def test():
model = SimpleNetChild()
dataset = [((numpy.ones((2, 5, 5)) * i).astype(numpy.float32),
numpy.int32(0)) for i in range(100)]
batch_size = 5
devices = tuple([chainer.get_device(d) for d in sys.argv[1].split(',')])
iters = [chainer.iterators.SerialIterator(i, batch_size) for i in
chainer.datasets.split_dataset_n_random(
dataset, len(devices))]
optimizer = chainer.optimizers.SGD(lr=1.0)
optimizer.setup(model)
# Initialize CUDA context.
cuda.cupy.cuda.runtime.runtimeGetVersion()
try:
mpu.MultiprocessParallelUpdater(iters, optimizer, devices=devices)
except RuntimeError as e:
if sys.argv[2] == 'fork':
assert 'CUDA context' in str(e)
return
updater = mpu.MultiprocessParallelUpdater(
iters, optimizer, devices=devices)
trainer = chainer.training.Trainer(updater, (1, 'epoch'), '/tmp')
trainer.run()
assert sys.argv[2] != 'fork'
def test():
model = chainer.Link()
dataset = [((numpy.ones(
(2, 5, 5)) * i).astype(numpy.float32), numpy.int32(0))
for i in range(100)]
batch_size = 5
devices = (0, )
iters = [
chainer.iterators.SerialIterator(i, batch_size)
for i in chainer.datasets.split_dataset_n_random(
dataset, len(devices))
]
optimizer = chainer.optimizers.SGD(lr=1.0)
optimizer.setup(model)
# Initialize CUDA context.
cuda.cupy.cuda.runtime.runtimeGetVersion()
try:
mpu.MultiprocessParallelUpdater(iters, optimizer, devices=devices)
except RuntimeError as e:
assert 'CUDA context' in str(e)
return
assert False
def test_update_uses_raw_array(self):
if mpu.MultiprocessParallelUpdater.available():
model = SimpleNetRawArray(self)
dataset = [((numpy.ones((2, 5, 5)) * i).astype(numpy.float32),
numpy.int32(0)) for i in range(100)]
batch_size = 5
devices = (1,)
iters = [chainer.iterators.SerialIterator(i, batch_size) for i in
chainer.datasets.split_dataset_n_random(
dataset, len(devices))]
optimizer = chainer.optimizers.SGD(lr=1.0)
optimizer.setup(model)
updater = mpu.MultiprocessParallelUpdater(
iters, optimizer, devices=devices)
updater.update()
self.assertEqual(model.call_called, 1)
def test():
model = SimpleNetRawArray()
dataset = [((numpy.ones(
(2, 5, 5)) * i).astype(numpy.float32), numpy.int32(0))
for i in range(100)]
batch_size = 5
devices = (0, )
iters = [
chainer.iterators.SerialIterator(i, batch_size)
for i in chainer.datasets.split_dataset_n_random(
dataset, len(devices))
]
optimizer = chainer.optimizers.SGD(lr=1.0)
optimizer.setup(model)
with testing.assert_warns(UserWarning):
updater = mpu.MultiprocessParallelUpdater(iters,
optimizer,
devices=devices)
updater.update()
assert model.call_called == 1
self.loss = chainer.functions.softmax_cross_entropy(y, t)
chainer.reporter.report({'loss': self.loss}, self)
return self.loss
if __name__ == '__main__':
model = SimpleNetChildReporter()
dataset = [(numpy.full((2, 5, 5), i, numpy.float32), numpy.int32(0))
for i in range(100)]
batch_size = 5
devices = tuple([int(x) for x in sys.argv[1].split(',')])
iters = [
chainer.iterators.SerialIterator(i, batch_size)
for i in chainer.datasets.split_dataset_n_random(
dataset, len(devices))
]
optimizer = chainer.optimizers.SGD(lr=1.0)
optimizer.setup(model)
updater = mpu.MultiprocessParallelUpdater(iters,
optimizer,
devices=devices)
trainer = trainer.Trainer(updater, (1, 'iteration'), '/tmp')
trainer.run()
assert model.call_called == 1
# This snippet is not a test code.
# testing.run_module(__name__, __file__)