tester_helper.h 28.6 KB
Newer Older
L
luotao1 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
// 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.

#pragma once

#include <gtest/gtest.h>
Y
Yan Chunwei 已提交
18

L
luotao1 已提交
19
#include <algorithm>
L
luotao1 已提交
20
#include <memory>
T
Tao Luo 已提交
21
#include <string>
L
luotao1 已提交
22
#include <thread>  // NOLINT
L
luotao1 已提交
23
#include <unordered_map>
L
luotao1 已提交
24
#include <vector>
Y
Yiqun Liu 已提交
25 26 27
#ifdef WITH_GPERFTOOLS
#include <gperftools/profiler.h>
#endif
L
luotao1 已提交
28
#include "paddle/fluid/framework/ir/fuse_pass_base.h"
29
#include "paddle/fluid/framework/scope.h"
L
luotao1 已提交
30 31 32
#include "paddle/fluid/inference/analysis/analyzer.h"
#include "paddle/fluid/inference/analysis/ut_helper.h"
#include "paddle/fluid/inference/api/analysis_predictor.h"
33
#include "paddle/fluid/inference/api/helper.h"
Y
Yan Chunwei 已提交
34
#include "paddle/fluid/inference/api/paddle_inference_pass.h"
35
#include "paddle/fluid/inference/tests/api/config_printer.h"
T
Tao Luo 已提交
36
#include "paddle/fluid/inference/tests/test_helper.h"
N
nhzlx 已提交
37
#include "paddle/fluid/inference/utils/benchmark.h"
L
luotao1 已提交
38 39
#include "paddle/fluid/platform/profiler.h"

N
nhzlx 已提交
40
DEFINE_string(model_name, "", "model name");
L
luotao1 已提交
41 42
DEFINE_string(infer_model, "", "model path");
DEFINE_string(infer_data, "", "data file");
T
Tao Luo 已提交
43
DEFINE_string(refer_result, "", "reference result for comparison");
44
DEFINE_int32(batch_size, 1, "batch size");
45 46
DEFINE_bool(enable_fp32, true, "Enable FP32 type prediction");
DEFINE_bool(enable_int8, true, "Enable INT8 type prediction");
47 48 49
DEFINE_int32(warmup_batch_size, 100, "batch size for quantization warmup");
// setting iterations to 0 means processing the whole dataset
DEFINE_int32(iterations, 0, "number of batches to process");
L
luotao1 已提交
50 51 52
DEFINE_int32(repeat, 1, "Running the inference program repeat times.");
DEFINE_bool(test_all_data, false, "Test the all dataset in data file.");
DEFINE_int32(num_threads, 1, "Running the inference program in multi-threads.");
T
Tao Luo 已提交
53 54
DEFINE_bool(use_analysis, true,
            "Running the inference program in analysis mode.");
N
nhzlx 已提交
55 56
DEFINE_bool(record_benchmark, false,
            "Record benchmark after profiling the model");
L
luotao1 已提交
57
DEFINE_double(accuracy, 1e-3, "Result Accuracy.");
58
DEFINE_double(quantized_accuracy, 1e-2, "Result Quantized Accuracy.");
L
luotao1 已提交
59
DEFINE_bool(zero_copy, false, "Use ZeroCopy to speedup Feed/Fetch.");
60 61 62
DEFINE_bool(warmup, false,
            "Use warmup to calculate elapsed_time more accurately. "
            "To reduce CI time, it sets false in default.");
L
luotao1 已提交
63

64
DECLARE_bool(profile);
L
luotao1 已提交
65
DECLARE_int32(paddle_num_threads);
66

L
luotao1 已提交
67 68 69
namespace paddle {
namespace inference {

70 71
using paddle::framework::proto::VarType;

72 73 74 75 76 77 78 79 80 81 82 83 84
template <typename T>
constexpr paddle::PaddleDType GetPaddleDType();

template <>
constexpr paddle::PaddleDType GetPaddleDType<int64_t>() {
  return paddle::PaddleDType::INT64;
}

template <>
constexpr paddle::PaddleDType GetPaddleDType<float>() {
  return paddle::PaddleDType::FLOAT32;
}

85
void PrintConfig(const PaddlePredictor::Config *config, bool use_analysis) {
86
  const auto *analysis_config =
87
      reinterpret_cast<const AnalysisConfig *>(config);
88
  if (use_analysis) {
89
    LOG(INFO) << *analysis_config;
90 91
    return;
  }
92
  LOG(INFO) << analysis_config->ToNativeConfig();
93
}
Y
Yan Chunwei 已提交
94

95 96 97 98 99 100 101 102
void CheckError(float data_ref, float data) {
  if (std::abs(data_ref) > 1) {
    CHECK_LE(std::abs((data_ref - data) / data_ref), FLAGS_accuracy);
  } else {
    CHECK_LE(std::abs(data_ref - data), FLAGS_accuracy);
  }
}

