Sacado Package Browser (Single Doxygen Collection)  Version of the Day
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
view/TestAssembly.hpp
Go to the documentation of this file.
1 // @HEADER
2 // ***********************************************************************
3 //
4 // Stokhos Package
5 // Copyright (2009) Sandia Corporation
6 //
7 // Under terms of Contract DE-AC04-94AL85000, there is a non-exclusive
8 // license for use of this work by or on behalf of the U.S. Government.
9 //
10 // Redistribution and use in source and binary forms, with or without
11 // modification, are permitted provided that the following conditions are
12 // met:
13 //
14 // 1. Redistributions of source code must retain the above copyright
15 // notice, this list of conditions and the following disclaimer.
16 //
17 // 2. Redistributions in binary form must reproduce the above copyright
18 // notice, this list of conditions and the following disclaimer in the
19 // documentation and/or other materials provided with the distribution.
20 //
21 // 3. Neither the name of the Corporation nor the names of the
22 // contributors may be used to endorse or promote products derived from
23 // this software without specific prior written permission.
24 //
25 // THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
26 // EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28 // PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
29 // CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
30 // EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
31 // PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
32 // PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
33 // LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
34 // NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
35 // SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
36 //
37 // Questions? Contact Eric T. Phipps (etphipp@sandia.gov).
38 //
39 // ***********************************************************************
40 // @HEADER
41 #include <iostream>
42 
43 // Utilities
46 
47 // FENL
48 #include <BoxElemFixture.hpp>
49 #include <fenl_functors.hpp>
50 
51 struct Perf {
52  size_t global_elem_count ;
53  size_t global_node_count ;
54  double fill_time ;
55 
58  fill_time(0) {}
59 
60  void increment(const Perf& p) {
63  fill_time += p.fill_time;
64  }
65 
66  void scale(double s) {
67  fill_time *= s;
68  }
69 };
70 
71 template <typename Scalar, typename Device,
75  const int use_print ,
76  const int use_trials ,
77  const int use_nodes[] ,
78  Kokkos::View< Scalar* , Device >& residual,
80 {
81  using Teuchos::RCP;
82  using Teuchos::rcp;
83  using Teuchos::rcpFromRef;
84  using Teuchos::arrayView;
85 
87 
89  typedef typename LocalMatrixType::StaticCrsGraphType LocalGraphType ;
90 
92 
94 
95  typedef typename ElementComputationType::vector_type VectorType ;
96 
97  //------------------------------------
98 
99  // Decompose by node to avoid parallel communication in assembly
100 
101  const double bubble_x = 1.0 ;
102  const double bubble_y = 1.0 ;
103  const double bubble_z = 1.0 ;
104 
105  const FixtureType fixture( Kokkos::Example::BoxElemPart::DecomposeNode ,
106  1 , 0 ,
107  use_nodes[0] , use_nodes[1] , use_nodes[2] ,
108  bubble_x , bubble_y , bubble_z );
109 
110  //------------------------------------
111 
112  Kokkos::Impl::Timer wall_clock ;
113 
114  Perf perf_stats = Perf() ;
115 
116  for ( int itrial = 0 ; itrial < use_trials ; ++itrial ) {
117 
118  Perf perf = Perf() ;
119 
120  perf.global_elem_count = fixture.elem_count_global();
121  perf.global_node_count = fixture.node_count_global();
122 
123  //----------------------------------
124  // Create the local sparse matrix graph and element-to-graph map
125  // from the element->to->node identifier array.
126  // The graph only has rows for the owned nodes.
