cinn_cache_key.cc 4.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
// Copyright (c) 2021 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.

#include "paddle/fluid/framework/paddle2cinn/cinn_cache_key.h"

17 18
#include <algorithm>
#include <functional>
19
#include <map>
20
#include <set>
21 22 23 24
#include <string>

#include "paddle/fluid/framework/ir/graph.h"
#include "paddle/fluid/framework/lod_tensor.h"
25
#include "paddle/pten/core/ddim.h"
26 27 28 29 30

namespace paddle {
namespace framework {
namespace paddle2cinn {

J
jiangcheng 已提交
31 32 33 34 35
using GraphHashStrategy = CinnCacheKey::GraphHashStrategy;

CinnCacheKey::CinnCacheKey(GraphHashStrategy graph_hash)
    : graph_hash_(graph_hash) {}

36 37
CinnCacheKey::CinnCacheKey(
    const ir::Graph& graph,
38
    const std::map<std::string, const LoDTensor*>& input_tensors,
J
jiangcheng 已提交
39 40
    const std::string& arch_str, GraphHashStrategy graph_hash)
    : graph_hash_(graph_hash) {
41
  this->SetKey(graph, input_tensors, arch_str);
42 43 44
}

CinnCacheKey::CinnCacheKey(const ir::Graph& graph,
45
                           const std::map<std::string, DDim>& input_shapes,
J
jiangcheng 已提交
46 47 48
                           const std::string& arch_str,
                           GraphHashStrategy graph_hash)
    : graph_hash_(graph_hash) {
49
  this->SetKey(graph, input_shapes, arch_str);
50 51 52 53
}

void CinnCacheKey::SetKey(
    const ir::Graph& graph,
54 55
    const std::map<std::string, const LoDTensor*>& input_tensors,
    const std::string& arch_str) {
J
jiangcheng 已提交
56
  graph_hash_val_ = graph_hash_(graph);
57 58
  for (const auto& name_tensor : input_tensors) {
    input_shapes_[name_tensor.first] = name_tensor.second->dims();
59
  }
60
  arch_str_ = arch_str;
61 62 63
}

void CinnCacheKey::SetKey(const ir::Graph& graph,
64 65
                          const std::map<std::string, DDim>& input_shapes,
                          const std::string& arch_str) {
J
jiangcheng 已提交
66
  graph_hash_val_ = graph_hash_(graph);
67 68
  input_shapes_ = input_shapes;
  arch_str_ = arch_str;
69 70 71 72 73 74 75
}

bool CinnCacheKey::operator!=(const CinnCacheKey& other) const {
  return !this->operator==(other);
}

bool CinnCacheKey::operator==(const CinnCacheKey& other) const {
J
jiangcheng 已提交
76
  return graph_hash_val_ == other.graph_hash_val_ &&
77
         input_shapes_ == other.input_shapes_ && arch_str_ == other.arch_str_;
78 79 80 81 82 83 84 85 86 87
}

size_t CinnCacheKey::Hash::hash_combine(size_t seed, size_t value) {
  return seed ^ (value + 0x9e3779b9 + (seed << 6) + (seed >> 2));
}

size_t CinnCacheKey::Hash::operator()(const CinnCacheKey& key) const {
  std::size_t ret = 0;

  std::hash<std::string> string_hasher;
88
  for (const auto& name_shape : key.input_shapes_) {
89 90 91 92
    ret = hash_combine(ret, string_hasher(name_shape.first));
    ret = hash_combine(ret, string_hasher(name_shape.second.to_str()));
  }

J
jiangcheng 已提交
93
  ret = hash_combine(ret, key.graph_hash_val_);
94
  ret = hash_combine(ret, string_hasher(key.arch_str_));
95 96 97
  return ret;
}

J
jiangcheng 已提交
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
size_t CinnCacheKeyByStructure::HashGraph(const ir::Graph& graph) {
  // sort grad node by name and id.
  auto compare = [](ir::Node* n1, ir::Node* n2) {
    return (n1->Name() == n2->Name()) ? (n1->id() < n2->id())
                                      : (n1->Name() < n2->Name());
  };

  // graph.Nodes() return unordered_set, here using set to avoid the same graph
  // may return different result
  std::set<ir::Node *, bool (*)(ir::Node *, ir::Node *)> node_set(compare),
      output_set(compare);
  node_set.insert(graph.Nodes().begin(), graph.Nodes().end());

  std::string hash_str;
  for (ir::Node* n : node_set) {
    hash_str.append(n->Name());

    output_set.clear();
    output_set.insert(n->outputs.begin(), n->outputs.end());
    for (auto* out : output_set) {
      hash_str.append(out->Name());
    }
  }

  VLOG(1) << "The hash graph:\n" << hash_str;

  size_t hash_val = std::hash<std::string>()(hash_str);
  VLOG(4) << "The graph's hash value by graph structure is: " << hash_val;
  return hash_val;
}

size_t CinnCacheKeyByAddress::HashGraph(const ir::Graph& graph) {
  size_t hash_val = reinterpret_cast<size_t>(&graph);
  VLOG(4) << "The graph's hash value by graph address is: " << hash_val;
  return hash_val;
}

135 136 137
}  // namespace paddle2cinn
}  // namespace framework
}  // namespace paddle