forked from rapidsai/cugraph
-
Notifications
You must be signed in to change notification settings - Fork 0
/
bfs_test.cpp
300 lines (256 loc) · 11.6 KB
/
bfs_test.cpp
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
/*
* Copyright (c) 2020-2022, NVIDIA CORPORATION.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governin_from_mtxg permissions and
* limitations under the License.
*/
#include <utilities/base_fixture.hpp>
#include <utilities/high_res_clock.h>
#include <utilities/test_graphs.hpp>
#include <utilities/test_utilities.hpp>
#include <utilities/thrust_wrapper.hpp>
#include <cugraph/algorithms.hpp>
#include <cugraph/graph.hpp>
#include <cugraph/graph_functions.hpp>
#include <cugraph/graph_view.hpp>
#include <raft/cudart_utils.h>
#include <raft/handle.hpp>
#include <rmm/device_scalar.hpp>
#include <rmm/device_uvector.hpp>
#include <rmm/mr/device/cuda_memory_resource.hpp>
#include <gtest/gtest.h>
#include <algorithm>
#include <iterator>
#include <limits>
#include <vector>
template <typename vertex_t, typename edge_t>
void bfs_reference(edge_t const* offsets,
vertex_t const* indices,
vertex_t* distances,
vertex_t* predecessors,
vertex_t num_vertices,
vertex_t source,
vertex_t depth_limit = std::numeric_limits<vertex_t>::max())
{
vertex_t depth{0};
std::fill(distances, distances + num_vertices, std::numeric_limits<vertex_t>::max());
std::fill(predecessors, predecessors + num_vertices, cugraph::invalid_vertex_id<vertex_t>::value);
*(distances + source) = depth;
std::vector<vertex_t> cur_frontier_rows{source};
std::vector<vertex_t> new_frontier_rows{};
while (cur_frontier_rows.size() > 0) {
for (auto const row : cur_frontier_rows) {
auto nbr_offset_first = *(offsets + row);
auto nbr_offset_last = *(offsets + row + 1);
for (auto nbr_offset = nbr_offset_first; nbr_offset != nbr_offset_last; ++nbr_offset) {
auto nbr = *(indices + nbr_offset);
if (*(distances + nbr) == std::numeric_limits<vertex_t>::max()) {
*(distances + nbr) = depth + 1;
*(predecessors + nbr) = row;
new_frontier_rows.push_back(nbr);
}
}
}
std::swap(cur_frontier_rows, new_frontier_rows);
new_frontier_rows.clear();
++depth;
if (depth >= depth_limit) { break; }
}
return;
}
struct BFS_Usecase {
size_t source{0};
bool check_correctness{true};
};
template <typename input_usecase_t>
class Tests_BFS : public ::testing::TestWithParam<std::tuple<BFS_Usecase, input_usecase_t>> {
public:
Tests_BFS() {}
static void SetUpTestCase() {}
static void TearDownTestCase() {}
virtual void SetUp() {}
virtual void TearDown() {}
template <typename vertex_t, typename edge_t>
void run_current_test(BFS_Usecase const& bfs_usecase, input_usecase_t const& input_usecase)
{
constexpr bool renumber = true;
using weight_t = float;
raft::handle_t handle{};
HighResClock hr_clock{};
if (cugraph::test::g_perf) {
RAFT_CUDA_TRY(cudaDeviceSynchronize()); // for consistent performance measurement
hr_clock.start();
}
auto [graph, d_renumber_map_labels] =
cugraph::test::construct_graph<vertex_t, edge_t, weight_t, false, false>(
handle, input_usecase, false, renumber);
if (cugraph::test::g_perf) {
RAFT_CUDA_TRY(cudaDeviceSynchronize()); // for consistent performance measurement
double elapsed_time{0.0};
hr_clock.stop(&elapsed_time);
std::cout << "construct_graph took " << elapsed_time * 1e-6 << " s.\n";
