未验证 提交 765c70a1 编写于 作者: L liuwei1031 提交者: GitHub

Unittest improve, test=develop (#16941)

* accelerate test_ir_memory_optimize_nlp, test=develop

* accelerate test_ir_memory_optimize_nlp, test=develop
上级 23df084b
...@@ -33,6 +33,13 @@ os.environ['CPU_NUM'] = '2' ...@@ -33,6 +33,13 @@ os.environ['CPU_NUM'] = '2'
class BuildIrMemOptBase(unittest.TestCase): class BuildIrMemOptBase(unittest.TestCase):
def setup_reader(self):
self.batch_size = 32
self.word_dict = paddle.dataset.imdb.word_dict()
self.train_reader = paddle.batch(
paddle.dataset.imdb.train(self.word_dict),
batch_size=self.batch_size)
def check_network_convergence(self, def check_network_convergence(self,
network, network,
use_cuda=True, use_cuda=True,
...@@ -51,35 +58,34 @@ class BuildIrMemOptBase(unittest.TestCase): ...@@ -51,35 +58,34 @@ class BuildIrMemOptBase(unittest.TestCase):
return return
fluid.default_startup_program().random_seed = 100 fluid.default_startup_program().random_seed = 100
fluid.default_main_program().random_seed = 100 fluid.default_main_program().random_seed = 100
batch_size = 32
batch_size *= fluid.core.get_cuda_device_count() if use_cuda else int(
os.environ.get('CPU_NUM', multiprocessing.cpu_count()))
# build network
word_dict = paddle.dataset.imdb.word_dict()
train_reader = paddle.batch(
paddle.dataset.imdb.train(word_dict), batch_size=batch_size)
data = fluid.layers.data( data = fluid.layers.data(
name="words", shape=[1], dtype="int64", lod_level=1) name="words", shape=[1], dtype="int64", lod_level=1)
label = fluid.layers.data(name="label", shape=[1], dtype="int64") label = fluid.layers.data(name="label", shape=[1], dtype="int64")
cost = network(data, label, len(word_dict)) cost = network(data, label, len(self.word_dict))
optimizer = fluid.optimizer.Adam(learning_rate=0.001) optimizer = fluid.optimizer.Adam(learning_rate=0.001)
optimizer.minimize(cost) optimizer.minimize(cost)
build_strategy = fluid.BuildStrategy()
build_strategy.enable_inplace = False
build_strategy.memory_optimize = False
if memory_opt: if memory_opt:
fluid.memory_optimize(fluid.default_main_program()) fluid.memory_optimize(fluid.default_main_program())
else:
build_strategy.enable_inplace = use_ir_memory_optimize
build_strategy.memory_optimize = enable_inplace
# execution # execution
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace() place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
feeder = fluid.DataFeeder(feed_list=[data, label], place=place) feeder = fluid.DataFeeder(feed_list=[data, label], place=place)
reader = feeder.decorate_reader(train_reader, multi_devices=True) reader = feeder.decorate_reader(self.train_reader, multi_devices=True)
exe = fluid.Executor(place) exe = fluid.Executor(place)
exe.run(fluid.default_startup_program()) exe.run(fluid.default_startup_program())
train_cp = compiler.CompiledProgram(fluid.default_main_program()) train_cp = compiler.CompiledProgram(fluid.default_main_program())
train_cp = train_cp.with_data_parallel(loss_name=cost.name) train_cp = train_cp.with_data_parallel(
loss_name=cost.name, build_strategy=build_strategy)
fetch_list = [cost.name] fetch_list = [cost.name]
begin = time.time() begin = time.time()
...@@ -100,7 +106,7 @@ class BuildIrMemOptBase(unittest.TestCase): ...@@ -100,7 +106,7 @@ class BuildIrMemOptBase(unittest.TestCase):
end = time.time() end = time.time()
print("%.4f Instance per second" % ( print("%.4f Instance per second" % (
(batch_size * iter) / (end - begin))) (self.batch_size * iter) / (end - begin)))
print(first_loss, last_loss) print(first_loss, last_loss)
avg_last_loss_val = np.array(last_loss).mean() avg_last_loss_val = np.array(last_loss).mean()
...@@ -120,31 +126,21 @@ class TestIrMemOptBase(BuildIrMemOptBase): ...@@ -120,31 +126,21 @@ class TestIrMemOptBase(BuildIrMemOptBase):
if self.network is None or not core.is_compiled_with_cuda(): if self.network is None or not core.is_compiled_with_cuda():
return return
baseline_first_loss, baseline_last_loss = None, None self.setup_reader()
for use_cuda in [True]:
for use_python_mem_opt in [True, False]:
print(
'network: {}, use_cuda: {}, use_python_mem_opt: {}, use_ir_mem_opt : {}'.
format(self.network.__name__, use_cuda, use_python_mem_opt,
not use_python_mem_opt))
with fluid.program_guard(fluid.Program(), fluid.Program()): with fluid.program_guard(fluid.Program(), fluid.Program()):
with fluid.scope_guard(core.Scope()): with fluid.scope_guard(core.Scope()):
if use_cuda is True and use_python_mem_opt is True:
baseline_first_loss, baseline_last_loss = self.check_network_convergence( baseline_first_loss, baseline_last_loss = self.check_network_convergence(
self.network, self.network)
use_cuda=use_cuda,
memory_opt=use_python_mem_opt)
else:
cur_first_loss, cur_last_loss = self.check_network_convergence( cur_first_loss, cur_last_loss = self.check_network_convergence(
self.network, self.network, memory_opt=False)
use_cuda=use_cuda,
memory_opt=use_python_mem_opt)
self.assertAlmostEquals( self.assertAlmostEquals(
np.mean(baseline_last_loss), np.mean(baseline_last_loss),
np.mean(cur_last_loss), np.mean(cur_last_loss),
delta=1e-2) delta=1e-6)
self.assertAlmostEquals( self.assertAlmostEquals(
np.mean(baseline_first_loss), np.mean(baseline_first_loss),
np.mean(cur_first_loss), np.mean(cur_first_loss),
delta=1e-2) delta=1e-6)
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册