theano_conv.py

import numpy as np
import theano
import theano.tensor as T
import time

x = T.tensor4('x')
x = theano.shared(
    np.random.rand(32, 128, 256, 256).astype(theano.config.floatX),
    'x')
filters = theano.shared(
    np.random.rand(256, 128, 8, 8).astype(theano.config.floatX),
    'filters')
# B x 1 x 1 x T
y = theano.gpuarray.dnn.dnn_conv(
    img=x,
    kerns=filters,
    border_mode='half',
    precision='float32')

f = theano.function([], y)
y_ = f()
#f.sync_shared()
t0 = time.time()
for i in range(50):
    print i
    y_ = f(x_)
 #   f.sync_shared()
t1 = time.time()
print t1 - t0

# rm -rf `theano-cache`
# export CUDA_LAUNCH_BLOCKING=1
# THEANO_FLAGS=cmodule.debug=True,device=cuda,floatX=float16,dnn.conv.algo_fwd=time_once python theano_conv.py
# THEANO_FLAGS=cmodule.debug=True,device=cuda,floatX=float32,dnn.conv.algo_fwd=time_once python theano_conv.py