fluid_benchmark.py 15.6 KB
Newer Older
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
# 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.

import argparse
import cProfile
import time
import os

import numpy as np

import paddle.fluid as fluid
import paddle.fluid.core as core
import paddle.fluid.profiler as profiler
import paddle.fluid.transpiler.distribute_transpiler as distribute_transpiler

BENCHMARK_MODELS = [
    "machine_translation", "resnet", "vgg", "mnist", "stacked_dynamic_lstm"
]


def parse_args():
    parser = argparse.ArgumentParser('Fluid model benchmarks.')
    parser.add_argument(
        '--model',
        type=str,
        choices=BENCHMARK_MODELS,
        default='resnet',
        help='The model to run benchmark with.')
    parser.add_argument(
        '--batch_size', type=int, default=32, help='The minibatch size.')
    parser.add_argument(
L
Luo Tao 已提交
43
        '--learning_rate', type=float, default=0.001, help='The learning rate.')
44 45 46 47 48 49 50
    parser.add_argument(
        '--skip_batch_num',
        type=int,
        default=5,
        help='The first num of minibatch num to skip, for better performance test'
    )
    parser.add_argument(
51 52 53 54
        '--iterations',
        type=int,
        default=80,
        help='The number of minibatches, set to -1 to run all batches.')
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73
    parser.add_argument(
        '--pass_num', type=int, default=100, help='The number of passes.')
    parser.add_argument(
        '--data_format',
        type=str,
        default='NCHW',
        choices=['NCHW', 'NHWC'],
        help='The data data_format, now only support NCHW.')
    parser.add_argument(
        '--device',
        type=str,
        default='GPU',
        choices=['CPU', 'GPU'],
        help='The device type.')
    parser.add_argument(
        '--gpus',
        type=int,
        default=1,
        help='If gpus > 1, will use ParallelExecutor to run, else use Executor.')
74
    # this option is available only for vgg and resnet.
75 76 77 78
    parser.add_argument(
        '--data_set',
        type=str,
        default='flowers',
79
        choices=['cifar10', 'flowers', 'imagenet'],
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
        help='Optional dataset for benchmark.')
    parser.add_argument(
        '--infer_only', action='store_true', help='If set, run forward only.')
    parser.add_argument(
        '--use_cprof', action='store_true', help='If set, use cProfile.')
    parser.add_argument(
        '--use_nvprof',
        action='store_true',
        help='If set, use nvprof for CUDA.')
    parser.add_argument(
        '--no_test',
        action='store_false',
        help='If set, test the testset during training.')
    parser.add_argument(
        '--memory_optimize',
        action='store_true',
        help='If set, optimize runtime memory before start.')
97 98 99 100
    parser.add_argument(
        '--use_fake_data',
        action='store_true',
        help='If set ommit the actual read data operators.')
X
Xin Pan 已提交
101 102
    parser.add_argument(
        '--profile', action='store_true', help='If set, profile a few steps.')
103 104 105 106 107 108
    parser.add_argument(
        '--update_method',
        type=str,
        default='local',
        choices=['local', 'pserver', 'nccl2'],
        help='Choose parameter update method, can be local, pserver, nccl2.')
Y
yi.wu 已提交
109 110 111 112 113 114 115 116 117 118
    parser.add_argument(
        '--use_reader_op',
        action='store_true',
        help='Whether to use reader op, and must specify the data path if set this to true.'
    )
    parser.add_argument(
        '--data_path',
        type=str,
        default="",
        help='Directory that contains all the training recordio files.')
119 120 121 122
    args = parser.parse_args()
    return args


