def test_outoforderqueue_clmath(ctx_factory):
context = ctx_factory()
try:
queue = cl.CommandQueue(context,
properties=cl.command_queue_properties.OUT_OF_ORDER_EXEC_MODE_ENABLE)
except Exception:
pytest.skip("out-of-order queue not available")
a = np.random.rand(10**6).astype(np.dtype('float32'))
a_gpu = cl_array.to_device(queue, a)
# testing that clmath functions wait for and create events
b_gpu = clmath.fabs(clmath.sin(a_gpu * 5))
queue.finish()
b1 = b_gpu.get()
b = np.abs(np.sin(a * 5))
assert np.abs(b1 - b).mean() < 1e-5
def test_outoforderqueue_clmath(ctx_factory):
context = ctx_factory()
try:
queue = cl.CommandQueue(context,
properties=cl.command_queue_properties.
OUT_OF_ORDER_EXEC_MODE_ENABLE)
except Exception:
pytest.skip("out-of-order queue not available")
a = np.random.rand(10**6).astype(np.dtype('float32'))
a_gpu = cl_array.to_device(queue, a)
# testing that clmath functions wait for and create events
b_gpu = clmath.fabs(clmath.sin(a_gpu * 5))
queue.finish()
b1 = b_gpu.get()
b = np.abs(np.sin(a * 5))
assert np.abs(b1 - b).mean() < 1e-5
def sin(arr):
#TODO: work with axis, out, keepdims
res = clmath.sin(arr, queue=queue) #np.sum(*args, **kwargs)
res.__class__ = myclArray
res.reinit()
return res
def test_target_specific_qbx(ctx_getter, op, helmholtz_k, qbx_order):
logging.basicConfig(level=logging.INFO)
cl_ctx = ctx_getter()
queue = cl.CommandQueue(cl_ctx)
target_order = 4
fmm_tol = 1e-3
from meshmode.mesh.generation import generate_icosphere
mesh = generate_icosphere(1, target_order)
from meshmode.discretization import Discretization
from meshmode.discretization.poly_element import \
InterpolatoryQuadratureSimplexGroupFactory
from pytential.qbx import QBXLayerPotentialSource
pre_density_discr = Discretization(
cl_ctx, mesh, InterpolatoryQuadratureSimplexGroupFactory(target_order))
from sumpy.expansion.level_to_order import SimpleExpansionOrderFinder
refiner_extra_kwargs = {}
if helmholtz_k != 0:
refiner_extra_kwargs["kernel_length_scale"] = 5 / abs(helmholtz_k)
qbx, _ = QBXLayerPotentialSource(
pre_density_discr,
4 * target_order,
qbx_order=qbx_order,
fmm_level_to_order=SimpleExpansionOrderFinder(fmm_tol),
fmm_backend="fmmlib",
_expansions_in_tree_have_extent=True,
_expansion_stick_out_factor=0.9,
_use_target_specific_qbx=False,
).with_refinement(**refiner_extra_kwargs)
density_discr = qbx.density_discr
nodes = density_discr.nodes().with_queue(queue)
u_dev = clmath.sin(nodes[0])
if helmholtz_k == 0:
kernel = LaplaceKernel(3)
kernel_kwargs = {}
else:
kernel = HelmholtzKernel(3, allow_evanescent=True)
kernel_kwargs = {"k": sym.var("k")}
u_sym = sym.var("u")
if op == "S":
op = sym.S
elif op == "D":
op = sym.D
elif op == "Sp":
op = sym.Sp
else:
raise ValueError("unknown operator: '%s'" % op)
expr = op(kernel, u_sym, qbx_forced_limit=-1, **kernel_kwargs)
bound_op = bind(qbx, expr)
pot_ref = bound_op(queue, u=u_dev, k=helmholtz_k).get()
qbx = qbx.copy(_use_target_specific_qbx=True)
bound_op = bind(qbx, expr)
pot_tsqbx = bound_op(queue, u=u_dev, k=helmholtz_k).get()
assert np.allclose(pot_tsqbx, pot_ref, atol=1e-13, rtol=1e-13)