}
auto graph_view = graph.view();
ASSERT_TRUE(static_cast<vertex_t>(bfs_usecase.source) >= 0 &&
static_cast<vertex_t>(bfs_usecase.source) < graph_view.number_of_vertices())
<< "Invalid starting source.";
rmm::device_uvector<vertex_t> d_distances(graph_view.number_of_vertices(), handle.get_stream());
rmm::device_uvector<vertex_t> d_predecessors(graph_view.number_of_vertices(),
handle.get_stream());
if (cugraph::test::g_perf) {
RAFT_CUDA_TRY(cudaDeviceSynchronize()); // for consistent performance measurement
hr_clock.start();
}
rmm::device_scalar<vertex_t> const d_source(bfs_usecase.source, handle.get_stream());
cugraph::bfs(handle,
graph_view,
d_distances.data(),
d_predecessors.data(),
d_source.data(),
size_t{1},
false,
std::numeric_limits<vertex_t>::max());
if (cugraph::test::g_perf) {
RAFT_CUDA_TRY(cudaDeviceSynchronize()); // for consistent performance measurement
double elapsed_time{0.0};
hr_clock.stop(&elapsed_time);
std::cout << "BFS took " << elapsed_time * 1e-6 << " s.\n";
}
if (bfs_usecase.check_correctness) {
cugraph::graph_t<vertex_t, edge_t, weight_t, false, false> unrenumbered_graph(handle);
if (renumber) {
std::tie(unrenumbered_graph, std::ignore) =
cugraph::test::construct_graph<vertex_t, edge_t, weight_t, false, false>(
handle, input_usecase, false, false);
}
auto unrenumbered_graph_view = renumber ? unrenumbered_graph.view() : graph_view;
auto h_offsets = cugraph::test::to_host(
handle, unrenumbered_graph_view.local_edge_partition_view().offsets());
auto h_indices = cugraph::test::to_host(
handle, unrenumbered_graph_view.local_edge_partition_view().indices());
auto unrenumbered_source = static_cast<vertex_t>(bfs_usecase.source);
if (renumber) {
auto h_renumber_map_labels = cugraph::test::to_host(handle, *d_renumber_map_labels);
unrenumbered_source = h_renumber_map_labels[bfs_usecase.source];
}
std::vector<vertex_t> h_reference_distances(unrenumbered_graph_view.number_of_vertices());
std::vector<vertex_t> h_reference_predecessors(unrenumbered_graph_view.number_of_vertices());
bfs_reference(h_offsets.data(),
h_indices.data(),
h_reference_distances.data(),
h_reference_predecessors.data(),
unrenumbered_graph_view.number_of_vertices(),
unrenumbered_source,
std::numeric_limits<vertex_t>::max());
std::vector<vertex_t> h_cugraph_distances{};
std::vector<vertex_t> h_cugraph_predecessors{};
if (renumber) {
cugraph::unrenumber_local_int_vertices(handle,
d_predecessors.data(),
d_predecessors.size(),
(*d_renumber_map_labels).data(),
vertex_t{0},
graph_view.number_of_vertices(),
true);
rmm::device_uvector<vertex_t> d_unrenumbered_distances(size_t{0}, handle.get_stream());
std::tie(std::ignore, d_unrenumbered_distances) =
cugraph::test::sort_by_key(handle, *d_renumber_map_labels, d_distances);
rmm::device_uvector<vertex_t> d_unrenumbered_predecessors(size_t{0}, handle.get_stream());
std::tie(std::ignore, d_unrenumbered_predecessors) =
cugraph::test::sort_by_key(handle, *d_renumber_map_labels, d_predecessors);
h_cugraph_distances = cugraph::test::to_host(handle, d_unrenumbered_distances);
h_cugraph_predecessors = cugraph::test::to_host(handle, d_unrenumbered_predecessors);
} else {
h_cugraph_distances = cugraph::test::to_host(handle, d_distances);
h_cugraph_predecessors = cugraph::test::to_host(handle, d_predecessors);