103
// Compare result between two PaddleTensor
L
luotao1 已提交
104
void CompareResult(const std::vector<PaddleTensor> &outputs,
T
tensor-tang 已提交
105
                   const std::vector<PaddleTensor> &ref_outputs) {
T
Tao Luo 已提交
106
  EXPECT_GT(outputs.size(), 0UL);
T
tensor-tang 已提交
107
  EXPECT_EQ(outputs.size(), ref_outputs.size());
L
luotao1 已提交
108 109
  for (size_t i = 0; i < outputs.size(); i++) {
    auto &out = outputs[i];
T
tensor-tang 已提交
110
    auto &ref_out = ref_outputs[i];
111 112
    size_t size = VecReduceToInt(out.shape);
    size_t ref_size = VecReduceToInt(ref_out.shape);
113
    EXPECT_GT(size, 0UL);
T
tensor-tang 已提交
114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
    EXPECT_EQ(size, ref_size);
    EXPECT_EQ(out.dtype, ref_out.dtype);
    switch (out.dtype) {
      case PaddleDType::INT64: {
        int64_t *pdata = static_cast<int64_t *>(out.data.data());
        int64_t *pdata_ref = static_cast<int64_t *>(ref_out.data.data());
        for (size_t j = 0; j < size; ++j) {
          EXPECT_EQ(pdata_ref[j], pdata[j]);
        }
        break;
      }
      case PaddleDType::FLOAT32: {
        float *pdata = static_cast<float *>(out.data.data());
        float *pdata_ref = static_cast<float *>(ref_out.data.data());
        for (size_t j = 0; j < size; ++j) {
129
          CheckError(pdata_ref[j], pdata[j]);
T
tensor-tang 已提交
130 131 132
        }
        break;
      }
133 134 135 136 137 138 139 140
      case PaddleDType::INT32: {
        int32_t *pdata = static_cast<int32_t *>(out.data.data());
        int32_t *pdata_ref = static_cast<int32_t *>(ref_out.data.data());
        for (size_t j = 0; j < size; ++j) {
          EXPECT_EQ(pdata_ref[j], pdata[j]);
        }
        break;
      }
141 142 143 144 145 146 147 148
      case PaddleDType::UINT8: {
        uint8_t *pdata = static_cast<uint8_t *>(out.data.data());
        uint8_t *pdata_ref = static_cast<uint8_t *>(ref_out.data.data());
        for (size_t j = 0; j < size; ++j) {
          EXPECT_EQ(pdata_ref[j], pdata[j]);
        }
        break;
      }
L
luotao1 已提交
149 150 151 152
    }
  }
}

153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
// Compare result between a PaddleTensor and a ZeroCopyTensor
void CompareResult(const std::vector<PaddleTensor> &outputs,
                   const std::vector<ZeroCopyTensor> &ref_outputs) {
  EXPECT_GT(outputs.size(), 0UL);
  EXPECT_EQ(outputs.size(), ref_outputs.size());
  for (size_t i = 0; i < outputs.size(); i++) {
    auto &out = outputs[i];
    auto &ref_out = ref_outputs[i];
    size_t size = VecReduceToInt(out.shape);
    EXPECT_GT(size, 0UL);
    int ref_size = 0;  // this is the number of elements not memory size
    PaddlePlace place;
    switch (out.dtype) {
      case PaddleDType::INT64: {
        int64_t *pdata = static_cast<int64_t *>(out.data.data());
        int64_t *pdata_ref = ref_out.data<int64_t>(&place, &ref_size);
169
        EXPECT_EQ(size, static_cast<size_t>(ref_size));
170 171 172 173 174 175 176 177 178 179
        for (size_t j = 0; j < size; ++j) {
          EXPECT_EQ(pdata_ref[j], pdata[j]);
        }
        break;
      }
      case PaddleDType::FLOAT32: {
        float *pdata = static_cast<float *>(out.data.data());
        float *pdata_ref = ref_out.data<float>(&place, &ref_size);
        EXPECT_EQ(size, ref_size);
        for (size_t j = 0; j < size; ++j) {
180
          CheckError(pdata_ref[j], pdata[j]);
181 182 183
        }
        break;
      }
L
luotao1 已提交
184 185 186 187 188 189 190 191 192
      case PaddleDType::INT32: {
        int32_t *pdata = static_cast<int32_t *>(out.data.data());
        int32_t *pdata_ref = ref_out.data<int32_t>(&place, &ref_size);
        EXPECT_EQ(size, ref_size);
        for (size_t j = 0; j < size; ++j) {
          EXPECT_EQ(pdata_ref[j], pdata[j]);
        }
        break;
      }
193 194 195 196 197 198 199 200 201
      case PaddleDType::UINT8: {
        uint8_t *pdata = static_cast<uint8_t *>(out.data.data());
        uint8_t *pdata_ref = ref_out.data<uint8_t>(&place, &ref_size);
        EXPECT_EQ(size, ref_size);
        for (size_t j = 0; j < size; ++j) {
          EXPECT_EQ(pdata_ref[j], pdata[j]);
        }
        break;
      }
202 203 204 205
    }
  }
}