X
Xin Pan 已提交
123 124
def append_nccl2_prepare(trainer_id):
    if trainer_id >= 0:
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
        # append gen_nccl_id at the end of startup program
        trainer_id = int(os.getenv("PADDLE_TRAINER_ID"))
        port = os.getenv("PADDLE_PSERVER_PORT")
        worker_ips = os.getenv("PADDLE_TRAINER_IPS")
        worker_endpoints = []
        for ip in worker_ips.split(","):
            worker_endpoints.append(':'.join([ip, port]))
        num_trainers = len(worker_endpoints)
        current_endpoint = os.getenv("PADDLE_CURRENT_IP") + ":" + port
        worker_endpoints.remove(current_endpoint)

        nccl_id_var = fluid.default_startup_program().global_block().create_var(
            name="NCCLID",
            persistable=True,
            type=fluid.core.VarDesc.VarType.RAW)
        fluid.default_startup_program().global_block().append_op(
            type="gen_nccl_id",
            inputs={},
            outputs={"NCCLID": nccl_id_var},
            attrs={
                "endpoint": current_endpoint,
                "endpoint_list": worker_endpoints,
                "trainer_id": trainer_id
            })
        return nccl_id_var, num_trainers, trainer_id
    else:
X
Xin Pan 已提交
151 152
        raise Exception("must set positive PADDLE_TRAINER_ID env variables for "
                        "nccl-based dist train.")
153 154


X
Xin Pan 已提交
155 156
def dist_transpile(trainer_id):
    if trainer_id < 0:
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
        return None, None

    # the port of all pservers, needed by both trainer and pserver
    port = os.getenv("PADDLE_PSERVER_PORT", "6174")
    # comma separated ips of all pservers, needed by trainer and
    # pserver
    pserver_ips = os.getenv("PADDLE_PSERVER_IPS", "")
    eplist = []
    for ip in pserver_ips.split(","):
        eplist.append(':'.join([ip, port]))
    pserver_endpoints = ",".join(eplist)
    # total number of workers/trainers in the job, needed by
    # trainer and pserver
    trainers = int(os.getenv("PADDLE_TRAINERS"))
    # the IP of the local machine, needed by pserver only
    current_endpoint = os.getenv("PADDLE_CURRENT_IP", "") + ":" + port
    # the role, should be either PSERVER or TRAINER
    training_role = os.getenv("PADDLE_TRAINING_ROLE")

    t = distribute_transpiler.DistributeTranspiler()
    t.transpile(trainer_id, pservers=pserver_endpoints, trainers=trainers)
    if training_role == "PSERVER":
        pserver_program = t.get_pserver_program(current_endpoint)
        pserver_startup_program = t.get_startup_program(current_endpoint,
                                                        pserver_program)
        return pserver_program, pserver_startup_program
    elif training_role == "TRAINER":
        train_program = t.get_trainer_program()
        return train_program, fluid.default_startup_program()
    else:
        raise ValueError(
            'TRAINING_ROLE environment variable must be either TRAINER or PSERVER'
        )


def test(exe, inference_program, test_reader, feeder, batch_acc):
    accuracy_evaluator = fluid.metrics.Accuracy()
    for batch_id, data in enumerate(test_reader()):
        acc = exe.run(inference_program,
                      feed=feeder.feed(data),
                      fetch_list=[batch_acc])
        accuracy_evaluator.update(value=np.array(acc), weight=len(data))

    return accuracy_evaluator.eval()


# TODO(wuyi): replace train, train_parallel, test functions with new trainer
# API once it is ready.
def train(avg_loss, infer_prog, optimizer, train_reader, test_reader, batch_acc,
          args, train_prog, startup_prog):
    if os.getenv("PADDLE_TRAINING_ROLE") == "PSERVER":
        place = core.CPUPlace()
        exe = fluid.Executor(place)
        exe.run(startup_prog)
        exe.run(train_prog)
        return

214 215 216 217
    if args.use_fake_data:
        raise Exception(
            "fake data is not supported in single GPU test for now.")

