graph_pattern_detector.cc 18.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
// Copyright (c) 2018 PaddlePaddle Authors. 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.

Q
Qiao Longfei 已提交
15
#include <array>
16 17 18 19
#include <string>
#include <vector>

#include "paddle/fluid/framework/ir/graph_helper.h"
20
#include "paddle/fluid/framework/ir/graph_pattern_detector.h"
21
#include "paddle/fluid/framework/ir/graph_traits.h"
22
#include "paddle/fluid/framework/ir/graph_viz_pass.h"
23 24 25 26 27 28
#include "paddle/fluid/platform/enforce.h"

namespace paddle {
namespace framework {
namespace ir {

29 30
size_t PDPattern::id_ = 0UL;

Y
Yan Chunwei 已提交
31 32 33 34 35 36 37 38 39 40 41 42 43
PDNode* PDPattern::NewNode(const std::string& name) {
  if (!name.empty()) {
    PADDLE_ENFORCE_EQ(node_map_.count(name), 0,
                      "PDNode's name should be unique, get duplicate [%s]",
                      name);
  }

  nodes_.emplace_back(new PDNode(this, name));
  auto* cur = nodes_.back().get();
  node_map_[name] = cur;
  return cur;
}

44
PDNode* PDPattern::NewNode(PDNode::teller_t&& teller, const std::string& name) {
45 46 47 48 49 50
  if (!name.empty()) {
    PADDLE_ENFORCE_EQ(node_map_.count(name), 0,
                      "PDNode's name should be unique, get duplicate [%s]",
                      name);
  }

51
  nodes_.emplace_back(new PDNode(std::move(teller), this, name));
52
  auto* cur = nodes_.back().get();
53
  node_map_[name] = cur;
54 55 56
  return cur;
}

Y
Yan Chunwei 已提交
57
PDNode* PDPattern::RetrieveNode(const std::string& id) const {
58 59 60 61 62 63 64 65
  auto it = node_map_.find(id);
  if (it == node_map_.end()) {
    return nullptr;
  }

  return it->second;
}

66 67 68 69 70 71 72
void PDPattern::AddEdge(PDNode* a, PDNode* b) {
  PADDLE_ENFORCE(a);
  PADDLE_ENFORCE(b);
  PADDLE_ENFORCE(a != b, "can't connect to the same nodes.");
  edges_.emplace_back(a, b);
}

73 74
void GraphPatternDetector::operator()(Graph* graph,
                                      GraphPatternDetector::handle_t handler) {
75 76 77 78
  if (!MarkPDNodesInGraph(*graph)) {
    return;
  }

79 80 81
  auto subgraphs = DetectPatterns();
  UniquePatterns(&subgraphs);
  RemoveOverlappedMatch(&subgraphs);
Y
Yan Chunwei 已提交
82
  ValidateByNodeRole(&subgraphs);
83

Y
Yan Chunwei 已提交
84
  if (subgraphs.empty()) return;
85 86
  LOG(INFO) << "detect " << subgraphs.size() << " subgraph matches the pattern";
  int id = 0;
87
  for (auto& g : subgraphs) {
88
    LOG(INFO) << "optimizing #" << id++ << " subgraph";
89 90 91 92
    handler(g, graph);
  }
}

93
bool GraphPatternDetector::MarkPDNodesInGraph(const ir::Graph& graph) {
94
  VLOG(3) << "mark pdnodes in graph";
95 96 97 98 99
  if (graph.Nodes().empty()) return false;

  for (auto& node : GraphTraits::DFS(graph)) {
    for (const auto& pdnode : pattern_.nodes()) {
      if (pdnode->Tell(&node)) {
100
        VLOG(4) << "pdnode " << pdnode->name() << " marked";
101 102 103 104
        pdnodes2nodes_[pdnode.get()].insert(&node);
      }
    }
  }
Y
Yan Chunwei 已提交
105 106 107 108 109 110 111 112
  // Check to early stop if some PDNode can't find matched Node.
  for (auto& pdnode : pattern_.nodes()) {
    if (!pdnodes2nodes_.count(pdnode.get())) {
      VLOG(4) << pdnode->name() << " can't find matched Node, early stop";

