def _profile():
seq_length = 50
batchsize = 48
feature_dimension = 128
data_cpu = np.random.normal(0,
1,
size=(batchsize, feature_dimension,
seq_length)).astype(np.float32)
data_gpu = cuda.to_gpu(data_cpu, gpu_device)
# CPU
layer = SRU(feature_dimension, feature_dimension)
for _ in range(100):
h_cpu, c_cpu = layer(data_cpu)
layer.reset_state()
# GPU (define-by-run)
layer = NaiveSRU(feature_dimension, feature_dimension)
layer.to_gpu(gpu_device)
for _ in range(100):
h, c = layer(data_gpu)
layer.reset_state()
# GPU (CUDA Kernel)
layer = SRU(feature_dimension, feature_dimension)
layer.to_gpu(gpu_device)
for _ in range(100):
h_gpu, c_gpu = layer(data_gpu)
layer.reset_state()
# GPU (PyTorch)
with torch.cuda.device(gpu_device):
from cuda_functional import SRU as PyTorchSRU
data_gpu_torch = torch.FloatTensor(seq_length, batchsize,
feature_dimension).cuda()
rnn = PyTorchSRU(128,
128,
num_layers=1,
dropout=0.0,
rnn_dropout=0.0,
use_tanh=0,
bidirectional=False)
rnn.cuda()
for _ in range(100):
output, hidden = rnn(torch.autograd.Variable(data_gpu_torch))
# LSTM (Chainer)
layer = links.LSTM(feature_dimension, feature_dimension)
layer.to_gpu(gpu_device)
for _ in range(100):
for t in range(seq_length):
h = layer(data_gpu[..., t])
layer.reset_state()
print(h_cpu)
print(h_gpu)
def check_outputs():
seq_length = 50
batchsize = 48
feature_dimension = 128
data_cpu = np.random.normal(0,
1,
size=(batchsize, feature_dimension,
seq_length)).astype(np.float32)
data_gpu = cuda.to_gpu(data_cpu, gpu_device)
# CPU
layer = SRU(feature_dimension, feature_dimension)
h_cpu, c_cpu = layer(data_cpu)
layer.reset_state()
# GPU
layer.to_gpu(gpu_device)
h_gpu, c_gpu = layer(data_gpu)
layer.reset_state()
print(np.mean(abs(c_cpu.data - cuda.to_cpu(c_gpu.data))))
print(np.mean(abs(h_cpu.data - cuda.to_cpu(h_gpu.data))))
