Sacado Package Browser (Single Doxygen Collection)  Version of the Day
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
TestAssembly.cpp
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 
42 #include <iostream>
43 
44 // Tests
45 #include "TestAssembly.hpp"
46 
47 // Devices
48 #include "Kokkos_Core.hpp"
49 
50 // Utilities
53 #ifdef KOKKOS_ENABLE_CUDA
54 #include "cuda_runtime_api.h"
55 #endif
56 
57 // For vtune
58 #include <sys/types.h>
59 #include <unistd.h>
60 
61 int main(int argc, char *argv[])
62 {
63  bool success = true;
64  bool verbose = false;
65  try {
66 
67  // Setup command line options
69  CLP.setDocString(
70  "This test performance of MP::Vector FEM assembly.\n");
71  int nGrid = 0;
72  CLP.setOption("n", &nGrid, "Number of mesh points in each direction. Set to zero to use a range");
73  int nGridBegin = 8;
74  CLP.setOption("n-begin", &nGridBegin, "Beginning number of mesh points in each direction.");
75  int nGridEnd = 48;
76  CLP.setOption("n-end", &nGridEnd, "Ending number of mesh points in each direction.");
77  int nGridStep = 8;
78  CLP.setOption("n-step", &nGridStep, "Increment in number of mesh points in each direction.");
79  int nIter = 10;
80  CLP.setOption("ni", &nIter, "Number of assembly iterations");
81  bool print = false;
82  CLP.setOption("print", "no-print", &print, "Print debugging output");
83  bool check = false;
84  CLP.setOption("check", "no-check", &check, "Check correctness");
85  bool quadratic = false;
86  CLP.setOption("quadratic", "linear", &quadratic, "Use quadratic basis functions");
87  int num_cores = -1;
88  CLP.setOption("cores", &num_cores,
89  "Number of CPU cores to use (defaults to all)");
90  int num_hyper_threads = -1;
91  CLP.setOption("hyperthreads", &num_hyper_threads,
92  "Number of hyper threads per core to use (defaults to all)");
93 #ifdef KOKKOS_ENABLE_THREADS
94  bool threads = true;
95  CLP.setOption("threads", "no-threads", &threads, "Enable Threads device");
96 #endif
97 #ifdef KOKKOS_ENABLE_OPENMP
98  bool openmp = true;
99  CLP.setOption("openmp", "no-openmp", &openmp, "Enable OpenMP device");
100 #endif
101 #ifdef KOKKOS_ENABLE_CUDA
102  bool cuda = true;
103  CLP.setOption("cuda", "no-cuda", &cuda, "Enable Cuda device");
104  int device_id = 0;
105  CLP.setOption("device", &device_id, "CUDA device ID.");
106 #endif
107  bool vtune = false;
108  CLP.setOption("vtune", "no-vtune", &vtune, "connect to vtune");
109  CLP.parse( argc, argv );
110 
111  if (nGrid > 0) {
112  nGridBegin = nGrid;
113  nGridEnd = nGrid;
114  }
115 
116  // Connect to VTune if requested
117  if (vtune) {
118  std::stringstream cmd;
119  pid_t my_os_pid=getpid();
120  const std::string vtune_loc =
121  "amplxe-cl";
122  const std::string output_dir = "./vtune/vtune.0";
123  cmd << vtune_loc
124  << " -collect hotspots -result-dir " << output_dir
125  << " -target-pid " << my_os_pid << " &";
126  std::cout << cmd.str() << std::endl;
127  system(cmd.str().c_str());
128  system("sleep 10");
129  }
130 
131  Kokkos::initialize(argc,argv);
132 #ifdef KOKKOS_ENABLE_THREADS
133  if (threads) {
134  typedef Kokkos::Threads Device;
135 
136  std::cout << std::endl
137  << "Threads performance with " << Kokkos::Threads::concurrency()
138  << " threads:" << std::endl;
139 
140  performance_test_driver<Device>(
141  print, nIter, nGridBegin, nGridEnd, nGridStep, quadratic, check);
142  }
143 #endif
144 
145 #ifdef KOKKOS_ENABLE_OPENMP
146  if (openmp) {
147  typedef Kokkos::OpenMP Device;
148 
149  std::cout << std::endl
150  << "OpenMP performance with " << Kokkos::OpenMP::concurrency()
151  << " threads:" << std::endl;
152 
153  performance_test_driver<Device>(
154  print, nIter, nGridBegin, nGridEnd, nGridStep, quadratic, check);
155 
156  }
157 #endif
158 
159 #ifdef KOKKOS_ENABLE_CUDA
160  if (cuda) {
161  typedef Kokkos::Cuda Device;
162 
163  cudaDeviceProp deviceProp;
164  cudaGetDeviceProperties(&deviceProp, device_id);
165  std::cout << std::endl
166  << "CUDA performance performance with device " << device_id
167  << " ("
168  << deviceProp.name << "):"
169  << std::endl;
170 
171  performance_test_driver<Device>(
172  print, nIter, nGridBegin, nGridEnd, nGridStep, quadratic, check);
173 
174  }
175 #endif
176  Kokkos::finalize();
177  }
178  TEUCHOS_STANDARD_CATCH_STATEMENTS(verbose, std::cerr, success);
179 
180  if (success)
181  return 0;
182  return -1;
183 }
std::enable_if< !Kokkos::is_view_fad< View2 >::value, bool >::type check(const View1 &v_gold, const View2 &v, const double tol)
void setOption(const char option_true[], const char option_false[], bool *option_val, const char documentation[]=NULL)
#define TEUCHOS_STANDARD_CATCH_STATEMENTS(VERBOSE, ERR_STREAM, SUCCESS_FLAG)
int main()
Definition: ad_example.cpp:191
EParseCommandLineReturn parse(int argc, char *argv[], std::ostream *errout=&std::cerr) const
void setDocString(const char doc_string[])