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
|
/*
* Copyright 2008-2021 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 governing permissions and
* limitations under the License.
*/
#pragma once
#include <thrust/detail/config.h>
#include <thrust/iterator/iterator_traits.h>
#include <thrust/detail/temporary_array.h>
#include <thrust/system/tbb/detail/execution_policy.h>
#include <thrust/merge.h>
#include <thrust/binary_search.h>
#include <thrust/detail/seq.h>
#include <tbb/parallel_for.h>
THRUST_NAMESPACE_BEGIN
namespace system
{
namespace tbb
{
namespace detail
{
namespace merge_detail
{
template<typename InputIterator1,
typename InputIterator2,
typename OutputIterator,
typename StrictWeakOrdering>
struct range
{
InputIterator1 first1, last1;
InputIterator2 first2, last2;
OutputIterator result;
StrictWeakOrdering comp;
size_t grain_size;
range(InputIterator1 first1, InputIterator1 last1,
InputIterator2 first2, InputIterator2 last2,
OutputIterator result,
StrictWeakOrdering comp,
size_t grain_size = 1024)
: first1(first1), last1(last1),
first2(first2), last2(last2),
result(result), comp(comp), grain_size(grain_size)
{}
range(range& r, ::tbb::split)
: first1(r.first1), last1(r.last1),
first2(r.first2), last2(r.last2),
result(r.result), comp(r.comp), grain_size(r.grain_size)
{
// we can assume n1 and n2 are not both 0
size_t n1 = thrust::distance(first1, last1);
size_t n2 = thrust::distance(first2, last2);
InputIterator1 mid1 = first1;
InputIterator2 mid2 = first2;
if (n1 > n2)
{
mid1 += n1 / 2;
mid2 = thrust::lower_bound(thrust::seq, first2, last2, raw_reference_cast(*mid1), comp);
}
else
{
mid2 += n2 / 2;
mid1 = thrust::upper_bound(thrust::seq, first1, last1, raw_reference_cast(*mid2), comp);
}
// set first range to [first1, mid1), [first2, mid2), result
r.last1 = mid1;
r.last2 = mid2;
// set second range to [mid1, last1), [mid2, last2), result + (mid1 - first1) + (mid2 - first2)
first1 = mid1;
first2 = mid2;
result += thrust::distance(r.first1, mid1) + thrust::distance(r.first2, mid2);
}
bool empty(void) const
{
return (first1 == last1) && (first2 == last2);
}
bool is_divisible(void) const
{
return static_cast<size_t>(thrust::distance(first1, last1) + thrust::distance(first2, last2)) > grain_size;
}
};
struct body
{
template <typename Range>
void operator()(Range& r) const
{
thrust::merge(thrust::seq,
r.first1, r.last1,
r.first2, r.last2,
r.result,
r.comp);
}
};
} // end namespace merge_detail
namespace merge_by_key_detail
{
template<typename InputIterator1,
typename InputIterator2,
typename InputIterator3,
typename InputIterator4,
typename OutputIterator1,
typename OutputIterator2,
typename StrictWeakOrdering>
struct range
{
InputIterator1 keys_first1, keys_last1;
InputIterator2 keys_first2, keys_last2;
InputIterator3 values_first1;
InputIterator4 values_first2;
OutputIterator1 keys_result;
OutputIterator2 values_result;
StrictWeakOrdering comp;
size_t grain_size;
range(InputIterator1 keys_first1, InputIterator1 keys_last1,
InputIterator2 keys_first2, InputIterator2 keys_last2,
InputIterator3 values_first1,
InputIterator4 values_first2,
OutputIterator1 keys_result,
OutputIterator2 values_result,
StrictWeakOrdering comp,
size_t grain_size = 1024)
: keys_first1(keys_first1), keys_last1(keys_last1),
keys_first2(keys_first2), keys_last2(keys_last2),
values_first1(values_first1),
values_first2(values_first2),
keys_result(keys_result), values_result(values_result),
comp(comp), grain_size(grain_size)
{}
range(range& r, ::tbb::split)
: keys_first1(r.keys_first1), keys_last1(r.keys_last1),
keys_first2(r.keys_first2), keys_last2(r.keys_last2),
values_first1(r.values_first1),
values_first2(r.values_first2),
keys_result(r.keys_result), values_result(r.values_result),
