PrintPerf(3P) Sun Performance Library PrintPerf(3P)NAME
PrintPerf: cPrintPerf, dPrintPerf, sPrintPerf, zPrintPerf - A utility
function of the SuperLU solver that prints statistics collected by the
computational routines.
SYNOPSIS
#include <sunperf.h>
void sPrintPerf(SuperMatrix *L, SuperMatrix *U, mem_usage_t *mem_usage,
float rpg, float rcond, float *ferr, float *berr, char
*equed, SuperLUStat_t *stat);
void dPrintPerf(SuperMatrix *L, SuperMatrix *U, mem_usage_t *mem_usage,
double rpg, double rcond, double *ferr, double *berr, char
*equed, SuperLUStat_t *stat);
void cPrintPerf(SuperMatrix *L, SuperMatrix *U, mem_usage_t *mem_usage,
complex rpg, complex rcond, complex *ferr, complex *berr,
char *equed, SuperLUStat_t *stat);
void zPrintPerf(SuperMatrix *L, SuperMatrix *U, mem_usage_t *mem_usage,
doublecomplex rpg, doublecomplex rcond, doublecomplex *ferr,
doublecomplex *berr, char *equed, SuperLUStat_t *stat);
void sPrintPerf_64(SuperMatrix_64 *L, SuperMatrix_64 *U, mem_usage_t_64
*mem_usage, float rpg, float rcond, float *ferr, float *berr,
char *equed, SuperLUStat_t_64 *stat);
void dPrintPerf_64(SuperMatrix_64 *L, SuperMatrix_64 *U, mem_usage_t_64
*mem_usage, double rpg, double rcond, double *ferr, double
*berr, char *equed, SuperLUStat_t_64 *stat);
void cPrintPerf_64(SuperMatrix_64 *L, SuperMatrix_64 *U, mem_usage_t_64
*mem_usage, complex rpg, complex rcond, complex *ferr, com‐
plex *berr, char *equed, SuperLUStat_t_64 *stat);
void zPrintPerf_64(SuperMatrix_64 *L, SuperMatrix_64 *U, mem_usage_t_64
*mem_usage, doublecomplex rpg, doublecomplex rcond, double‐
complex *ferr, doublecomplex *berr, char *equed, SuperLUS‐
tat_t_64 *stat);
PURPOSEPrintPerf() is called following the execution of expert driver routine
sgssvx to print various statistics collected during the computation.
Statistics printed include:
Floating point operation count of the factorization
Floating point operation count of the solve
Number of nonzeros in factor L
Number of nonzeros in factor U
Total memory needed (in MB) for factors L and U
Total memory needed (in MB) needed for the factorization
Amount of memory allocated (in MB) in the expansions
Time and Megaflop rate of the factorization
Time and Megaflop rate of the solve
The reciprocal pivot growth factor
The estimate of the reciprocal condition number of the input matrix A
The estimated forward error bound for the solution vector
The componentwise relative backward error of the solution vector
Whether equilibration was performed
ARGUMENTS
SuperMatrix *L (input)
Factor L from the factorization
SuperMatrix *U (input)
Factor U from the factorization
mem_usage_t *mem_usage (input)
Data structure holding the memory requirement of the factor‐
ization
float rpg (input)
The reciprocal pivot growth factor computed during the fac‐
torization
float rcond (input)
The estimate of the reciprocal condition number of the matrix
A after equilibration (if done).
float *ferr (input)
Address of the estimated forward error bound for a particular
solution vector
float *berr (input)
Address of the componentwise relative backward error of a
particular solution vector
char *equed (input)
Address of the flag that specified the form of equilibration
that was done
SuperLUStat_t *stat (input)
Data structure that holds the statistics collected during the
computation
COPYRIGHT
Copyright (c) 2003, The Regents of the University of California,
through Lawrence Berkeley National Laboratory (subject to receipt of
any required approvals from U.S. Dept. of Energy)
SEE ALSO
SuperMatrix
StatInit
QuerySpace
gssvx
http://crd.lbl.gov/~xiaoye/SuperLU/
James W. Demmel, Stanley C. Eisenstat, John R. Gilbert, Xiaoye S. Li
and Joseph W. H. Liu, "A supernodal approach to sparse partial pivot‐
ing", SIAM J. Matrix Analysis and Applications, Vol. 20, Num. 3, 1999,
pp. 720-755.
6 Mar 2009 PrintPerf(3P)