}
ASSERT_TRUE(std::equal(
h_reference_distances.begin(), h_reference_distances.end(), h_cugraph_distances.begin()))
<< "distances do not match with the reference values.";
for (auto it = h_cugraph_predecessors.begin(); it != h_cugraph_predecessors.end(); ++it) {
auto i = std::distance(h_cugraph_predecessors.begin(), it);
if (*it == cugraph::invalid_vertex_id<vertex_t>::value) {
ASSERT_TRUE(h_reference_predecessors[i] == *it)
<< "vertex reachability does not match with the reference.";
} else {
ASSERT_TRUE(h_reference_distances[*it] + 1 == h_reference_distances[i])
<< "distance to this vertex != distance to the predecessor vertex + 1.";
bool found{false};
for (auto j = h_offsets[*it]; j < h_offsets[*it + 1]; ++j) {
if (h_indices[j] == i) {
found = true;
break;
}
}
ASSERT_TRUE(found) << "no edge from the predecessor vertex to this vertex.";
}
}
}
}
};
using Tests_BFS_File = Tests_BFS<cugraph::test::File_Usecase>;
using Tests_BFS_Rmat = Tests_BFS<cugraph::test::Rmat_Usecase>;
// FIXME: add tests for type combinations
TEST_P(Tests_BFS_File, CheckInt32Int32)
{
auto param = GetParam();
run_current_test<int32_t, int32_t>(std::get<0>(param), std::get<1>(param));
}
TEST_P(Tests_BFS_Rmat, CheckInt32Int32)
{
auto param = GetParam();
run_current_test<int32_t, int32_t>(
std::get<0>(param), override_Rmat_Usecase_with_cmd_line_arguments(std::get<1>(param)));
}
TEST_P(Tests_BFS_Rmat, CheckInt32Int64)
{
auto param = GetParam();
run_current_test<int32_t, int64_t>(
std::get<0>(param), override_Rmat_Usecase_with_cmd_line_arguments(std::get<1>(param)));
}
TEST_P(Tests_BFS_Rmat, CheckInt64Int64)
{
auto param = GetParam();
run_current_test<int64_t, int64_t>(
std::get<0>(param), override_Rmat_Usecase_with_cmd_line_arguments(std::get<1>(param)));
}
INSTANTIATE_TEST_SUITE_P(
file_test,
Tests_BFS_File,
::testing::Values(
// enable correctness checks
std::make_tuple(BFS_Usecase{0}, cugraph::test::File_Usecase("test/datasets/karate.mtx")),
std::make_tuple(BFS_Usecase{0}, cugraph::test::File_Usecase("test/datasets/polbooks.mtx")),
std::make_tuple(BFS_Usecase{0}, cugraph::test::File_Usecase("test/datasets/netscience.mtx")),
std::make_tuple(BFS_Usecase{100}, cugraph::test::File_Usecase("test/datasets/netscience.mtx")),
std::make_tuple(BFS_Usecase{1000}, cugraph::test::File_Usecase("test/datasets/wiki2003.mtx")),
std::make_tuple(BFS_Usecase{1000},
cugraph::test::File_Usecase("test/datasets/wiki-Talk.mtx"))));
INSTANTIATE_TEST_SUITE_P(
rmat_small_test,
Tests_BFS_Rmat,
::testing::Values(
// enable correctness checks
std::make_tuple(BFS_Usecase{0},
cugraph::test::Rmat_Usecase(10, 16, 0.57, 0.19, 0.19, 0, false, false))));
INSTANTIATE_TEST_SUITE_P(
rmat_benchmark_test, /* note that scale & edge factor can be overridden in benchmarking (with
--gtest_filter to select only the rmat_benchmark_test with a specific
vertex & edge type combination) by command line arguments and do not
include more than one Rmat_Usecase that differ only in scale or edge
factor (to avoid running same benchmarks more than once) */
Tests_BFS_Rmat,
::testing::Values(
// disable correctness checks for large graphs
std::make_pair(BFS_Usecase{0, false},
cugraph::test::Rmat_Usecase(20, 32, 0.57, 0.19, 0.19, 0, false, false))));
CUGRAPH_TEST_PROGRAM_MAIN()