206
std::unique_ptr<PaddlePredictor> CreateTestPredictor(
207
    const PaddlePredictor::Config *config, bool use_analysis = true) {
208
  const auto *analysis_config =
209
      reinterpret_cast<const AnalysisConfig *>(config);
T
Tao Luo 已提交
210
  if (use_analysis) {
211
    return CreatePaddlePredictor<AnalysisConfig>(*analysis_config);
T
Tao Luo 已提交
212
  }
213 214
  auto native_config = analysis_config->ToNativeConfig();
  return CreatePaddlePredictor<NativeConfig>(native_config);
T
Tao Luo 已提交
215 216
}

217
size_t GetSize(const PaddleTensor &out) { return VecReduceToInt(out.shape); }
T
Tao Luo 已提交
218

219
std::unordered_map<std::string, int> GetFuseStatis(PaddlePredictor *predictor,
T
Tao Luo 已提交
220
                                                   int *num_ops) {
221
  std::unordered_map<std::string, int> res;
222
  auto *analysis_predictor = static_cast<AnalysisPredictor *>(predictor);
223 224 225 226 227 228
  auto *fusion_status =
      analysis_predictor->analysis_argument().fusion_statis_ptr();
  if (!fusion_status) {
    return res;
  }
  for (auto &item : *fusion_status) {
T
Tao Luo 已提交
229 230 231 232
    LOG(INFO) << "fused " << item.first << " " << item.second;
  }
  int num = 0;
  for (auto &node :
233 234
       analysis_predictor->analysis_argument().main_graph().Nodes()) {
    if (node->IsOp()) {
T
Tao Luo 已提交
235 236 237 238
      ++num;
    }
  }
  *num_ops = num;
239
  return *fusion_status;
T
Tao Luo 已提交
240 241
}

T
Tao Luo 已提交
242
void SetFakeImageInput(std::vector<std::vector<PaddleTensor>> *inputs,
243 244
                       const std::string &dirname, bool is_combined = true,
                       std::string model_filename = "model",
T
tensor-tang 已提交
245
                       std::string params_filename = "params",
N
nhzlx 已提交
246 247
                       const std::vector<std::string> *feed_names = nullptr,
                       const int continuous_inuput_index = 0) {
T
Tao Luo 已提交
248 249
  // Set fake_image_data
  PADDLE_ENFORCE_EQ(FLAGS_test_all_data, 0, "Only have single batch of data.");
250 251 252 253 254 255 256 257 258 259 260
  std::vector<std::vector<int64_t>> feed_target_shapes = GetFeedTargetShapes(
      dirname, is_combined, model_filename, params_filename);
  std::ostringstream os;
  for (size_t i = 0; i < feed_target_shapes.size(); ++i) {
    os << "feed target " << i << ": {" << feed_target_shapes[i][0];
    for (size_t j = 1; j < feed_target_shapes[i].size(); ++j) {
      os << ", " << feed_target_shapes[i][j];
    }
    os << "}\n";
  }
  LOG(INFO) << os.str();
T
tensor-tang 已提交
261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
  if (feed_names) {
    PADDLE_ENFORCE_EQ(feed_names->size(), feed_target_shapes.size());
  }
  std::vector<PaddleTensor> input_slots(feed_target_shapes.size());
  for (size_t i = 0; i < feed_target_shapes.size(); ++i) {
    const auto &feed_shape = feed_target_shapes[i];
    auto &input = input_slots[i];
    std::vector<int> shape({FLAGS_batch_size});
    for (size_t s = 1; s < feed_shape.size(); ++s) {
      shape.push_back(static_cast<int>(feed_shape[s]));
    }
    if (feed_names) {
      input.name = (*feed_names)[i];
    }
    input.shape = shape;
    input.dtype = PaddleDType::FLOAT32;
277
    size_t len = std::accumulate(shape.begin(), shape.end(), size_t{1},
T
tensor-tang 已提交
278 279 280 281 282 283
                                 [](int a, int b) { return a * b; });
    input.data.Resize(len * sizeof(float));
    input.lod.assign({{0, static_cast<size_t>(FLAGS_batch_size)}});
    float *input_data = static_cast<float *>(input.data.data());
    // fill input data, for profile easily, do not use random data here.
    for (size_t j = 0; j < len; ++j) {
N
nhzlx 已提交
284 285
      *(input_data + j) =
          static_cast<float>((j + continuous_inuput_index) % len) / len;
T
tensor-tang 已提交
286
    }
T
Tao Luo 已提交
287 288 289 290
  }
  (*inputs).emplace_back(input_slots);
}

