMemoryDeviceLimit

Device-memory budget in bytes for hybridMemory; negative keeps the cuDSS default heuristic, which is what cuDSS itself spells as -1. Zero is a budget of its own and is passed through.

◆ pivotEpsilon

double Eigen::gpu::SparseSolverConfig::pivotEpsilon

Perturbation applied to near-zero pivots; negative keeps the cuDSS default.

◆ pivotThreshold

double Eigen::gpu::SparseSolverConfig::pivotThreshold

Pivot admissibility threshold in [0, 1]; negative keeps the cuDSS default.

◆ refinementSteps

int Eigen::gpu::SparseSolverConfig::refinementSteps

Iterative-refinement steps during solve(); negative keeps the cuDSS default (0).

◆ refinementTolerance

double Eigen::gpu::SparseSolverConfig::refinementTolerance

Iterative-refinement stopping tolerance; negative keeps the cuDSS default.


The documentation for this struct was generated from the following file: