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
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
|
/*
Copyright (c) 2005-2024 Intel 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.
*/
#ifndef __TBB_parallel_for_H
#define __TBB_parallel_for_H
#include "detail/_config.h"
#include "detail/_namespace_injection.h"
#include "detail/_exception.h"
#include "detail/_task.h"
#include "detail/_small_object_pool.h"
#include "profiling.h"
#include "partitioner.h"
#include "blocked_range.h"
#include "task_group.h"
#include <cstddef>
#include <new>
namespace tbb {
namespace detail {
#if __TBB_CPP20_CONCEPTS_PRESENT
inline namespace d0 {
template <typename Body, typename Range>
concept parallel_for_body = std::copy_constructible<Body> && std::invocable<const std::remove_reference_t<Body>&, Range&>;
template <typename Index>
concept parallel_for_index = std::constructible_from<Index, int> &&
std::copyable<Index> &&
requires( const std::remove_reference_t<Index>& lhs, const std::remove_reference_t<Index>& rhs ) {
{ lhs < rhs } -> adaptive_same_as<bool>;
{ lhs - rhs } -> std::convertible_to<std::size_t>;
{ lhs + (rhs - lhs) } -> std::convertible_to<Index>;
};
template <typename Function, typename Index>
concept parallel_for_function = std::invocable<const std::remove_reference_t<Function>&, Index>;
} // namespace d0
#endif // __TBB_CPP20_CONCEPTS_PRESENT
namespace d1 {
//! Task type used in parallel_for
/** @ingroup algorithms */
template<typename Range, typename Body, typename Partitioner>
struct start_for : public task {
Range my_range;
const Body my_body;
node* my_parent;
typename Partitioner::task_partition_type my_partition;
small_object_allocator my_allocator;
task* execute(execution_data&) override;
task* cancel(execution_data&) override;
void finalize(const execution_data&);
//! Constructor for root task.
start_for( const Range& range, const Body& body, Partitioner& partitioner, small_object_allocator& alloc ) :
my_range(range),
my_body(body),
my_parent(nullptr),
my_partition(partitioner),
my_allocator(alloc) {}
//! Splitting constructor used to generate children.
/** parent_ becomes left child. Newly constructed object is right child. */
start_for( start_for& parent_, typename Partitioner::split_type& split_obj, small_object_allocator& alloc ) :
my_range(parent_.my_range, get_range_split_object<Range>(split_obj)),
my_body(parent_.my_body),
my_parent(nullptr),
my_partition(parent_.my_partition, split_obj),
my_allocator(alloc) {}
//! Construct right child from the given range as response to the demand.
/** parent_ remains left child. Newly constructed object is right child. */
start_for( start_for& parent_, const Range& r, depth_t d, small_object_allocator& alloc ) :
my_range(r),
my_body(parent_.my_body),
my_parent(nullptr),
my_partition(parent_.my_partition, split()),
my_allocator(alloc)
{
my_partition.align_depth( d );
}
static void run(const Range& range, const Body& body, Partitioner& partitioner) {
task_group_context context(PARALLEL_FOR);
run(range, body, partitioner, context);
}
static void run(const Range& range, const Body& body, Partitioner& partitioner, task_group_context& context) {
if ( !range.empty() ) {
small_object_allocator alloc{};
start_for& for_task = *alloc.new_object<start_for>(range, body, partitioner, alloc);
// defer creation of the wait node until task allocation succeeds
wait_node wn;
for_task.my_parent = &wn;
execute_and_wait(for_task, context, wn.m_wait, context);
}
}
//! Run body for range, serves as callback for partitioner
void run_body( Range &r ) {
tbb::detail::invoke(my_body, r);
}
//! spawn right task, serves as callback for partitioner
void offer_work(typename Partitioner::split_type& split_obj, execution_data& ed) {
offer_work_impl(ed, *this, split_obj);
}
//! spawn right task, serves as callback for partitioner
void offer_work(const Range& r, depth_t d, execution_data& ed) {
offer_work_impl(ed, *this, r, d);
}
private:
template <typename... Args>
void offer_work_impl(execution_data& ed, Args&&... constructor_args) {
// New right child
small_object_allocator alloc{};
start_for& right_child = *alloc.new_object<start_for>(ed, std::forward<Args>(constructor_args)..., alloc);
// New root node as a continuation and ref count. Left and right child attach to the new parent.
right_child.my_parent = my_parent = alloc.new_object<tree_node>(ed, my_parent, 2, alloc);
// Spawn the right sibling
right_child.spawn_self(ed);
}
void spawn_self(execution_data& ed) {
my_partition.spawn_task(*this, *context(ed));
}
};
//! fold the tree and deallocate the task
template<typename Range, typename Body, typename Partitioner>
void start_for<Range, Body, Partitioner>::finalize(const execution_data& ed) {
// Get the current parent and allocator an object destruction
node* parent = my_parent;
auto allocator = my_allocator;
// Task execution finished - destroy it
this->~start_for();
// Unwind the tree decrementing the parent`s reference count
fold_tree<tree_node>(parent, ed);
allocator.deallocate(this, ed);
}
//! execute task for parallel_for
template<typename Range, typename Body, typename Partitioner>
task* start_for<Range, Body, Partitioner>::execute(execution_data& ed) {
if (!is_same_affinity(ed)) {
my_partition.note_affinity(execution_slot(ed));
}
my_partition.check_being_stolen(*this, ed);
my_partition.execute(*this, my_range, ed);
finalize(ed);
return nullptr;
}
//! cancel task for parallel_for
template<typename Range, typename Body, typename Partitioner>
task* start_for<Range, Body, Partitioner>::cancel(execution_data& ed) {
finalize(ed);
return nullptr;
}
//! Calls the function with values from range [begin, end) with a step provided
template<typename Function, typename Index>
class parallel_for_body_wrapper : detail::no_assign {
const Function &my_func;
const Index my_begin;
const Index my_step;
public:
parallel_for_body_wrapper( const Function& _func, Index& _begin, Index& _step )
: my_func(_func), my_begin(_begin), my_step(_step) {}
void operator()( const blocked_range<Index>& r ) const {
// A set of local variables to help the compiler with vectorization of the following loop.
Index b = r.begin();
Index e = r.end();
Index ms = my_step;
Index k = my_begin + b*ms;
#if __INTEL_COMPILER
#pragma ivdep
#if __TBB_ASSERT_ON_VECTORIZATION_FAILURE
#pragma vector always assert
#endif
#endif
for ( Index i = b; i < e; ++i, k += ms ) {
tbb::detail::invoke(my_func, k);
}
}
};
// Requirements on Range concept are documented in blocked_range.h
/** \page parallel_for_body_req Requirements on parallel_for body
Class \c Body implementing the concept of parallel_for body must define:
- \code Body::Body( const Body& ); \endcode Copy constructor
- \code Body::~Body(); \endcode Destructor
- \code void Body::operator()( Range& r ) const; \endcode Function call operator applying the body to range \c r.