218 219 220
    place = core.CPUPlace() if args.device == 'CPU' else core.CUDAPlace(0)
    exe = fluid.Executor(place)
    exe.run(startup_prog)
Y
yi.wu 已提交
221 222 223 224 225 226 227

    if not args.use_reader_op:
        feed_var_list = [
            var for var in train_prog.global_block().vars.itervalues()
            if var.is_data
        ]
        feeder = fluid.DataFeeder(feed_var_list, place)
228 229 230 231

    iters, num_samples, start_time = 0, 0, time.time()
    for pass_id in range(args.pass_num):
        train_losses = []
232 233 234 235 236 237 238 239
        reader_generator = train_reader()
        batch_id = 0
        data = None
        while True:
            if not args.use_reader_op:
                data = next(reader_generator, None)
            if iters == args.iterations or data == None:
                break
240 241 242
            if iters == args.skip_batch_num:
                start_time = time.time()
                num_samples = 0
243

Y
yi.wu 已提交
244 245 246 247 248 249
            if args.use_reader_op:
                loss = exe.run(train_prog, fetch_list=[avg_loss])
            else:
                loss = exe.run(train_prog,
                               feed=feeder.feed(data),
                               fetch_list=[avg_loss])
250
            iters += 1
251 252 253
            batch_id += 1
            # FIXME(wuyi): last batch size maybe different
            num_samples += len(args.batch_size)
254 255 256
            train_losses.append(loss)
            print("Pass: %d, Iter: %d, Loss: %f\n" %
                  (pass_id, iters, np.mean(train_losses)))
L
Luo Tao 已提交
257
        print_train_time(start_time, time.time(), num_samples)
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
        print("Pass: %d, Loss: %f" % (pass_id, np.mean(train_losses)))
        # evaluation
        if not args.no_test and batch_acc != None:
            pass_test_acc = test(exe, infer_prog, test_reader, feeder,
                                 batch_acc)
            print(", Test Accuracy: %f" % pass_test_acc)
        print("\n")
        # TODO(wuyi): add warmup passes to get better perf data.
        exit(0)


# TODO(wuyi): replace train, train_parallel, test functions with new trainer
# API once it is ready.
def train_parallel(avg_loss, infer_prog, optimizer, train_reader, test_reader,
                   batch_acc, args, train_prog, startup_prog, nccl_id_var,
                   num_trainers, trainer_id):
Y
yi.wu 已提交
274 275 276 277 278 279 280 281
    place = core.CPUPlace() if args.device == 'CPU' else core.CUDAPlace(0)
    if not args.use_reader_op:
        feed_var_list = [
            var for var in train_prog.global_block().vars.itervalues()
            if var.is_data
        ]
        feeder = fluid.DataFeeder(feed_var_list, place)

282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301
    # generate fake:
    if args.use_fake_data:
        for var in feed_var_list:
            v = startup_prog.global_block().clone_variable(var)
            var.persistable = True
            v.persistable = True

            real_shape = list(var.shape)
            real_shape[0] = args.batch_size / args.gpus
            startup_prog.global_block().append_op(
                outputs={"Out": v},
                type="fill_constant",
                attrs={"shape": real_shape,
                       "value": 1.0,
                       "dtype": var.dtype})

    if nccl_id_var and trainer_id == 0:
        #FIXME(wuyi): wait other trainer to start listening
        time.sleep(30)

302 303 304 305 306 307 308 309 310 311 312
    startup_exe = fluid.Executor(place)
    startup_exe.run(startup_prog)
    strategy = fluid.ExecutionStrategy()
    strategy.num_threads = 1
    strategy.allow_op_delay = False
    exe = fluid.ParallelExecutor(
        True,
        avg_loss.name,
        exec_strategy=strategy,
        num_trainers=num_trainers,
        trainer_id=trainer_id)
313

314 315 316 317
    for pass_id in range(args.pass_num):
        num_samples = 0
        iters = 0
        start_time = time.time()
318 319 320 321 322 323 324 325
        reader_generator = train_reader()
        batch_id = 0
        data = None
        while True:
            if not args.use_reader_op:
                data = next(reader_generator, None)
            if iters == args.iterations or data == None:
                break
X
Xin Pan 已提交
326 327 328 329 330
            if args.profile and pass_id == 0 and batch_id == 5:
                profiler.start_profiler("All")
            elif args.profile and pass_id == 0 and batch_id == 10:
                profiler.stop_profiler("total", "/tmp/profile_%d" % trainer_id)