127 
128  typename NodeNodeGraphType::Times graph_times;
129  const NodeNodeGraphType
130  mesh_to_graph( fixture.elem_node() , fixture.node_count_owned(),
131  graph_times );
132 
133  // Create the local sparse matrix from the graph:
134  jacobian = LocalMatrixType( mesh_to_graph.graph );
135 
136  //----------------------------------
137 
138  // Allocate solution vector for each node in the mesh and residual vector for each owned node
139  VectorType solution( "solution" , fixture.node_count() );
140  residual = VectorType( "residual" , fixture.node_count_owned() );
141 
142  // Create element computation functor
143  const ElementComputationType elemcomp( fixture , solution ,
144  mesh_to_graph.elem_graph ,
145  jacobian , residual );
146 
147  Kokkos::deep_copy( solution , Scalar(1.2345) );
148 
149  //--------------------------------
150  // Element contributions to residual and jacobian
151 
152  Kokkos::deep_copy( residual , Scalar(0) );
153  Kokkos::deep_copy( jacobian.coeff , Scalar(0) );
154 
155  wall_clock.reset();
156 
157  elemcomp.apply();
158 
159  Device().fence();
160  perf.fill_time = wall_clock.seconds();
161 
162  //--------------------------------
163 
164  perf_stats.increment(perf);
165 
166  }
167 
168  return perf_stats ;
169 }
170 
171 template<class ValueType>
172 bool compareValues(const ValueType& a1,
173  const std::string& a1_name,
174  const ValueType&a2,
175  const std::string& a2_name,
176  const ValueType& rel_tol, const ValueType& abs_tol,
178 {
179  bool success = true;
180 
181  ValueType err = std::abs(a1 - a2);
182  ValueType tol = abs_tol + rel_tol*std::max(std::abs(a1),std::abs(a2));
183  if (err > tol) {
184  out << "\nError, relErr(" << a1_name <<","
185  << a2_name << ") = relErr(" << a1 <<"," << a2 <<") = "
186  << err << " <= tol = " << tol << ": failed!\n";
187  success = false;
188  }
189 
190  return success;
191 }
192 
193 template <typename VectorType, typename MatrixType>
194 bool check_assembly(const VectorType& analytic_residual,
195  const MatrixType& analytic_jacobian,
196  const VectorType& fad_residual,
197  const MatrixType& fad_jacobian,
198  const std::string& test_name)
199 {
200  const double tol = 1e-14;
201  bool success = true;
203  Teuchos::VerboseObjectBase::getDefaultOStream();
204  std::stringstream buf;
205  Teuchos::FancyOStream fbuf(Teuchos::rcp(&buf,false));
206 
207  typename VectorType::HostMirror host_analytic_residual =
208  Kokkos::create_mirror_view(analytic_residual);
209  typename VectorType::HostMirror host_fad_residual =
210  Kokkos::create_mirror_view(fad_residual);
211  Kokkos::deep_copy( host_analytic_residual, analytic_residual );
212  Kokkos::deep_copy( host_fad_residual, fad_residual );
213 
214  fbuf << test_name << ":" << std::endl;
215 
216  if (host_analytic_residual.extent(0) != host_fad_residual.extent(0)) {
217  fbuf << "Analytic residual dimension "
218  << host_analytic_residual.extent(0)
219  << " does not match Fad residual dimension "
220  << host_fad_residual.extent(0) << std::endl;
221  success = false;
222  }
223  else {
224  const size_t num_node = host_analytic_residual.extent(0);
225  for (size_t i=0; i<num_node; ++i) {
226  success = success && compareValues(
227  host_analytic_residual(i), "analytic residual",
228  host_fad_residual(i), "Fad residual",
229  tol, tol, fbuf );
230  }
231  }
232 
233  typename MatrixType::HostMirror host_analytic_jacobian =
234  Kokkos::create_mirror_view(analytic_jacobian);
235  typename MatrixType::HostMirror host_fad_jacobian =
236  Kokkos::create_mirror_view(fad_jacobian);
237  Kokkos::deep_copy( host_analytic_jacobian, analytic_jacobian );
238  Kokkos::deep_copy( host_fad_jacobian, fad_jacobian );
239 
240  if (host_analytic_jacobian.extent(0) != host_fad_jacobian.extent(0)) {
241  fbuf << "Analytic Jacobian dimension "
242  << host_analytic_jacobian.extent(0)
243  << " does not match Fad Jacobian dimension "
244  << host_fad_jacobian.extent(0) << std::endl;
245  success = false;
246  }
247  else {
248  const size_t num_entry = host_analytic_jacobian.extent(0);
249  for (size_t i=0; i<num_entry; ++i) {
250  success = success && compareValues(
251  host_analytic_jacobian(i), "analytic Jacobian",
252  host_fad_jacobian(i), "Fad Jacobian",
253  tol, tol, fbuf );
254  }
255  }
256 
257  if (!success)
258  *out << buf.str();
259 
260  return success;
261 }
262 
263 template <class Device>
265  const int use_print ,
266  const int use_trials ,
267  const int n_begin ,
268  const int n_end ,
269  const int n_step ,
270  const bool quadratic ,
271  const bool check )
