def test_computation_performance(thr_and_double, fast_math, test_sampler_float):
thr, double = thr_and_double
size = 2 ** 15
batch = 2 ** 6
bijection = philox(64, 4)
sampler = test_sampler_float.get_sampler(bijection, double)
rng = CBRNG(Type(sampler.dtype, shape=(batch, size)), 1, sampler)
dest_dev = thr.empty_like(rng.parameter.randoms)
counters = rng.create_counters()
counters_dev = thr.to_device(counters)
rngc = rng.compile(thr, fast_math=fast_math)
attempts = 10
times = []
for i in range(attempts):
t1 = time.time()
rngc(counters_dev, dest_dev)
thr.synchronize()
times.append(time.time() - t1)
byte_size = size * batch * sampler.dtype.itemsize
return min(times), byte_size
def test_computation_performance(thr_and_double, fast_math,
test_sampler_float):
thr, double = thr_and_double
size = 2**15
batch = 2**6
sampler = test_sampler_float.get_sampler(double)
rng = CBRNG(Type(sampler.dtype, shape=(batch, size)), 1, sampler)
dest_dev = thr.empty_like(rng.parameter.randoms)
counters = rng.create_counters()
counters_dev = thr.to_device(counters)
rngc = rng.compile(thr, fast_math=fast_math)
attempts = 10
times = []
for i in range(attempts):
t1 = time.time()
rngc(counters_dev, dest_dev)
thr.synchronize()
times.append(time.time() - t1)
byte_size = size * batch * sampler.dtype.itemsize
return min(times), byte_size
def test_computation_convenience(thr):
size = 10000
batch = 101
ref = UniformIntegerHelper(0, 511)
rng = CBRNG.uniform_integer(Type(numpy.int32, shape=(batch, size)), 1,
sampler_kwds=dict(low=ref.extent[0], high=ref.extent[1]))
check_computation(thr, rng, ref)
def test_computation_convenience(thr):
size = 10000
batch = 101
extent = (0, 511)
mean, std = uniform_discrete_mean_and_std(*extent)
rng = CBRNG.uniform_integer(Type(numpy.int32, shape=(batch, size)), 1,
sampler_kwds=dict(low=extent[0], high=extent[1] + 1))
check_computation(thr, rng, extent=extent, mean=mean, std=std)
def test_computation_convenience(thr):
size = 10000
batch = 101
extent = (0, 511)
mean, std = uniform_discrete_mean_and_std(*extent)
rng = CBRNG.uniform_integer(Type(numpy.int32, shape=(batch, size)),
1,
sampler_kwds=dict(low=extent[0],
high=extent[1] + 1))
check_computation(thr, rng, extent=extent, mean=mean, std=std)
def test_computation_general(thr_and_double):
size = 10000
batch = 101
thr, double = thr_and_double
dtype = numpy.float64 if double else numpy.float32
mean, std = -2, 10
bijection = philox(64, 4)
sampler = normal_bm(bijection, dtype, mean=mean, std=std)
rng = CBRNG(Type(dtype, shape=(batch, size)), 1, sampler)
check_computation(thr, rng, mean=mean, std=std)
def test_computation_general(thr_and_double):
size = 10000
batch = 101
thr, double = thr_and_double
bijection = philox(64, 4)
ref = NormalBMHelper(mean=-2, std=10)
sampler = ref.get_sampler(bijection, double)
rng = CBRNG(Type(sampler.dtype, shape=(batch, size)), 1, sampler)
check_computation(thr, rng, ref)
def randint(self, array, minval, maxval, seed=None):
kernel_cache, thread = self.kernel_cache, self.thread
key = (self.randint, array.shape, array.dtype, minval, maxval, thread)
if key not in kernel_cache.keys():
log.info("compiling " + str(key))
rng = CBRNG.uniform_integer(Type(array.dtype, shape=array.shape), len(array.shape), # @UndefinedVariable
sampler_kwds=dict(low=numpy.int32(minval), high=numpy.int32(maxval)), seed=seed)
counters = thread.to_device(rng.create_counters())
kernel_cache[key] = (rng.compile(thread), counters)
(rng, counters) = kernel_cache[key]
rng(counters, array)
def test_computation_uniqueness(thr):
"""
A regression test for the bug with a non-updating counter.
"""
size = 10000
batch = 1
rng = CBRNG.normal_bm(Type(numpy.complex64, shape=(batch, size)), 1)
dest1_dev = thr.empty_like(rng.parameter.randoms)
dest2_dev = thr.empty_like(rng.parameter.randoms)
counters = rng.create_counters()
counters_dev = thr.to_device(counters)
rngc = rng.compile(thr)
rngc(counters_dev, dest1_dev)
rngc(counters_dev, dest2_dev)
assert not diff_is_negligible(dest1_dev.get(), dest2_dev.get())
def test_computation_uniqueness(thr):
"""
A regression test for the bug with a non-updating counter.
"""
size = 10000
batch = 1
rng = CBRNG.normal_bm(Type(numpy.complex64, shape=(batch, size)), 1)
dest1_dev = thr.empty_like(rng.parameter.randoms)
dest2_dev = thr.empty_like(rng.parameter.randoms)
counters = rng.create_counters()
counters_dev = thr.to_device(counters)
rngc = rng.compile(thr)
rngc(counters_dev, dest1_dev)
rngc(counters_dev, dest2_dev)
assert not diff_is_negligible(dest1_dev.get(), dest2_dev.get(), verbose=False)
def normal(self, array, mean, std, seed=None):
kernel_cache, thread = self.kernel_cache, self.thread
key = (self.normal, array.shape, array.dtype, mean, std, thread)
if key not in kernel_cache.keys():
log.info("compiling " + str(key))
rng = CBRNG.normal_bm(Type(array.dtype, shape=array.shape), len(array.shape), # @UndefinedVariable
sampler_kwds=dict(mean=numpy.float32(mean), std=numpy.float32(std)), seed=seed)
counters = thread.to_device(rng.create_counters())
kernel_cache[key] = (rng.compile(thread), counters)
(rng, counters) = kernel_cache[key]
rng(counters, array)
return array
def randint(self, array, minval, maxval, seed=None):
kernel_cache, thread = self.kernel_cache, self.thread
key = (self.randint, array.shape, array.dtype, minval, maxval, thread)
if key not in kernel_cache.keys():
log.info("compiling " + str(key))
rng = CBRNG.uniform_integer(
Type(array.dtype, shape=array.shape),
len(array.shape), # @UndefinedVariable
sampler_kwds=dict(low=numpy.int32(minval),
high=numpy.int32(maxval)),
seed=seed)
counters = thread.to_device(rng.create_counters())
kernel_cache[key] = (rng.compile(thread), counters)
(rng, counters) = kernel_cache[key]
rng(counters, array)
def normal(self, array, mean, std, seed=None):
kernel_cache, thread = self.kernel_cache, self.thread
key = (self.normal, array.shape, array.dtype, mean, std, thread)
if key not in kernel_cache.keys():
log.info("compiling " + str(key))
rng = CBRNG.normal_bm(
Type(array.dtype, shape=array.shape),
len(array.shape), # @UndefinedVariable
sampler_kwds=dict(mean=numpy.float32(mean),
std=numpy.float32(std)),
seed=seed)
counters = thread.to_device(rng.create_counters())
kernel_cache[key] = (rng.compile(thread), counters)
(rng, counters) = kernel_cache[key]
rng(counters, array)
return array