291 292 293 294 295 296 297 298 299 300 301 302
void GetInputPerBatch(const std::vector<std::vector<int64_t>> &in,
                      std::vector<std::vector<int64_t>> *out,
                      std::vector<size_t> *lod, size_t batch_iter,
                      size_t batch_end) {
  lod->clear();
  lod->push_back(0);
  for (auto it = in.begin() + batch_iter; it < in.begin() + batch_end; it++) {
    out->push_back(*it);
    lod->push_back(lod->back() + (*it).size());  // calculate lod
  }
}

L
luotao1 已提交
303 304 305 306 307 308 309 310 311 312 313
void ConvertPaddleTensorToZeroCopyTensor(
    PaddlePredictor *predictor, const std::vector<PaddleTensor> &inputs) {
  for (size_t i = 0; i < inputs.size(); i++) {
    auto input = inputs[i];
    auto tensor = predictor->GetInputTensor(input.name);
    tensor->Reshape(input.shape);
    tensor->SetLoD({input.lod});
    if (input.dtype == PaddleDType::INT64) {
      ZeroCopyTensorAssignData<int64_t>(tensor.get(), input.data);
    } else if (input.dtype == PaddleDType::FLOAT32) {
      ZeroCopyTensorAssignData<float>(tensor.get(), input.data);
L
luotao1 已提交
314 315
    } else if (input.dtype == PaddleDType::INT32) {
      ZeroCopyTensorAssignData<int32_t>(tensor.get(), input.data);
316 317
    } else if (input.dtype == PaddleDType::UINT8) {
      ZeroCopyTensorAssignData<uint8_t>(tensor.get(), input.data);
L
luotao1 已提交
318 319 320 321 322
    } else {
      LOG(ERROR) << "unsupported feed type " << input.dtype;
    }
  }
}
323

L
luotao1 已提交
324 325
void PredictionWarmUp(PaddlePredictor *predictor,
                      const std::vector<std::vector<PaddleTensor>> &inputs,
326
                      std::vector<std::vector<PaddleTensor>> *outputs,
327 328
                      int num_threads, int tid,
                      const VarType::Type data_type = VarType::FP32) {
L
luotao1 已提交
329 330 331 332 333
  int batch_size = FLAGS_batch_size;
  LOG(INFO) << "Running thread " << tid << ", warm up run...";
  if (FLAGS_zero_copy) {
    ConvertPaddleTensorToZeroCopyTensor(predictor, inputs[0]);
  }
334
  outputs->resize(1);
L
luotao1 已提交
335 336 337
  Timer warmup_timer;
  warmup_timer.tic();
  if (!FLAGS_zero_copy) {
338
    predictor->Run(inputs[0], &(*outputs)[0], batch_size);
L
luotao1 已提交
339 340
  } else {
    predictor->ZeroCopyRun();
341
  }
342
  PrintTime(batch_size, 1, num_threads, tid, warmup_timer.toc(), 1, data_type);
L
luotao1 已提交
343 344 345 346
  if (FLAGS_profile) {
    paddle::platform::ResetProfiler();
  }
}
347

L
luotao1 已提交
348 349
void PredictionRun(PaddlePredictor *predictor,
                   const std::vector<std::vector<PaddleTensor>> &inputs,
350
                   std::vector<std::vector<PaddleTensor>> *outputs,
351
                   int num_threads, int tid,
352 353
                   const VarType::Type data_type = VarType::FP32,
                   float *sample_latency = nullptr) {
L
luotao1 已提交
354
  int num_times = FLAGS_repeat;
355
  int iterations = inputs.size();  // process the whole dataset ...
356 357
  if (FLAGS_iterations > 0 &&
      FLAGS_iterations < static_cast<int64_t>(inputs.size()))
358 359 360 361 362
    iterations =
        FLAGS_iterations;  // ... unless the number of iterations is set
  outputs->resize(iterations);
  LOG(INFO) << "Thread " << tid << ", number of threads " << num_threads
            << ", run " << num_times << " times...";
L
luotao1 已提交
363 364
  Timer run_timer;
  double elapsed_time = 0;
Y
Yiqun Liu 已提交
365
#ifdef WITH_GPERFTOOLS
L
luotao1 已提交
366
  ProfilerStart("paddle_inference.prof");
Y
Yiqun Liu 已提交
367
#endif
368
  int predicted_num = 0;
L
luotao1 已提交
369
  if (!FLAGS_zero_copy) {
370
    for (int i = 0; i < iterations; i++) {
371
      run_timer.tic();
L
luotao1 已提交
372
      for (int j = 0; j < num_times; j++) {
373
        predictor->Run(inputs[i], &(*outputs)[i], FLAGS_batch_size);
374
      }
375 376 377 378 379 380
      elapsed_time += run_timer.toc();

      predicted_num += FLAGS_batch_size;
      if (predicted_num % 100 == 0) {
        LOG(INFO) << predicted_num << " samples";
      }
L
luotao1 已提交
381
    }
L
luotao1 已提交
382
  } else {
383
    for (int i = 0; i < iterations; i++) {
L
luotao1 已提交
384 385 386 387 388 389
      ConvertPaddleTensorToZeroCopyTensor(predictor, inputs[i]);
      run_timer.tic();
      for (int j = 0; j < num_times; j++) {
        predictor->ZeroCopyRun();
      }
      elapsed_time += run_timer.toc();
390 391 392 393 394

      predicted_num += FLAGS_batch_size;
      if (predicted_num % 100 == 0) {
        LOG(INFO) << predicted_num << " samples";
      }
L
luotao1 已提交
395 396
    }
  }
397

