def test_save():
perfplot.save(
"out.png",
setup=np.random.rand,
kernels=kernels,
n_range=r,
xlabel="len(a)",
relative_to=0,
)
def test_save():
perfplot.save(
"out.png",
setup=numpy.random.rand,
kernels=kernels,
n_range=r,
xlabel="len(a)",
title="mytest",
)
return
def test_save():
def mytest(a):
return numpy.c_[a, a]
kernels = [mytest]
r = [2**k for k in range(4)]
perfplot.save(
'out.png',
setup=numpy.random.rand,
kernels=kernels, n_range=r,
xlabel='len(a)', title='mytest'
)
return
def test_save():
def mytest(a):
return numpy.c_[a, a]
kernels = [mytest]
r = [2**k for k in range(4)]
perfplot.save(
"out.png",
setup=numpy.random.rand,
kernels=kernels,
n_range=r,
xlabel="len(a)",
title="mytest",
)
return
import perfplot
import scipy.fftpack
import pyfftw
# numpy.use_fastnumpy = True
perfplot.save(
"rfftperf.png",
transparent=False,
setup=lambda n:
(numpy.random.rand(int(n))), # or simply setup=numpy.random.rand
kernels=[
lambda a: numpy.fft.rfft(a),
lambda a: scipy.fftpack.rfft(a),
lambda a: pyfftw.interfaces.numpy_fft.rfft(a),
],
labels=["numpy", "scipy", "pyfftw"],
n_range=[2000, 4000, 8000, 16000, 32000, 48000, 48000 * 2],
xlabel="len(a)",
# More optional arguments with their default values:
title="Comparison between different rfft functions",
# logx=False,
# logy=False,
equality_check=None, # set to None to disable "correctness" assertion
# automatic_order=True,
# colors=None,
# target_time_per_measurement=1.0,
# time_unit="auto", # set to one of ("s", "ms", "us", or "ns") to force plot units
# relative_to=1, # plot the timings relative to one of the measurements
)
"""
def sample(n):
_rb.memory.wait_priority_after_sampling = False
return _rb.sample(n)
return sample
# ReplayBuffer.add
perfplot.save(filename="ReplayBuffer_add.png",
setup = env,
time_unit="ms",
kernels = [add_b(brb),
add_r(rrb),
add_c(crb),
lambda e: rb.add(**e)],
labels = ["OpenAI/Baselines","Ray/RLlib","Chainer/ChainerRL","cpprb"],
n_range = [n for n in range(1,102,10)],
xlabel = "Step size added at once",
title = "Replay Buffer Add Speed",
logx = False,
logy = False,
equality_check = None)
# Fill Buffers
for _ in range(buffer_size):
o = np.random.rand(obs_shape) # [0,1)
a = np.random.rand(act_shape)
r = np.random.rand(1)
d = np.random.randint(2) # [0,2) == 0 or 1
brb.add(obs_t=o,action=a,reward=r,obs_tp1=o,done=d)