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
|
/*
* Copyright (c) 2019 Samsung Electronics Co., Ltd. All Rights Reserved
*
* 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.
*/
#include "DataflowExecutor.h"
#include <cassert>
#include "util/logging.h"
namespace onert
{
namespace exec
{
int64_t DataflowExecutor::calculateRank(const std::vector<ir::Element> &operations)
{
int64_t rank = 0;
if (!_indexed_ranks)
{
return rank;
}
for (const auto &element : operations)
{
auto it = _indexed_ranks->find(element.index);
if (it == _indexed_ranks->end())
{
assert(element.node->opcode() == ir::OpCode::Permute && operations.size() == 1);
// run Permute ASAP for next operations to be ready for other backends
return std::numeric_limits<int64_t>::max();
}
else
{
rank += it->second;
}
}
return rank;
}
void DataflowExecutor::emplaceToReadyJobs(const uint32_t &id)
{
auto &job = _waiting_jobs[id];
assert(job != nullptr);
auto &op_seq = _lowered_graph->op_seqs().at(_job_to_op_seq[job->index()]);
auto rank = calculateRank(op_seq.operations());
_ready_jobs.emplace(rank, std::move(job));
}
void DataflowExecutor::notify(uint32_t finished_job_id)
{
for (auto id : _output_info[finished_job_id])
{
assert(_input_info[id] > 0);
auto count = --_input_info[id];
if (count == 0) // No dependent jobs left, ready for execution
{
emplaceToReadyJobs(id);
}
}
}
bool DataflowExecutor::noWaitingJobs()
{
return std::all_of(_waiting_jobs.begin(), _waiting_jobs.end(),
[](const std::unique_ptr<Job> &job) { return job == nullptr; });
}
DataflowExecutor::DataflowExecutor(std::unique_ptr<ir::LoweredGraph> lowered_graph,
const backend::TensorBuilderSet &tensor_builders,
compiler::CodeMap &&code_map)
: ExecutorBase{std::move(lowered_graph), tensor_builders}, _code_map{std::move(code_map)},
_profiling{false}
{
VERBOSE(DataflowExecutor) << "Constructing Dataflow Executor" << std::endl;
const auto &op_seqs = _lowered_graph->op_seqs();
// Assign jobs convert OpSequenceIndex to job index(uint32_t)
uint32_t next_job_index = 0;
std::unordered_map<ir::OpSequenceIndex, uint32_t> op_seq_to_job;
op_seqs.iterate([&](const ir::OpSequenceIndex &op_seq_index, const ir::OpSequence &) {
VERBOSE(DataflowExecutor) << "Create a job #" << next_job_index << " with OpSequenceIndex "
<< op_seq_index.value() << std::endl;
_finished_jobs.emplace_back(
std::make_unique<Job>(next_job_index, _code_map.at(op_seq_index).fn_seq.get()));
op_seq_to_job[op_seq_index] = next_job_index++;
});
_waiting_jobs.resize(next_job_index);
_output_info.resize(next_job_index);
_initial_input_info.resize(next_job_index, 0);
op_seqs.iterate([&](const ir::OpSequenceIndex &op_seq_index, const ir::OpSequence &op_seq) {
auto job_index = op_seq_to_job[op_seq_index];
for (auto output : op_seq.getOutputs())
{
// Update output and input info
op_seqs.iterate(
[&](const ir::OpSequenceIndex &op_seq_cur_index, const ir::OpSequence &op_seq_cur) {
if (op_seq_cur.getInputs().contains(output))
{
auto dep_index = op_seq_to_job[op_seq_cur_index];
++_initial_input_info[dep_index];
_output_info[job_index].push_back(dep_index);
}
});
}
});
for (const auto &s : op_seq_to_job)
_job_to_op_seq.emplace(s.second, s.first);
_input_info = _initial_input_info;
}
void DataflowExecutor::executeImpl()
{
assert(noWaitingJobs());
// Execution setup
_waiting_jobs.swap(_finished_jobs); // Move finished jobs to waiting jobs
for (uint32_t i = 0; i < _waiting_jobs.size(); ++i)
{
if (_input_info[i] == 0)
{
emplaceToReadyJobs(i);
}
}
assert(!_ready_jobs.empty()); // Cannot begin if there is no initial jobs
_subject.notifyModelBegin(this);
while (!_ready_jobs.empty())
{
auto job = std::move((_ready_jobs.begin())->second);
_ready_jobs.erase(_ready_jobs.begin());
auto job_index = job->index();
VERBOSE(DataflowExecutor) << "Run job #" << job_index << std::endl;
auto op_seq_index = _job_to_op_seq[job_index];
auto op_seq = &_lowered_graph->op_seqs().at(op_seq_index);
const backend::Backend *backend =
_lowered_graph->getLowerInfo()->operation.at(op_seq_index)->backend();
_subject.notifyJobBegin(this, op_seq, backend);
if (_profiling)
job->fn()->runSync();
else
job->run();
_subject.notifyJobEnd(this, op_seq, backend);
notify(job_index);
_finished_jobs[job_index] = std::move(job);
}
assert(noWaitingJobs());
_subject.notifyModelEnd(this);
// Reset input info for the next execution
_input_info = _initial_input_info;
}
} // namespace exec
} // namespace onert
|