Y
Yiqun Liu 已提交
398
#ifdef WITH_GPERFTOOLS
L
luotao1 已提交
399
  ProfilerStop();
Y
Yiqun Liu 已提交
400
#endif
N
nhzlx 已提交
401

402 403
  auto batch_latency = elapsed_time / (iterations * num_times);
  PrintTime(FLAGS_batch_size, num_times, num_threads, tid, batch_latency,
404
            iterations, data_type);
405 406 407 408

  if (sample_latency != nullptr)
    *sample_latency = batch_latency / FLAGS_batch_size;

L
luotao1 已提交
409 410 411
  if (FLAGS_record_benchmark) {
    Benchmark benchmark;
    benchmark.SetName(FLAGS_model_name);
412 413
    benchmark.SetBatchSize(FLAGS_batch_size);
    benchmark.SetLatency(batch_latency);
L
luotao1 已提交
414
    benchmark.PersistToFile("benchmark_record.txt");
L
luotao1 已提交
415 416 417
  }
}

L
luotao1 已提交
418 419 420
void TestOneThreadPrediction(
    const PaddlePredictor::Config *config,
    const std::vector<std::vector<PaddleTensor>> &inputs,
421
    std::vector<std::vector<PaddleTensor>> *outputs, bool use_analysis = true,
422 423
    const VarType::Type data_type = VarType::FP32,
    float *sample_latency = nullptr) {
L
luotao1 已提交
424
  auto predictor = CreateTestPredictor(config, use_analysis);
425
  if (FLAGS_warmup) {
426
    PredictionWarmUp(predictor.get(), inputs, outputs, 1, 0, data_type);
427
  }
428 429
  PredictionRun(predictor.get(), inputs, outputs, 1, 0, data_type,
                sample_latency);
L
luotao1 已提交
430 431
}

L
luotao1 已提交
432
void TestMultiThreadPrediction(
433
    const PaddlePredictor::Config *config,
434
    const std::vector<std::vector<PaddleTensor>> &inputs,
435
    std::vector<std::vector<PaddleTensor>> *outputs, int num_threads,
T
Tao Luo 已提交
436
    bool use_analysis = true) {
L
luotao1 已提交
437
  std::vector<std::thread> threads;
L
luotao1 已提交
438 439 440 441 442
  std::vector<std::unique_ptr<PaddlePredictor>> predictors;
  predictors.emplace_back(CreateTestPredictor(config, use_analysis));
  for (int tid = 1; tid < num_threads; tid++) {
    predictors.emplace_back(predictors.front()->Clone());
  }
443

L
luotao1 已提交
444 445 446 447
  for (int tid = 0; tid < num_threads; ++tid) {
    threads.emplace_back([&, tid]() {
      // Each thread should have local inputs and outputs.
      // The inputs of each thread are all the same.
448
      std::vector<std::vector<PaddleTensor>> outputs_tid;
L
luotao1 已提交
449
      auto &predictor = predictors[tid];
450 451 452 453
      if (FLAGS_warmup) {
        PredictionWarmUp(predictor.get(), inputs, &outputs_tid, num_threads,
                         tid);
      }
454
      PredictionRun(predictor.get(), inputs, &outputs_tid, num_threads, tid);
L
luotao1 已提交
455 456 457 458 459 460 461
    });
  }
  for (int i = 0; i < num_threads; ++i) {
    threads[i].join();
  }
}

462
void TestPrediction(const PaddlePredictor::Config *config,
463
                    const std::vector<std::vector<PaddleTensor>> &inputs,
464 465
                    std::vector<std::vector<PaddleTensor>> *outputs,
                    int num_threads, bool use_analysis = FLAGS_use_analysis) {
466
  PrintConfig(config, use_analysis);
L
luotao1 已提交
467
  if (num_threads == 1) {
T
Tao Luo 已提交
468
    TestOneThreadPrediction(config, inputs, outputs, use_analysis);
L
luotao1 已提交
469
  } else {
T
Tao Luo 已提交
470 471
    TestMultiThreadPrediction(config, inputs, outputs, num_threads,
                              use_analysis);
L
luotao1 已提交
472 473 474
  }
}