      return false;
    }
  }
Y
Yan Chunwei 已提交
113 114 115 116 117
  for (auto& item : pdnodes2nodes_) {
    for (auto& n : item.second) {
      GetMarkedNodes(const_cast<Graph*>(&graph)).insert(n);
    }
  }
118
  VLOG(3) << pdnodes2nodes_.size() << " nodes marked";
119

120 121 122
  return !pdnodes2nodes_.empty();
}

Y
Yan Chunwei 已提交
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
// The intermediate Nodes can only link to the nodes inside the pattern, or this
// subgraph will be droped.
void GraphPatternDetector::ValidateByNodeRole(
    std::vector<GraphPatternDetector::subgraph_t>* subgraphs) {
  std::vector<GraphPatternDetector::subgraph_t> result;

  subgraphs->erase(
      std::remove_if(
          subgraphs->begin(), subgraphs->end(),
          [](const GraphPatternDetector::subgraph_t& subgraph) -> bool {
            // Collect the inputs and outputs.
            std::unordered_set<Node*> ios;
            for (auto& item : subgraph) {
              if (!item.first->IsIntermediate()) {
                ios.insert(item.second);
              }
            }
            for (auto& item : subgraph) {
              if (item.first->IsIntermediate()) {
                for (auto* x : item.second->inputs) {
                  if (!ios.count(x)) {
                    return true;
                  }
                }
                for (auto* x : item.second->outputs) {
                  if (!ios.count(x)) {
                    return true;
                  }
                }
              }
            }
            return false;
          }),
      subgraphs->end());
}

159 160 161 162
struct HitGroup {
  std::unordered_map<PDNode*, Node*> roles;

  bool Match(Node* node, PDNode* pat) {
163 164 165 166
    if (nodes_.count(node)) {
      if (!roles.count(pat)) return false;
      return roles[pat] == node;
    }
167 168 169
    return !roles.count(pat) || roles.at(pat) == node;
  }

170 171 172 173 174 175 176
  void Register(Node* node, PDNode* pat) {
    roles[pat] = node;
    nodes_.insert(node);
  }

 private:
  std::unordered_set<Node*> nodes_;
177 178 179 180 181 182 183 184 185 186 187 188
};

// Tell whether Node a links to b.
bool IsNodesLink(Node* a, Node* b) {
  for (auto* node : a->outputs) {
    if (b == node) {
      return true;
    }
  }
  return false;
}

189 190
std::vector<GraphPatternDetector::subgraph_t>
GraphPatternDetector::DetectPatterns() {
191
  // Init empty subgraphs.
192
  std::vector<GraphPatternDetector::subgraph_t> result;
193
  std::vector<HitGroup> init_groups;
194 195 196 197
  std::array<std::vector<HitGroup>, 2> bi_records;
  // PADDLE_ENFORCE(!pattern_.edges().empty(), "At least one edge is needed");
  auto* first_pnode = pattern_.edges().empty() ? pattern().nodes().front().get()
                                               : pattern_.edges().front().first;
198 199 200 201 202 203 204 205 206 207 208 209 210
  if (!pdnodes2nodes_.count(first_pnode)) return result;
  for (auto* node : pdnodes2nodes_[first_pnode]) {
    HitGroup group;
    group.roles[first_pnode] = node;
    init_groups.emplace_back(group);
  }

  int step = 0;
  bi_records[0] = std::move(init_groups);

