1hpcg_kernel(1) Utility Commands hpcg_kernel(1)
2
6 hpcg_kernel - high performance conjugate gradient kernel benchmark
7
10 hpcg_kernel matrix_type solution_filename rhistory_filename [options]
11
14 This program solves the linear equation Ax = b with additive Schwarz,
15 symmetric Gauss-Seidel preconditioned conjugate gradient solver, where
16 the coefficient matrix A of size lmn is derived from a discretized
17 three dimensional Poisson's equation using the twenty-seven point cen‐
18 tral difference scheme, with the coefficient matrix in the storage for‐
19 mat specified by matrix_type and the solver specified by options. It
20 outputs the solution to solution_filename in the extended Matrix Market
21 format and the residual history to rhistory_filename in the PLAIN for‐
22 mat (see Appendix of the Lis User Guide). The right-hand side vector
23 is set such that the values of the elements of the solution are 1. The
24 values l, m and n represent the numbers of grid points in each dimen‐
25 sion.
26
29 The following options are supported:
30 -i linear solver
32 The following options are supported for linear solver:
33 -i {cg|1}
35 CG
36 -i {bicg|2}
38 BiCG
39 -i {cgs|3}
41 CGS
42 -i {bicgstab|4}
44 BiCGSTAB
45 -i {bicgstabl|5}
47 BiCGSTAB(l)
48 -ell [2]
50 The degree l
51 -i {gpbicg|6}
53 GPBiCG
54 -i {tfqmr|7}
56 TFQMR
57 -i {orthomin|8}
59 Orthomin(m)
60 -restart [40]
62 The restart value m
63 -i {gmres|9}
65 GMRES(m)
66 -restart [40]
68 The restart value m
69 -i {jacobi|10}
71 Jacobi
72 -i {gs|11}
74 Gauss-Seidel
75 -i {sor|12}
77 SOR
78 -omega [1.9]
80 The relaxation coefficient omega (0<omega<2)
81 -i {bicgsafe|13}
83 BiCGSafe
84 -i {cr|14}
86 CR
87 -i {bicr|15}
89 BiCR
90 -i {crs|16}
92 CRS
93 -i {bicrstab|17}
95 BiCRSTAB
96 -i {gpbicr|18}
98 GPBiCR
99 -i {bicrsafe|19}
101 BiCRSafe
102 -i {fgmres|20}
104 FGMRES(m)
105 -restart [40]
107 The restart value m
108 -i {idrs|21}
110 IDR(s)
111 -irestart [2]
113 The restart value s
114 -i {idr1|22}
116 IDR(1)
117 -i {minres|23}
119 MINRES
120 -i {COCG|24}
122 COCG
123 -i {COCR|25}
125 COCR
126 -p preconditioner
130 The following options are supported for preconditioner:
131 -p {none|0}
133 None
134 -p {jacobi|1}
136 Jacobi
137 -p {ilu|2}
139 ILU(k)
140 -ilu_fill [0]
142 The fill level k
143 -p {ssor|3}
145 SSOR
146 -ssor_omega [1.0]
148 The relaxation coefficient omega (0<omega<2)
149 -p {hybrid|4}
151 Hybrid
152 -hybrid_i [sor]
154 The linear solver
155 -hybrid_maxiter [25]
157 The maximum number of the iterations
158 -hybrid_tol [1.0e-3]
160 The convergence criterion
161 -hybrid_omega [1.5]
163 The relaxation coefficient omega of the SOR
164 (0<omega<2)
165 -hybrid_ell [2]
167 The degree l of the BiCGSTAB(l)
168 -hybrid_restart [40]
170 The restart values of the GMRES and Orthomin
171 -p {is|5}
173 I+S
174 -is_alpha [1.0]
176 The parameter alpha of I+alpha*S(m)
177 -is_m [3]
179 The parameter m of I+alpha*S(m)
180 -p {sainv|6}
182 SAINV
183 -sainv_drop [0.05]
185 The drop criterion
186 -p {saamg|7}
188 SA-AMG
189 -saamg_unsym [false]
191 Select the unsymmetric version (The matrix struc‐
192 ture must be symmetric)
193 -saamg_theta [0.05|0.12]
195 The drop criterion
196 -p {iluc|8}
198 Crout ILU
199 -iluc_drop [0.05]
201 The drop criterion
202 -iluc_rate [5.0]
204 The ration of maximum fill-in
205 -p {ilut|9}
207 ILUT
208 -ilut_drop [0.05]
210 The drop criterion
211 -ilut_rate [5.0]
213 The ration of maximum fill-in
214 -adds true
216 Additive Schwarz
217 -adds_iter [1]
219 The number of the iteration
220 Other Options:
222 -maxiter [1000]
224 The maximum number of the iterations
225 -tol [1.0e-12]
227 The convergence criterion
228 -print [0]
230 The display of the residual history
231 -print {none|0}
233 None
234 -print {mem|1}
236 Save the residual history
237 -print {out|2}
239 Display the residual history
240 -print {all|3}
242 Save the residual history and output it to the standard
243 output
244 -scale [0]
246 The scaling
247 -scale {none|0}
249 No scaling
250 -scale {jacobi|1}
252 The Jacobi scaling
253 -scale {symm_diag|2}
255 The diagonal scaling
256 -initx_zeros [true]
258 The behavior of the initial vector x_0
259 -initx_zero {false|0}
261 Given values
262 -initx_zero {true|1}
264 All values are set to 0
265 -omp_num_threads [t]
267 The number of the threads (t represents the maximum number of
268 the threads)
269 -storage [0]
271 The matrix storage format
272 -storage_block [2]
274 The block size of the BSR and BSC formats
275 -f [0] The precision of the linear solver
277 -f {double|0}
279 Double precision
280 -f {quad|1}
282 Double-double (quadruple) precision
283 See Lis User Guide for full description.
285
288 The following exit values are returned:
289 0 The process is normally terminated
291 unspecified
293 An error occurred
294
297 lis(3), lsolve(1), esolve(1), hpcg_spmvtest(1), spmvtest1(1),
298 spmvtest2(1), spmvtest2b(1), spmvtest3(1), spmvtest3b(1), spmvtest4(1),
299 spmvtest5(1)
300 http://www.ssisc.org/lis/
302 http://software.sandia.gov/hpcg/
303Man Page 14 Sep 2017 hpcg_kernel(1)