475 476
void SummarizeAccuracy(float avg_acc_fp32, float avg_acc_int8,
                       int compared_idx) {
477 478 479 480 481 482
  PADDLE_ENFORCE_LE(compared_idx, 2,
                    "Compare either top1 accuracy or mAP (top5), the "
                    "compared_idx is out of range");
  PADDLE_ENFORCE_GE(compared_idx, 1,
                    "Compare either top1 accuracy or mAP (top5), the "
                    "compared_idx is out of range");
483
  std::string prefix = (compared_idx == 1) ? "top1_accuracy " : "mAP ";
484
  LOG(INFO) << "--- Accuracy summary --- ";
485 486 487 488 489 490 491 492
  LOG(INFO) << "Accepted " << prefix
            << "drop threshold: " << FLAGS_quantized_accuracy
            << ". (condition: (FP32_" << prefix << " - INT8_" << prefix
            << ") <= threshold)";
  LOG(INFO) << "FP32: avg " << prefix << std::fixed << std::setw(6)
            << std::setprecision(4) << avg_acc_fp32;
  LOG(INFO) << "INT8: avg " << prefix << std::fixed << std::setw(6)
            << std::setprecision(4) << avg_acc_int8;
493 494
}

495 496 497 498 499 500 501 502
void SummarizePerformance(const char *title, float sample) {
  CHECK_GT(sample, 0.0);
  auto throughput = 1000.0 / sample;
  LOG(INFO) << title << ": avg fps: " << std::fixed << std::setw(6)
            << std::setprecision(4) << throughput << ", avg latency: " << sample
            << " ms";
}

503 504
void SummarizePerformance(float sample_latency_fp32,
                          float sample_latency_int8) {
505 506
  if (FLAGS_enable_fp32) SummarizePerformance("FP32", sample_latency_fp32);
  if (FLAGS_enable_int8) SummarizePerformance("INT8", sample_latency_int8);
507 508
}

509 510
float CompareAccuracyOne(
    const std::vector<std::vector<PaddleTensor>> &output_slots,
511
    int compared_idx) {
512
  if (output_slots.size() == 0)
513
    throw std::invalid_argument(
514
        "CompareAccuracy: output_slots vector is empty.");
515

516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
  float total_accs{0};

  for (size_t i = 0; i < output_slots.size(); ++i) {
    switch (compared_idx) {
      case 1:
        PADDLE_ENFORCE_GE(
            output_slots[i].size(), 2UL,
            "To achieve top 1 accuracy, output_slots_quant[i].size()>=2");
        break;
      case 2:
        PADDLE_ENFORCE_GE(
            output_slots[i].size(), 2UL,
            "To achieve top 1 accuracy, output_slots_ref[i].size()>=2");
        break;
      default:
        throw std::invalid_argument(
            "CompareAccuracy: compared_idx is out of range.");
533 534
    }

535
    if (output_slots[i][compared_idx].lod.size() > 0)
536
      throw std::invalid_argument("CompareAccuracy: output has nonempty LoD.");
537 538

    if (output_slots[i][compared_idx].dtype != paddle::PaddleDType::FLOAT32)
539
      throw std::invalid_argument(
540
          "CompareAccuracy: output is of a wrong type.");
541 542 543

    total_accs +=
        *static_cast<float *>(output_slots[i][compared_idx].data.data());
544
  }
545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567

  CHECK_GT(output_slots.size(), 0);

  return total_accs / output_slots.size();
}

void CompareAccuracy(
    const std::vector<std::vector<PaddleTensor>> &output_slots_quant,
    const std::vector<std::vector<PaddleTensor>> &output_slots_ref,
    int compared_idx) {
  if ((FLAGS_enable_fp32 && FLAGS_enable_int8) &&
      (output_slots_quant.size() == 0 || output_slots_ref.size()) == 0)
    throw std::invalid_argument(
        "CompareAccuracy: output_slots vector is empty.");

  float avg_acc_quant = 0.0;
  float avg_acc_ref = 0.0;

  if (FLAGS_enable_int8)
    avg_acc_quant = CompareAccuracyOne(output_slots_quant, compared_idx);

  if (FLAGS_enable_fp32)
    avg_acc_ref = CompareAccuracyOne(output_slots_ref, compared_idx);
568

569
  SummarizeAccuracy(avg_acc_ref, avg_acc_quant, compared_idx);
570 571 572 573 574 575 576

  if (FLAGS_enable_fp32) CHECK_GT(avg_acc_ref, 0.0);

  if (FLAGS_enable_int8) CHECK_GT(avg_acc_quant, 0.0);

  if (FLAGS_enable_fp32 && FLAGS_enable_int8)
    CHECK_LE(avg_acc_ref - avg_acc_quant, FLAGS_quantized_accuracy);
577 578
}

L
luotao1 已提交
579 580 581 582 583 584 585 586 587
void CompareDeterministic(
    const PaddlePredictor::Config *config,
    const std::vector<std::vector<PaddleTensor>> &inputs) {
  int batch_size = FLAGS_batch_size;
  int num_times = FLAGS_repeat;
  auto predictor = CreateTestPredictor(config, FLAGS_use_analysis);

  std::vector<PaddleTensor> warmup_outputs, outputs;
  // run num_times to Compare Deterministic Result.
588 589 590 591
  for (size_t j = 0; j < inputs.size(); j++) {
    // warmup run
    predictor->Run(inputs[j], &warmup_outputs, batch_size);
    for (int i = 0; i < num_times; i++) {
L
luotao1 已提交
592 593 594 595 596 597
      predictor->Run(inputs[j], &outputs, batch_size);
      CompareResult(outputs, warmup_outputs);
    }
  }
}