  // Extend a PDNode to subgraphs by deducing the connection relations defined
  // in edges of PDNodes.
  for (const auto& edge : pattern_.edges()) {
211
    VLOG(4) << "check " << edge.first->name() << " -> " << edge.second->name();
Y
Yan Chunwei 已提交
212
    // TODO(Superjomn) Fix bug here, the groups might be duplicate here.
213 214 215 216 217
    // Each role has two PDNodes, which indicates two roles.
    // Detect two Nodes that can match these two roles and they are connected.
    auto& pre_groups = bi_records[step % 2];
    auto& cur_groups = bi_records[1 - (step++ % 2)];
    cur_groups.clear();
218
    if (pre_groups.empty()) break;
219 220 221
    // source -> target
    for (Node* source : pdnodes2nodes_[edge.first]) {
      for (Node* target : pdnodes2nodes_[edge.second]) {
Y
Yan Chunwei 已提交
222
        VLOG(8) << "check " << source->id() << " -- " << target->id();
223 224 225 226 227 228 229 230 231 232 233 234 235 236 237
        // TODO(Superjomn) add some prune strategies.
        for (const auto& group : pre_groups) {
          HitGroup new_group = group;
          if (IsNodesLink(source, target) &&
              new_group.Match(source, edge.first)) {
            new_group.Register(source, edge.first);
            if (new_group.Match(target, edge.second)) {
              new_group.Register(target, edge.second);
              cur_groups.push_back(new_group);
              // TODO(Superjomn) need to unique
            }
          }
        }
      }
    }
238
    VLOG(3) << "step " << step << " get records: " << cur_groups.size();
Y
Yan Chunwei 已提交
239 240 241 242 243 244
    for (auto& group : cur_groups) {
      for (auto& item : group.roles) {
        VLOG(4) << "node " << item.second->id() << " as " << item.first->name();
      }
      VLOG(4) << "=========================================================";
    }
245 246 247
  }

  for (auto& group : bi_records[step % 2]) {
248
    GraphPatternDetector::subgraph_t subgraph;
249 250 251 252 253 254 255 256
    for (auto& role : group.roles) {
      subgraph.emplace(role.first, role.second);
    }
    result.emplace_back(subgraph);
  }
  return result;
}

257 258
void GraphPatternDetector::UniquePatterns(
    std::vector<GraphPatternDetector::subgraph_t>* subgraphs) {
259
  if (subgraphs->empty()) return;
260
  std::vector<GraphPatternDetector::subgraph_t> result;
261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276

  std::unordered_set<size_t> set;
  for (auto& g : *subgraphs) {
    size_t key = 0;
    for (auto& item : g) {
      key ^= std::hash<void*>{}(item.first);
      key ^= std::hash<void*>{}(item.second);
    }
    if (!set.count(key)) {
      result.emplace_back(g);
      set.insert(key);
    }
  }
  *subgraphs = result;
}

277
void GraphPatternDetector::RemoveOverlappedMatch(
278 279 280 281 282 283 284
    std::vector<subgraph_t>* subgraphs) {
  std::vector<subgraph_t> result;
  std::unordered_set<Node*> node_set;

  for (const auto& subgraph : *subgraphs) {
    bool valid = true;
    for (auto& item : subgraph) {
Y
Yan Chunwei 已提交
285
      if (item.first->IsIntermediate() && node_set.count(item.second)) {
286 287 288 289 290 291 292 293 294 295 296 297 298 299
        valid = false;
        break;
      }
    }
    if (valid) {
      for (auto& item : subgraph) {
        node_set.insert(item.second);
      }
      result.push_back(subgraph);
    }
  }
  *subgraphs = result;
}

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
std::string PDPattern::DotString() const {
  using inference::analysis::Dot;
  Dot dot;
  int id = 0;
  // Create Nodes
  std::unordered_map<PDNode*, std::string> node2dot;
  for (const auto& node : nodes()) {
    std::string node_id = "Node" + std::to_string(id++);
    dot.AddNode(node_id, {}, node->name());
    node2dot[node.get()] = node_id;
  }
  // Create Edges
  for (const auto& edge : edges()) {
    if (!node2dot.count(edge.first) || !node2dot.count(edge.second)) {
      LOG(ERROR) << "no node " << edge.first << " " << edge.second;
      continue;
    }
    auto& src = node2dot.at(edge.first);
    auto& trg = node2dot.at(edge.second);
    dot.AddEdge(src, trg, {});
  }
  return dot.Build();
}