comp(r.comp), grain_size(r.grain_size)
{
// we can assume n1 and n2 are not both 0
size_t n1 = thrust::distance(keys_first1, keys_last1);
size_t n2 = thrust::distance(keys_first2, keys_last2);
InputIterator1 mid1 = keys_first1;
InputIterator2 mid2 = keys_first2;
if (n1 > n2)
{
mid1 += n1 / 2;
mid2 = thrust::lower_bound(thrust::seq, keys_first2, keys_last2, raw_reference_cast(*mid1), comp);
}
else
{
mid2 += n2 / 2;
mid1 = thrust::upper_bound(thrust::seq, keys_first1, keys_last1, raw_reference_cast(*mid2), comp);
}
// set first range to [keys_first1, mid1), [keys_first2, mid2), keys_result, values_result
r.keys_last1 = mid1;
r.keys_last2 = mid2;
// set second range to [mid1, keys_last1), [mid2, keys_last2), keys_result + (mid1 - keys_first1) + (mid2 - keys_first2), values_result + (mid1 - keys_first1) + (mid2 - keys_first2)
keys_first1 = mid1;
keys_first2 = mid2;
values_first1 += thrust::distance(r.keys_first1, mid1);
values_first2 += thrust::distance(r.keys_first2, mid2);
keys_result += thrust::distance(r.keys_first1, mid1) + thrust::distance(r.keys_first2, mid2);
values_result += thrust::distance(r.keys_first1, mid1) + thrust::distance(r.keys_first2, mid2);
}
bool empty(void) const
{
return (keys_first1 == keys_last1) && (keys_first2 == keys_last2);
}
bool is_divisible(void) const
{
return static_cast<size_t>(thrust::distance(keys_first1, keys_last1) + thrust::distance(keys_first2, keys_last2)) > grain_size;
}
};
struct body
{
template <typename Range>
void operator()(Range& r) const
{
thrust::merge_by_key(thrust::seq,
r.keys_first1, r.keys_last1,
r.keys_first2, r.keys_last2,
r.values_first1,
r.values_first2,
r.keys_result,
r.values_result,
r.comp);
}
};
} // end namespace merge_by_key_detail
template<typename DerivedPolicy,
typename InputIterator1,
typename InputIterator2,
typename OutputIterator,
typename StrictWeakOrdering>
OutputIterator merge(execution_policy<DerivedPolicy> &,
InputIterator1 first1,
InputIterator1 last1,
InputIterator2 first2,
InputIterator2 last2,
OutputIterator result,
StrictWeakOrdering comp)
{
typedef typename merge_detail::range<InputIterator1,InputIterator2,OutputIterator,StrictWeakOrdering> Range;
typedef merge_detail::body Body;
Range range(first1, last1, first2, last2, result, comp);
Body body;
::tbb::parallel_for(range, body);
thrust::advance(result, thrust::distance(first1, last1) + thrust::distance(first2, last2));
return result;
} // end merge()
template <typename DerivedPolicy,
typename InputIterator1,
typename InputIterator2,
typename InputIterator3,
typename InputIterator4,
typename OutputIterator1,
typename OutputIterator2,
typename StrictWeakOrdering>
thrust::pair<OutputIterator1,OutputIterator2>
merge_by_key(execution_policy<DerivedPolicy> &,
InputIterator1 keys_first1,
InputIterator1 keys_last1,
InputIterator2 keys_first2,
InputIterator2 keys_last2,
InputIterator3 values_first3,
InputIterator4 values_first4,
OutputIterator1 keys_result,
OutputIterator2 values_result,
StrictWeakOrdering comp)
{
typedef typename merge_by_key_detail::range<InputIterator1,InputIterator2,InputIterator3,InputIterator4,OutputIterator1,OutputIterator2,StrictWeakOrdering> Range;
typedef merge_by_key_detail::body Body;
Range range(keys_first1, keys_last1, keys_first2, keys_last2, values_first3, values_first4, keys_result, values_result, comp);
Body body;
::tbb::parallel_for(range, body);
thrust::advance(keys_result, thrust::distance(keys_first1, keys_last1) + thrust::distance(keys_first2, keys_last2));
thrust::advance(values_result, thrust::distance(keys_first1, keys_last1) + thrust::distance(keys_first2, keys_last2));
return thrust::make_pair(keys_result,values_result);
}
} // end namespace detail
} // end namespace tbb
} // end namespace system
THRUST_NAMESPACE_END
|