T
Tao Luo 已提交
598
void CompareNativeAndAnalysis(
599
    const PaddlePredictor::Config *config,
600
    const std::vector<std::vector<PaddleTensor>> &inputs) {
601
  PrintConfig(config, true);
602
  std::vector<std::vector<PaddleTensor>> native_outputs, analysis_outputs;
603
  TestOneThreadPrediction(config, inputs, &native_outputs, false);
T
Tao Luo 已提交
604
  TestOneThreadPrediction(config, inputs, &analysis_outputs, true);
605 606
  PADDLE_ENFORCE_GT(native_outputs.size(), 0, "Native output is empty.");
  PADDLE_ENFORCE_GT(analysis_outputs.size(), 0, "Analysis output is empty.");
607
  CompareResult(analysis_outputs.back(), native_outputs.back());
T
Tao Luo 已提交
608 609
}

610
void CompareQuantizedAndAnalysis(
611
    const AnalysisConfig *config, const AnalysisConfig *qconfig,
612 613
    const std::vector<std::vector<PaddleTensor>> &inputs,
    const int compared_idx = 1) {
614 615 616 617 618 619 620 621 622
  PADDLE_ENFORCE_EQ(inputs[0][0].shape[0], FLAGS_batch_size,
                    "Input data has to be packed batch by batch.");
  LOG(INFO) << "FP32 & INT8 prediction run: batch_size " << FLAGS_batch_size
            << ", warmup batch size " << FLAGS_warmup_batch_size << ".";

  LOG(INFO) << "--- FP32 prediction start ---";
  auto *cfg = reinterpret_cast<const PaddlePredictor::Config *>(config);
  PrintConfig(cfg, true);
  std::vector<std::vector<PaddleTensor>> analysis_outputs;
623
  float sample_latency_fp32{-1};
624 625 626 627 628

  if (FLAGS_enable_fp32) {
    TestOneThreadPrediction(cfg, inputs, &analysis_outputs, true, VarType::FP32,
                            &sample_latency_fp32);
  }
629 630 631 632 633

  LOG(INFO) << "--- INT8 prediction start ---";
  auto *qcfg = reinterpret_cast<const PaddlePredictor::Config *>(qconfig);
  PrintConfig(qcfg, true);
  std::vector<std::vector<PaddleTensor>> quantized_outputs;
634
  float sample_latency_int8{-1};
635

636 637 638 639
  if (FLAGS_enable_int8) {
    TestOneThreadPrediction(qcfg, inputs, &quantized_outputs, true,
                            VarType::INT8, &sample_latency_int8);
  }
640
  SummarizePerformance(sample_latency_fp32, sample_latency_int8);
641

642
  CompareAccuracy(quantized_outputs, analysis_outputs, compared_idx);
643 644
}

N
nhzlx 已提交
645 646 647 648 649 650 651 652 653 654
void CompareNativeAndAnalysis(
    PaddlePredictor *native_pred, PaddlePredictor *analysis_pred,
    const std::vector<std::vector<PaddleTensor>> &inputs) {
  int batch_size = FLAGS_batch_size;
  std::vector<PaddleTensor> native_outputs, analysis_outputs;
  native_pred->Run(inputs[0], &native_outputs, batch_size);
  analysis_pred->Run(inputs[0], &analysis_outputs, batch_size);
  CompareResult(analysis_outputs, native_outputs);
}

655
void CompareAnalysisAndZeroCopy(
656
    PaddlePredictor::Config *config, PaddlePredictor::Config *config1,
657 658 659 660 661 662 663 664 665
    const std::vector<std::vector<PaddleTensor>> &inputs,
    const std::vector<std::string> &outputs_name) {
  int batch_size = FLAGS_batch_size;
  // analysis
  std::vector<PaddleTensor> analysis_outputs;
  auto predictor = CreateTestPredictor(config, true);
  predictor->Run(inputs[0], &analysis_outputs, batch_size);
  // analysis + zero_copy
  std::vector<ZeroCopyTensor> zerocopy_outputs;
666 667
  reinterpret_cast<AnalysisConfig *>(config1)->SwitchUseFeedFetchOps(false);
  predictor = CreateTestPredictor(config1, true);
668 669 670 671 672 673
  ConvertPaddleTensorToZeroCopyTensor(predictor.get(), inputs[0]);
  predictor->ZeroCopyRun();
  for (size_t i = 0; i < outputs_name.size(); i++) {
    ZeroCopyTensor zerocopy_output =
        *predictor->GetOutputTensor(outputs_name[i]).get();
    zerocopy_outputs.emplace_back(zerocopy_output);
L
luotao1 已提交
674
    LOG(INFO) << "ZeroCopy output: " << DescribeZeroCopyTensor(zerocopy_output);
675 676 677 678 679
  }
  // compare
  CompareResult(analysis_outputs, zerocopy_outputs);
}