PDNode& PDNode::LinksTo(const std::vector<PDNode*>& others) {
  // extend outlinks.
  for (PDNode* x : others) {
    pattern_->AddEdge(this, x);
  }
  return *this;
}

PDNode& PDNode::LinksFrom(const std::vector<PDNode*>& others) {
  // extend outlinks.
  for (PDNode* x : others) {
    pattern_->AddEdge(x, this);
  }
  return *this;
}

Y
Yan Chunwei 已提交
340
PDNode* PDNode::assert_is_op() {
341
  asserts_.emplace_back([](Node* x) { return x && x->IsOp(); });
Y
Yan Chunwei 已提交
342 343 344
  return this;
}
PDNode* PDNode::assert_is_op(const std::string& op_type) {
345
  asserts_.emplace_back([op_type](Node* x) {
Y
Yan Chunwei 已提交
346 347 348 349 350
    return x && x->IsOp() && x->Op()->Type() == op_type;
  });
  return this;
}
PDNode* PDNode::assert_is_var() {
351
  asserts_.emplace_back([](Node* x) { return x && x->IsVar(); });
Y
Yan Chunwei 已提交
352 353 354 355
  return this;
}
PDNode* PDNode::assert_var_not_persistable() {
  assert_is_var();
356
  asserts_.emplace_back([](Node* x) { return !x->Var()->Persistable(); });
Y
Yan Chunwei 已提交
357 358 359 360 361 362 363 364 365 366 367 368 369
  return this;
}
PDNode* PDNode::assert_is_persistable_var() {
  assert_is_var();
  asserts_.emplace_back([=](Node* x) { return x->Var()->Persistable(); });
  return this;
}
PDNode* PDNode::assert_is_op_nth_input(const std::string& op_type,
                                       const std::string& argument, int nth) {
  assert_is_var();
  assert_is_op_input(op_type);
  asserts_.emplace_back([=](Node* x) {
    for (auto* op : x->outputs) {
370 371 372
      if (op->IsOp() && op->Op()->Type() == op_type &&
          IsNthInput(x, op, argument, nth))
        return true;
Y
Yan Chunwei 已提交
373 374 375 376 377 378 379 380 381 382
    }
    return false;
  });
  return this;
}
PDNode* PDNode::assert_is_op_nth_output(const std::string& op_type,
                                        const std::string& argument, int nth) {
  assert_is_var();
  asserts_.emplace_back([=](Node* x) {
    for (auto* op : x->inputs) {
383 384 385
      if (op->IsOp() && op->Op()->Type() == op_type &&
          IsNthOutput(x, op, argument, nth))
        return true;
Y
Yan Chunwei 已提交
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
    }
    return false;
  });
  return this;
}
PDNode* PDNode::assert_is_only_input_of_op(const std::string& op_type) {
  assert_is_var();
  asserts_.emplace_back([=](Node* x) {
    for (auto* op : x->outputs) {
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type &&
          op->inputs.size() == 1) {
        return true;
      }
    }
    return false;
  });
  return this;
}
PDNode* PDNode::assert_is_only_output_of_op(const std::string& op_type) {
  assert_is_var();
  asserts_.emplace_back([=](Node* x) {
    for (auto* op : x->inputs) {
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type &&
          op->outputs.size() == 1) {
        return true;
      }
    }
    return false;
  });
  return this;
}
PDNode* PDNode::assert_is_op_output(const std::string& op_type) {
  assert_is_var();
  asserts_.emplace_back([=](Node* x) {
    for (auto* op : x->inputs) {
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type) {
        return true;
      }
    }
    return false;
  });
  return this;
}
429 430 431 432 433 434
PDNode* PDNode::assert_is_op_output(const std::string& op_type,
                                    const std::string& argument) {
  assert_is_var();
  assert_is_op_nth_output(op_type, argument, 0);
  return this;
}
Y
Yan Chunwei 已提交
435 436 437 438 439 440 441 442 443 444 445 446
PDNode* PDNode::assert_is_op_input(const std::string& op_type) {
  assert_is_var();
  asserts_.emplace_back([=](Node* x) {
    for (auto* op : x->outputs) {
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type) {
        return true;
      }
    }
    return false;
  });
  return this;
}
447 448 449 450 451 452
PDNode* PDNode::assert_is_op_input(const std::string& op_type,
                                   const std::string& argument) {
  assert_is_var();
  assert_is_op_nth_input(op_type, argument, 0);
  return this;
}
Y
Yan Chunwei 已提交
453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
PDNode* PDNode::assert_op_has_n_inputs(const std::string& op_type, size_t n) {
  assert_is_op(op_type);
  asserts_.emplace_back([=](Node* x) { return x->inputs.size() == n; });
  return this;
}
PDNode* PDNode::assert_op_has_n_outputs(const std::string& op_type, size_t n) {
  assert_is_op(op_type);
  asserts_.emplace_back([=](Node* x) { return x->outputs.size() == n; });
  return this;
}
PDNode* PDNode::assert_more(PDNode::teller_t&& teller) {
  asserts_.emplace_back(std::move(teller));
  return this;
}