**/
/** \name parallel_for
See also requirements on \ref range_req "Range" and \ref parallel_for_body_req "parallel_for Body". **/
//@{
//! Parallel iteration over range with default partitioner.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body ) {
start_for<Range,Body,const __TBB_DEFAULT_PARTITIONER>::run(range,body,__TBB_DEFAULT_PARTITIONER());
}
//! Parallel iteration over range with simple partitioner.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, const simple_partitioner& partitioner ) {
start_for<Range,Body,const simple_partitioner>::run(range,body,partitioner);
}
//! Parallel iteration over range with auto_partitioner.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, const auto_partitioner& partitioner ) {
start_for<Range,Body,const auto_partitioner>::run(range,body,partitioner);
}
//! Parallel iteration over range with static_partitioner.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, const static_partitioner& partitioner ) {
start_for<Range,Body,const static_partitioner>::run(range,body,partitioner);
}
//! Parallel iteration over range with affinity_partitioner.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, affinity_partitioner& partitioner ) {
start_for<Range,Body,affinity_partitioner>::run(range,body,partitioner);
}
//! Parallel iteration over range with default partitioner and user-supplied context.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, task_group_context& context ) {
start_for<Range,Body,const __TBB_DEFAULT_PARTITIONER>::run(range, body, __TBB_DEFAULT_PARTITIONER(), context);
}
//! Parallel iteration over range with simple partitioner and user-supplied context.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, const simple_partitioner& partitioner, task_group_context& context ) {
start_for<Range,Body,const simple_partitioner>::run(range, body, partitioner, context);
}
//! Parallel iteration over range with auto_partitioner and user-supplied context.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, const auto_partitioner& partitioner, task_group_context& context ) {
start_for<Range,Body,const auto_partitioner>::run(range, body, partitioner, context);
}
//! Parallel iteration over range with static_partitioner and user-supplied context.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, const static_partitioner& partitioner, task_group_context& context ) {
start_for<Range,Body,const static_partitioner>::run(range, body, partitioner, context);
}
//! Parallel iteration over range with affinity_partitioner and user-supplied context.
/** @ingroup algorithms **/
template<typename Range, typename Body>
__TBB_requires(tbb_range<Range> && parallel_for_body<Body, Range>)
void parallel_for( const Range& range, const Body& body, affinity_partitioner& partitioner, task_group_context& context ) {
start_for<Range,Body,affinity_partitioner>::run(range,body,partitioner, context);
}
//! Implementation of parallel iteration over stepped range of integers with explicit step and partitioner
template <typename Index, typename Function, typename Partitioner>
void parallel_for_impl(Index first, Index last, Index step, const Function& f, Partitioner& partitioner) {
if (step <= 0 )
throw_exception(exception_id::nonpositive_step); // throws std::invalid_argument
else if (first < last) {
// Above "else" avoids "potential divide by zero" warning on some platforms
Index end = Index(last - first - 1ul) / step + Index(1);
blocked_range<Index> range(static_cast<Index>(0), end);
parallel_for_body_wrapper<Function, Index> body(f, first, step);
parallel_for(range, body, partitioner);
}
}
//! Parallel iteration over a range of integers with a step provided and default partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f) {
parallel_for_impl<Index,Function,const __TBB_DEFAULT_PARTITIONER>(first, last, step, f, __TBB_DEFAULT_PARTITIONER());
}
//! Parallel iteration over a range of integers with a step provided and simple partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, const simple_partitioner& partitioner) {
parallel_for_impl<Index,Function,const simple_partitioner>(first, last, step, f, partitioner);
}
//! Parallel iteration over a range of integers with a step provided and auto partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, const auto_partitioner& partitioner) {
parallel_for_impl<Index,Function,const auto_partitioner>(first, last, step, f, partitioner);
}
//! Parallel iteration over a range of integers with a step provided and static partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, const static_partitioner& partitioner) {
parallel_for_impl<Index,Function,const static_partitioner>(first, last, step, f, partitioner);
}
//! Parallel iteration over a range of integers with a step provided and affinity partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, affinity_partitioner& partitioner) {
parallel_for_impl(first, last, step, f, partitioner);
}
//! Parallel iteration over a range of integers with a default step value and default partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f) {
parallel_for_impl<Index,Function,const __TBB_DEFAULT_PARTITIONER>(first, last, static_cast<Index>(1), f, __TBB_DEFAULT_PARTITIONER());
}
//! Parallel iteration over a range of integers with a default step value and simple partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, const simple_partitioner& partitioner) {
parallel_for_impl<Index,Function,const simple_partitioner>(first, last, static_cast<Index>(1), f, partitioner);