680 681 682 683 684 685 686
void SaveOptimModel(AnalysisConfig *cfg, const std::string &dstPath) {
  auto predictor = CreateTestPredictor(
      reinterpret_cast<const PaddlePredictor::Config *>(cfg),
      FLAGS_use_analysis);
  (static_cast<AnalysisPredictor *>(predictor.get()))->SaveOptimModel(dstPath);
}

L
luotao1 已提交
687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757
template <typename T>
std::string LoDTensorSummary(const framework::LoDTensor &tensor) {
  std::stringstream ss;
  ss << "\n---- tensor ---" << '\n';
  ss << "lod: [";
  for (const auto &level : tensor.lod()) {
    ss << "[ ";
    for (auto i : level) {
      ss << i << ", ";
    }
    ss << "]";
  }
  ss << "]\n";

  ss << "shape: [";
  int size = 1;
  for (int i = 0; i < tensor.dims().size(); i++) {
    int dim = tensor.dims()[i];
    ss << dim << ", ";
    size *= dim;
  }
  ss << "]\n";

  ss << "data: ";
  for (int i = 0; i < std::min(20, size); i++) {
    ss << tensor.data<T>()[i] << " ";
  }
  ss << "\n";

  return ss.str();
}

static bool CompareLoD(const framework::LoD &a, const framework::LoD &b) {
  if (a.size() != b.size()) {
    LOG(ERROR) << string::Sprintf("lod size not match %d != %d", a.size(),
                                  b.size());
    return false;
  }
  for (size_t i = 0; i < a.size(); i++) {
    auto &al = a[i];
    auto &bl = b[i];
    if (al.size() != bl.size()) {
      LOG(ERROR) << string::Sprintf("level size %d != %d", al.size(),
                                    bl.size());
      return false;
    }
  }
  return true;
}

static bool CompareShape(const std::vector<int64_t> &a,
                         const std::vector<int64_t> &b) {
  if (a.size() != b.size()) {
    LOG(ERROR) << string::Sprintf("shape size not match %d != %d", a.size(),
                                  b.size());
    return false;
  }
  for (size_t i = 0; i < a.size(); i++) {
    if (a[i] != b[i]) {
      LOG(ERROR) << string::Sprintf("shape %d-th element not match %d != %d", i,
                                    a[i], b[i]);
      return false;
    }
  }
  return true;
}

static bool CompareTensorData(const framework::LoDTensor &a,
                              const framework::LoDTensor &b) {
  auto a_shape = framework::vectorize(a.dims());
  auto b_shape = framework::vectorize(b.dims());
758
  size_t a_size = std::accumulate(a_shape.begin(), a_shape.end(), size_t{1},
L
luotao1 已提交
759
                                  [](int a, int b) { return a * b; });
760
  size_t b_size = std::accumulate(b_shape.begin(), b_shape.end(), size_t{1},
L
luotao1 已提交
761 762 763 764 765 766 767
                                  [](int a, int b) { return a * b; });
  if (a_size != b_size) {
    LOG(ERROR) << string::Sprintf("tensor data size not match, %d != %d",
                                  a_size, b_size);
  }

  for (size_t i = 0; i < a_size; i++) {
768
    if (a.type() == VarType::FP32) {
L
luotao1 已提交
769 770 771 772 773 774 775 776
      const auto *a_data = a.data<float>();
      const auto *b_data = b.data<float>();
      if (std::abs(a_data[i] - b_data[i]) > 1e-3) {
        LOG(ERROR) << string::Sprintf(
            "tensor data %d-th element not match, %f != %f", i, a_data[i],
            b_data[i]);
        return false;
      }
777
    } else if (a.type() == VarType::INT64) {
L
luotao1 已提交
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808
      const auto *a_data = a.data<int64_t>();
      const auto *b_data = b.data<int64_t>();
      if (std::abs(a_data[i] - b_data[i]) > 1e-3) {
        LOG(ERROR) << string::Sprintf(
            "tensor data %d-th element not match, %f != %f", i, a_data[i],
            b_data[i]);
        return false;
      }
    }
  }

  return true;
}

static bool CompareTensor(const framework::LoDTensor &a,
                          const framework::LoDTensor &b) {
  if (!CompareLoD(a.lod(), b.lod())) {
    return false;
  }
  if (!CompareShape(framework::vectorize(a.dims()),
                    framework::vectorize(b.dims()))) {
    return false;
  }

  if (!CompareTensorData(a, b)) {
    return false;
  }

  return true;
}

L
luotao1 已提交
809 810
}  // namespace inference
}  // namespace paddle