468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497
bool VarLinksToOp(Node* node, const std::string& op_type) {
  for (auto* out : node->outputs) {
    if (out->IsOp() && out->Op()->Type() == op_type) {
      return true;
    }
  }
  return false;
}
bool IsNthInput(Node* var, Node* op, const std::string& argument, size_t nth) {
  PADDLE_ENFORCE(var->IsVar());
  PADDLE_ENFORCE(op->IsOp());
  if (op->Op()->Input(argument).size() <= nth) return false;
  return var->Name() == op->Op()->Input(argument)[nth];
}
bool IsNthOutput(Node* var, Node* op, const std::string& argument, size_t nth) {
  PADDLE_ENFORCE(var->IsVar());
  PADDLE_ENFORCE(op->IsOp());
  if (op->Op()->Output(argument).size() <= nth) return false;
  return var->Name() == op->Op()->Output(argument)[nth];
}
void GraphSafeRemoveNodes(Graph* graph,
                          const std::unordered_set<const Node*>& nodes) {
  for (auto* node : nodes) {
    graph->RemoveNode(const_cast<Node*>(node));
  }

  for (auto* node : graph->Nodes()) {
    for (auto it = node->inputs.begin(); it != node->inputs.end();) {
      if (nodes.count(*it)) {
        it = const_cast<Node*>(node)->inputs.erase(it);
498
      } else {
499
        it++;
500
      }
501 502 503 504
    }
    for (auto it = node->outputs.begin(); it != node->outputs.end();) {
      if (nodes.count(*it)) {
        it = const_cast<Node*>(node)->outputs.erase(it);
505
      } else {
506
        it++;
507
      }
508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567
    }
  }
}
bool VarLinksFromOp(Node* node, const std::string& op_type) {
  for (auto* out : node->inputs) {
    if (out->IsOp() && out->Op()->Type() == op_type) {
      return true;
    }
  }
  return false;
}