331 332 333
            if iters == args.skip_batch_num:
                start_time = time.time()
                num_samples = 0
Y
yi.wu 已提交
334 335 336 337
            # NOTE: if use reader ops, the input data is not splited to multiple cards
            if args.use_reader_op and iters >= args.iterations / args.gpus:
                break
            if args.use_fake_data or args.use_reader_op:
338 339 340
                loss, = exe.run([avg_loss.name])
            else:
                loss, = exe.run([avg_loss.name], feed=feeder.feed(data))
341 342 343 344 345 346 347
            if args.update_method == "pserver":
                exe.bcast_params()
            num_samples += len(data)
            iters += 1
            if batch_id % 1 == 0:
                print("Pass %d, batch %d, loss %s" %
                      (pass_id, batch_id, np.array(loss)))
348
            batch_id += 1
Y
yi.wu 已提交
349 350
        if args.use_reader_op:
            num_samples = num_samples * args.gpus
L
Luo Tao 已提交
351
        print_train_time(start_time, time.time(), num_samples)
352 353 354 355 356 357 358 359 360 361
        if not args.no_test and batch_acc != None:
            test_acc = test(startup_exe, infer_prog, test_reader, feeder,
                            batch_acc)
            print("Pass: %d, Test Accuracy: %f\n" % (pass_id, test_acc))
        exit(0)


def print_arguments(args):
    vars(args)['use_nvprof'] = (vars(args)['use_nvprof'] and
                                vars(args)['device'] == 'GPU')
L
Luo Tao 已提交
362
    print('----------- Configuration Arguments -----------')
363 364 365 366 367
    for arg, value in sorted(vars(args).iteritems()):
        print('%s: %s' % (arg, value))
    print('------------------------------------------------')


L
Luo Tao 已提交
368 369 370 371 372 373 374
def print_train_time(start_time, end_time, num_samples):
    train_elapsed = end_time - start_time
    examples_per_sec = num_samples / train_elapsed
    print('\nTotal examples: %d, total time: %.5f, %.5f examples/sed\n' %
          (num_samples, train_elapsed, examples_per_sec))


375 376 377
def main():
    args = parse_args()
    print_arguments(args)
X
Xin Pan 已提交
378 379 380 381

    # the unique trainer id, starting from 0, needed by trainer
    # only
    nccl_id_var, num_trainers, trainer_id = (
Y
yi.wu 已提交
382
        None, 1, int(os.getenv("PADDLE_TRAINER_ID", "0")))
383 384 385 386 387 388 389 390 391 392 393 394 395

    if args.use_cprof:
        pr = cProfile.Profile()
        pr.enable()
    model_def = __import__("models.%s" % args.model, fromlist=["models"])
    train_args = list(model_def.get_model(args))
    train_args.append(args)
    # Run optimizer.minimize(avg_loss)
    train_args[2].minimize(train_args[0])
    if args.memory_optimize:
        fluid.memory_optimize(fluid.default_main_program())

    if args.update_method == "pserver":
X
Xin Pan 已提交
396
        train_prog, startup_prog = dist_transpile(trainer_id)
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
        if not train_prog:
            raise Exception(
                "Must configure correct environments to run dist train.")
        train_args.extend([train_prog, startup_prog])
        if args.gpus > 1 and os.getenv("PADDLE_TRAINING_ROLE") == "TRAINER":
            train_args.extend([nccl_id_var, num_trainers, trainer_id])
            train_parallel(*train_args)
        train(*train_args)
        exit(0)

    # for other update methods, use default programs
    train_args.append(fluid.default_main_program())
    train_args.append(fluid.default_startup_program())

    if args.update_method == "nccl2":
X
Xin Pan 已提交
412
        nccl_id_var, num_trainers, trainer_id = append_nccl2_prepare(trainer_id)
413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
    if args.gpus == 1:
        # NOTE: parallel executor use profiler interanlly
        if args.use_nvprof and args.device == 'GPU':
            with profiler.cuda_profiler("cuda_profiler.txt", 'csv') as nvprof:
                train(*train_args)
        else:
            train(*train_args)
    else:
        if args.device == "CPU":
            raise Exception("Only support GPU perf with parallel exe")
        train_args.extend([nccl_id_var, num_trainers, trainer_id])
        train_parallel(*train_args)


if __name__ == "__main__":
    main()