272 {
278 
279  std::cout.precision(8);
280  std::cout << std::endl
281  << "\"Grid Size\" , "
282  << "\"FEM Size\" , "
283  << "\"Analytic Fill Time\" , "
284  << "\"Fad Element Fill Slowdown\" , "
285  << "\"Fad Optimized Element Fill Slowdown\" , "
286  << "\"Fad QP Fill Slowdown\" , "
287  << std::endl;
288 
289  typedef Kokkos::View< double* , Device > vector_type ;
291  vector_type analytic_residual, fad_residual, fad_opt_residual,
292  fad_qp_residual;
293  matrix_type analytic_jacobian, fad_jacobian, fad_opt_jacobian,
294  fad_qp_jacobian;
295 
296  for (int n=n_begin; n<=n_end; n+=n_step) {
297  const int use_nodes[] = { n, n, n };
298  Perf perf_analytic, perf_fad, perf_fad_opt, perf_fad_qp;
299 
300  if (quadratic) {
301  perf_analytic =
302  fenl_assembly<double,Device,BoxElemPart::ElemQuadratic,Analytic>(
303  use_print, use_trials, use_nodes,
304  analytic_residual, analytic_jacobian );
305 
306  perf_fad =
307  fenl_assembly<double,Device,BoxElemPart::ElemQuadratic,FadElement>(
308  use_print, use_trials, use_nodes,
309  fad_residual, fad_jacobian);
310 
311  perf_fad_opt =
312  fenl_assembly<double,Device,BoxElemPart::ElemQuadratic,FadElementOptimized>(
313  use_print, use_trials, use_nodes,
314  fad_opt_residual, fad_opt_jacobian);
315 
316  perf_fad_qp =
317  fenl_assembly<double,Device,BoxElemPart::ElemQuadratic,FadQuadPoint>(
318  use_print, use_trials, use_nodes,
319  fad_qp_residual, fad_qp_jacobian);
320  }
321  else {
322  perf_analytic =
323  fenl_assembly<double,Device,BoxElemPart::ElemLinear,Analytic>(
324  use_print, use_trials, use_nodes,
325  analytic_residual, analytic_jacobian );
326 
327  perf_fad =
328  fenl_assembly<double,Device,BoxElemPart::ElemLinear,FadElement>(
329  use_print, use_trials, use_nodes,
330  fad_residual, fad_jacobian);
331 
332  perf_fad_opt =
333  fenl_assembly<double,Device,BoxElemPart::ElemLinear,FadElementOptimized>(
334  use_print, use_trials, use_nodes,
335  fad_opt_residual, fad_opt_jacobian);
336 
337  perf_fad_qp =
338  fenl_assembly<double,Device,BoxElemPart::ElemLinear,FadQuadPoint>(
339  use_print, use_trials, use_nodes,
340  fad_qp_residual, fad_qp_jacobian);
341  }
342  if (check) {
343  check_assembly( analytic_residual, analytic_jacobian.coeff,
344  fad_residual, fad_jacobian.coeff,
345  "Fad" );
346  check_assembly( analytic_residual, analytic_jacobian.coeff,
347  fad_opt_residual, fad_opt_jacobian.coeff,
348  "Optimized Fad" );
349  check_assembly( analytic_residual, analytic_jacobian.coeff,
350  fad_qp_residual, fad_qp_jacobian.coeff,
351  "QP Fad" );
352  }
353 
354  double s =
355  1000.0 / ( use_trials * perf_analytic.global_elem_count );
356  perf_analytic.scale(s);
357  perf_fad.scale(s);
358  perf_fad_opt.scale(s);
359  perf_fad_qp.scale(s);
360 
361  std::cout.precision(3);
362  std::cout << n << " , "
363  << perf_analytic.global_node_count << " , "
364  << std::setw(2)
365  << std::scientific
366  << perf_analytic.fill_time << " , "
367  << std::fixed << std::setw(6)
368  << perf_fad.fill_time / perf_analytic.fill_time << " , "
369  << perf_fad_opt.fill_time / perf_analytic.fill_time << " , "
370  << perf_fad_qp.fill_time / perf_analytic.fill_time << " , "
371  << std::endl;
372  }
373 }
Perf fenl_assembly(const int use_print, const int use_trials, const int use_nodes[], Kokkos::View< Scalar *, Device > &residual, Kokkos::Example::FENL::CrsMatrix< Scalar, Device > &jacobian)
size_t global_node_count
abs(expr.val())
std::enable_if< !Kokkos::is_view_fad< View2 >::value, bool >::type check(const View1 &v_gold, const View2 &v, const double tol)
bool compareValues(const ValueType &a1, const std::string &a1_name, const ValueType &a2, const std::string &a2_name, const ValueType &rel_tol, const ValueType &abs_tol, Teuchos::FancyOStream &out)
double fill_time
TEUCHOS_DEPRECATED RCP< T > rcp(T *p, Dealloc_T dealloc, bool owns_mem)
size_t global_elem_count
void performance_test_driver(const int use_print, const int use_trials, const int n_begin, const int n_end, const int n_step, const bool quadratic, const bool check)
void increment(const Perf &p)
void scale(double s)
const double tol
bool check_assembly(const VectorType &analytic_residual, const MatrixType &analytic_jacobian, const VectorType &fad_residual, const MatrixType &fad_jacobian, const std::string &test_name)
SimpleFad< ValueT > max(const SimpleFad< ValueT > &a, const SimpleFad< ValueT > &b)
Partition a box of hexahedral elements among subdomains.
int n
Generate a distributed unstructured finite element mesh from a partitioned NX*NY*NZ box of elements...