PDNode* patterns::FC(PDPattern* pattern, const std::string& name_scope,
                     PDNode* x, bool with_bias) {
  // Create Operators
  PDNode* elementwise_add_op{nullptr};
  auto* mul_op = pattern->NewNode(name_scope, "mul")->assert_is_op("mul");
  if (with_bias) {
    elementwise_add_op = pattern->NewNode(name_scope, "elementwise_add")
                             ->assert_is_op("elementwise_add");
  }
  // Create variables
  // w
  auto* mul_weight_var = pattern->NewNode(name_scope, "w")
                             ->AsInput()
                             ->assert_is_persistable_var()
                             ->assert_is_op_nth_input("mul", "Y", 0);
  PDNode* mul_out_var{nullptr};
  if (with_bias) {
    // intermediate variable, will be removed in the IR after fuse.
    mul_out_var = pattern->NewNode(name_scope, "mul_out")
                      ->AsIntermediate()
                      ->assert_is_only_output_of_op("mul")
                      ->assert_is_op_input("elementwise_add");
  }
  PDNode *bias{nullptr}, *fc_out{nullptr};
  if (with_bias) {
    // bias
    bias = pattern->NewNode(name_scope, "fc_bias")
               ->assert_is_op_input("elementwise_add")
               ->AsInput();
    // output
    fc_out = pattern->NewNode(name_scope, "fc_out")
                 ->AsOutput()
                 ->assert_is_op_output("elementwise_add");
  } else {
    fc_out = pattern->NewNode(name_scope, "fc_out")
                 ->AsOutput()
                 ->assert_is_op_output("mul");
  }

  if (with_bias) {
    mul_op->LinksFrom({mul_weight_var, x}).LinksTo({mul_out_var});
    elementwise_add_op->LinksFrom({mul_out_var, bias}).LinksTo({fc_out});
  } else {
    mul_op->LinksFrom({mul_weight_var, x}).LinksTo({fc_out});
  }

  return fc_out;
}
T
tensor-tang 已提交
568

569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592
PDNode* patterns::LSTM(PDPattern* pattern, const std::string& name_scope,
                       PDNode* x) {
  x->assert_is_op_input("lstm", "Input");
  auto* lstm_op = pattern->NewNode(name_scope, "lstm")->assert_is_op("lstm");
#define NEW_NODE(arg__, io__)                        \
  auto* arg__ = pattern->NewNode(name_scope, #arg__) \
                    ->assert_is_op_##io__("lstm", #arg__);

  // Currently, the H0 and C0 are optional
  // TODO(Superjomn) upgrade the fuse framework to support optional.
  // NEW_NODE(H0, input);
  // NEW_NODE(C0, input);
  NEW_NODE(Weight, input);
  NEW_NODE(Bias, input);

  NEW_NODE(Hidden, output);
  NEW_NODE(Cell, output);
  NEW_NODE(BatchGate, output);
  NEW_NODE(BatchCellPreAct, output);

  lstm_op->LinksFrom({x, Weight, Bias});
  lstm_op->LinksTo({Hidden, Cell, BatchGate, BatchCellPreAct});
  return Hidden;
}
T
tensor-tang 已提交
593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618

PDNode* patterns::GRU(PDPattern* pattern, const std::string& name_scope,
                      PDNode* x) {
  x->assert_is_op_input("gru", "Input");
  auto* gru_op = pattern->NewNode(name_scope, "gru")->assert_is_op("gru");
#define NEW_NODE(arg__, io__)                        \
  auto* arg__ = pattern->NewNode(name_scope, #arg__) \
                    ->assert_is_op_##io__("gru", #arg__);

  NEW_NODE(Weight, input);
  // TODO(Superjomn): upgrade the fuse framework to support optional.
  // H0 and bias are optional
  NEW_NODE(Bias, input);  // also optional
  // NEW_NODE(H0, input);

  NEW_NODE(Hidden, output);
  // below are intermediate
  NEW_NODE(BatchGate, output);
  NEW_NODE(BatchResetHiddenPrev, output);
  NEW_NODE(BatchHidden, output);

  gru_op->LinksFrom({x, Weight, Bias});
  gru_op->LinksTo({Hidden, BatchGate, BatchResetHiddenPrev, BatchHidden});
  return Hidden;
}

619 620 621
}  // namespace ir
}  // namespace framework
}  // namespace paddle