}
//! Parallel iteration over a range of integers with a default step value and auto partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, const auto_partitioner& partitioner) {
parallel_for_impl<Index,Function,const auto_partitioner>(first, last, static_cast<Index>(1), f, partitioner);
}
//! Parallel iteration over a range of integers with a default step value and static partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, const static_partitioner& partitioner) {
parallel_for_impl<Index,Function,const static_partitioner>(first, last, static_cast<Index>(1), f, partitioner);
}
//! Parallel iteration over a range of integers with a default step value and affinity partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, affinity_partitioner& partitioner) {
parallel_for_impl(first, last, static_cast<Index>(1), f, partitioner);
}
//! Implementation of parallel iteration over stepped range of integers with explicit step, task group context, and partitioner
template <typename Index, typename Function, typename Partitioner>
void parallel_for_impl(Index first, Index last, Index step, const Function& f, Partitioner& partitioner, task_group_context &context) {
if (step <= 0 )
throw_exception(exception_id::nonpositive_step); // throws std::invalid_argument
else if (first < last) {
// Above "else" avoids "potential divide by zero" warning on some platforms
Index end = (last - first - Index(1)) / step + Index(1);
blocked_range<Index> range(static_cast<Index>(0), end);
parallel_for_body_wrapper<Function, Index> body(f, first, step);
parallel_for(range, body, partitioner, context);
}
}
//! Parallel iteration over a range of integers with explicit step, task group context, and default partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, task_group_context &context) {
parallel_for_impl<Index,Function,const __TBB_DEFAULT_PARTITIONER>(first, last, step, f, __TBB_DEFAULT_PARTITIONER(), context);
}
//! Parallel iteration over a range of integers with explicit step, task group context, and simple partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, const simple_partitioner& partitioner, task_group_context &context) {
parallel_for_impl<Index,Function,const simple_partitioner>(first, last, step, f, partitioner, context);
}
//! Parallel iteration over a range of integers with explicit step, task group context, and auto partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, const auto_partitioner& partitioner, task_group_context &context) {
parallel_for_impl<Index,Function,const auto_partitioner>(first, last, step, f, partitioner, context);
}
//! Parallel iteration over a range of integers with explicit step, task group context, and static partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, const static_partitioner& partitioner, task_group_context &context) {
parallel_for_impl<Index,Function,const static_partitioner>(first, last, step, f, partitioner, context);
}
//! Parallel iteration over a range of integers with explicit step, task group context, and affinity partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, Index step, const Function& f, affinity_partitioner& partitioner, task_group_context &context) {
parallel_for_impl(first, last, step, f, partitioner, context);
}
//! Parallel iteration over a range of integers with a default step value, explicit task group context, and default partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, task_group_context &context) {
parallel_for_impl<Index,Function,const __TBB_DEFAULT_PARTITIONER>(first, last, static_cast<Index>(1), f, __TBB_DEFAULT_PARTITIONER(), context);
}
//! Parallel iteration over a range of integers with a default step value, explicit task group context, and simple partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, const simple_partitioner& partitioner, task_group_context &context) {
parallel_for_impl<Index,Function,const simple_partitioner>(first, last, static_cast<Index>(1), f, partitioner, context);
}
//! Parallel iteration over a range of integers with a default step value, explicit task group context, and auto partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, const auto_partitioner& partitioner, task_group_context &context) {
parallel_for_impl<Index,Function,const auto_partitioner>(first, last, static_cast<Index>(1), f, partitioner, context);
}
//! Parallel iteration over a range of integers with a default step value, explicit task group context, and static partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, const static_partitioner& partitioner, task_group_context &context) {
parallel_for_impl<Index,Function,const static_partitioner>(first, last, static_cast<Index>(1), f, partitioner, context);
}
//! Parallel iteration over a range of integers with a default step value, explicit task group context, and affinity_partitioner
template <typename Index, typename Function>
__TBB_requires(parallel_for_index<Index> && parallel_for_function<Function, Index>)
void parallel_for(Index first, Index last, const Function& f, affinity_partitioner& partitioner, task_group_context &context) {
parallel_for_impl(first, last, static_cast<Index>(1), f, partitioner, context);
}
// @}
} // namespace d1
} // namespace detail
inline namespace v1 {
using detail::d1::parallel_for;
// Split types
using detail::split;
using detail::proportional_split;
} // namespace v1
} // namespace tbb
#endif /* __TBB_parallel_for_H */
|
